|
|
|
/*
|
|
|
|
* Compton - a compositor for X11
|
|
|
|
*
|
|
|
|
* Based on `xcompmgr` - Copyright (c) 2003, Keith Packard
|
|
|
|
*
|
|
|
|
* Copyright (c) 2011, Christopher Jeffrey
|
|
|
|
* See LICENSE for more information.
|
|
|
|
*
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include "compton.h"
|
|
|
|
|
|
|
|
// === Global constants ===
|
|
|
|
|
|
|
|
/// Name strings for window types.
|
|
|
|
const char *WINTYPES[NUM_WINTYPES] = {
|
|
|
|
"unknown",
|
|
|
|
"desktop",
|
|
|
|
"dock",
|
|
|
|
"toolbar",
|
|
|
|
"menu",
|
|
|
|
"utility",
|
|
|
|
"splash",
|
|
|
|
"dialog",
|
|
|
|
"normal",
|
|
|
|
"dropdown_menu",
|
|
|
|
"popup_menu",
|
|
|
|
"tooltip",
|
|
|
|
"notify",
|
|
|
|
"combo",
|
|
|
|
"dnd",
|
|
|
|
};
|
|
|
|
|
|
|
|
/// Names of root window properties that could point to a pixmap of
|
|
|
|
/// background.
|
|
|
|
const char *background_props_str[] = {
|
|
|
|
"_XROOTPMAP_ID",
|
|
|
|
"_XSETROOT_ID",
|
|
|
|
0,
|
|
|
|
};
|
|
|
|
|
|
|
|
// === Global variables ===
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
|
|
|
|
/// Pointer to current session, as a global variable. Only used by
|
|
|
|
/// <code>error()</code> and <code>reset_enable()</code>, which could not
|
|
|
|
/// have a pointer to current session passed in.
|
|
|
|
session_t *ps_g = NULL;
|
|
|
|
|
|
|
|
// === Fading ===
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Get the time left before next fading point.
|
|
|
|
*
|
|
|
|
* In milliseconds.
|
|
|
|
*/
|
|
|
|
static int
|
|
|
|
fade_timeout(session_t *ps) {
|
|
|
|
int diff = ps->o.fade_delta - get_time_ms() + ps->fade_time;
|
|
|
|
|
|
|
|
diff = normalize_i_range(diff, 0, ps->o.fade_delta * 2);
|
|
|
|
|
|
|
|
return diff;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Run fading on a window.
|
|
|
|
*
|
|
|
|
* @param steps steps of fading
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
run_fade(session_t *ps, win *w, unsigned steps) {
|
|
|
|
// If we have reached target opacity, return
|
|
|
|
if (w->opacity == w->opacity_tgt) {
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (!w->fade)
|
|
|
|
w->opacity = w->opacity_tgt;
|
|
|
|
else if (steps) {
|
|
|
|
// Use double below because opacity_t will probably overflow during
|
|
|
|
// calculations
|
|
|
|
if (w->opacity < w->opacity_tgt)
|
|
|
|
w->opacity = normalize_d_range(
|
|
|
|
(double) w->opacity + (double) ps->o.fade_in_step * steps,
|
|
|
|
0.0, w->opacity_tgt);
|
|
|
|
else
|
|
|
|
w->opacity = normalize_d_range(
|
|
|
|
(double) w->opacity - (double) ps->o.fade_out_step * steps,
|
|
|
|
w->opacity_tgt, OPAQUE);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (w->opacity != w->opacity_tgt) {
|
|
|
|
ps->idling = false;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Set fade callback of a window, and possibly execute the previous
|
|
|
|
* callback.
|
|
|
|
*
|
|
|
|
* @param exec_callback whether the previous callback is to be executed
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
set_fade_callback(session_t *ps, win *w,
|
|
|
|
void (*callback) (session_t *ps, win *w), bool exec_callback) {
|
|
|
|
void (*old_callback) (session_t *ps, win *w) = w->fade_callback;
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
|
|
|
|
w->fade_callback = callback;
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
// Must be the last line as the callback could destroy w!
|
|
|
|
if (exec_callback && old_callback) {
|
|
|
|
old_callback(ps, w);
|
|
|
|
// Although currently no callback function affects window state on
|
|
|
|
// next paint, it could, in the future
|
|
|
|
ps->idling = false;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
// === Shadows ===
|
|
|
|
|
|
|
|
static double __attribute__((const))
|
|
|
|
gaussian(double r, double x, double y) {
|
|
|
|
return ((1 / (sqrt(2 * M_PI * r))) *
|
|
|
|
exp((- (x * x + y * y)) / (2 * r * r)));
|
|
|
|
}
|
|
|
|
|
|
|
|
static conv *
|
|
|
|
make_gaussian_map(double r) {
|
|
|
|
conv *c;
|
|
|
|
int size = ((int) ceil((r * 3)) + 1) & ~1;
|
|
|
|
int center = size / 2;
|
|
|
|
int x, y;
|
|
|
|
double t;
|
|
|
|
double g;
|
|
|
|
|
|
|
|
c = malloc(sizeof(conv) + size * size * sizeof(double));
|
|
|
|
c->size = size;
|
|
|
|
c->data = (double *) (c + 1);
|
|
|
|
t = 0.0;
|
|
|
|
|
|
|
|
for (y = 0; y < size; y++) {
|
|
|
|
for (x = 0; x < size; x++) {
|
|
|
|
g = gaussian(r, x - center, y - center);
|
|
|
|
t += g;
|
|
|
|
c->data[y * size + x] = g;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
for (y = 0; y < size; y++) {
|
|
|
|
for (x = 0; x < size; x++) {
|
|
|
|
c->data[y * size + x] /= t;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return c;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* A picture will help
|
|
|
|
*
|
|
|
|
* -center 0 width width+center
|
|
|
|
* -center +-----+-------------------+-----+
|
|
|
|
* | | | |
|
|
|
|
* | | | |
|
|
|
|
* 0 +-----+-------------------+-----+
|
|
|
|
* | | | |
|
|
|
|
* | | | |
|
|
|
|
* | | | |
|
|
|
|
* height +-----+-------------------+-----+
|
|
|
|
* | | | |
|
|
|
|
* height+ | | | |
|
|
|
|
* center +-----+-------------------+-----+
|
|
|
|
*/
|
|
|
|
|
|
|
|
static unsigned char
|
|
|
|
sum_gaussian(conv *map, double opacity,
|
|
|
|
int x, int y, int width, int height) {
|
|
|
|
int fx, fy;
|
|
|
|
double *g_data;
|
|
|
|
double *g_line = map->data;
|
|
|
|
int g_size = map->size;
|
|
|
|
int center = g_size / 2;
|
|
|
|
int fx_start, fx_end;
|
|
|
|
int fy_start, fy_end;
|
|
|
|
double v;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Compute set of filter values which are "in range",
|
|
|
|
* that's the set with:
|
|
|
|
* 0 <= x + (fx-center) && x + (fx-center) < width &&
|
|
|
|
* 0 <= y + (fy-center) && y + (fy-center) < height
|
|
|
|
*
|
|
|
|
* 0 <= x + (fx - center) x + fx - center < width
|
|
|
|
* center - x <= fx fx < width + center - x
|
|
|
|
*/
|
|
|
|
|
|
|
|
fx_start = center - x;
|
|
|
|
if (fx_start < 0) fx_start = 0;
|
|
|
|
fx_end = width + center - x;
|
|
|
|
if (fx_end > g_size) fx_end = g_size;
|
|
|
|
|
|
|
|
fy_start = center - y;
|
|
|
|
if (fy_start < 0) fy_start = 0;
|
|
|
|
fy_end = height + center - y;
|
|
|
|
if (fy_end > g_size) fy_end = g_size;
|
|
|
|
|
|
|
|
g_line = g_line + fy_start * g_size + fx_start;
|
|
|
|
|
|
|
|
v = 0;
|
|
|
|
|
|
|
|
for (fy = fy_start; fy < fy_end; fy++) {
|
|
|
|
g_data = g_line;
|
|
|
|
g_line += g_size;
|
|
|
|
|
|
|
|
for (fx = fx_start; fx < fx_end; fx++) {
|
|
|
|
v += *g_data++;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
if (v > 1) v = 1;
|
|
|
|
|
|
|
|
return ((unsigned char) (v * opacity * 255.0));
|
|
|
|
}
|
|
|
|
|
|
|
|
/* precompute shadow corners and sides
|
|
|
|
to save time for large windows */
|
|
|
|
|
|
|
|
static void
|
|
|
|
presum_gaussian(session_t *ps, conv *map) {
|
|
|
|
int center = map->size / 2;
|
|
|
|
int opacity, x, y;
|
|
|
|
|
|
|
|
ps->cgsize = map->size;
|
|
|
|
|
|
|
|
if (ps->shadow_corner)
|
|
|
|
free(ps->shadow_corner);
|
|
|
|
if (ps->shadow_top)
|
|
|
|
free(ps->shadow_top);
|
|
|
|
|
|
|
|
ps->shadow_corner = malloc((ps->cgsize + 1) * (ps->cgsize + 1) * 26);
|
|
|
|
ps->shadow_top = malloc((ps->cgsize + 1) * 26);
|
|
|
|
|
|
|
|
for (x = 0; x <= ps->cgsize; x++) {
|
|
|
|
ps->shadow_top[25 * (ps->cgsize + 1) + x] =
|
|
|
|
sum_gaussian(map, 1, x - center, center, ps->cgsize * 2, ps->cgsize * 2);
|
|
|
|
|
|
|
|
for (opacity = 0; opacity < 25; opacity++) {
|
|
|
|
ps->shadow_top[opacity * (ps->cgsize + 1) + x] =
|
|
|
|
ps->shadow_top[25 * (ps->cgsize + 1) + x] * opacity / 25;
|
|
|
|
}
|
|
|
|
|
|
|
|
for (y = 0; y <= x; y++) {
|
|
|
|
ps->shadow_corner[25 * (ps->cgsize + 1) * (ps->cgsize + 1) + y * (ps->cgsize + 1) + x]
|
|
|
|
= sum_gaussian(map, 1, x - center, y - center, ps->cgsize * 2, ps->cgsize * 2);
|
|
|
|
ps->shadow_corner[25 * (ps->cgsize + 1) * (ps->cgsize + 1) + x * (ps->cgsize + 1) + y]
|
|
|
|
= ps->shadow_corner[25 * (ps->cgsize + 1) * (ps->cgsize + 1) + y * (ps->cgsize + 1) + x];
|
|
|
|
|
|
|
|
for (opacity = 0; opacity < 25; opacity++) {
|
|
|
|
ps->shadow_corner[opacity * (ps->cgsize + 1) * (ps->cgsize + 1)
|
|
|
|
+ y * (ps->cgsize + 1) + x]
|
|
|
|
= ps->shadow_corner[opacity * (ps->cgsize + 1) * (ps->cgsize + 1)
|
|
|
|
+ x * (ps->cgsize + 1) + y]
|
|
|
|
= ps->shadow_corner[25 * (ps->cgsize + 1) * (ps->cgsize + 1)
|
|
|
|
+ y * (ps->cgsize + 1) + x] * opacity / 25;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static XImage *
|
|
|
|
make_shadow(session_t *ps, double opacity,
|
|
|
|
int width, int height) {
|
|
|
|
XImage *ximage;
|
|
|
|
unsigned char *data;
|
|
|
|
int ylimit, xlimit;
|
|
|
|
int swidth = width + ps->cgsize;
|
|
|
|
int sheight = height + ps->cgsize;
|
|
|
|
int center = ps->cgsize / 2;
|
|
|
|
int x, y;
|
|
|
|
unsigned char d;
|
|
|
|
int x_diff;
|
|
|
|
int opacity_int = (int)(opacity * 25);
|
|
|
|
|
|
|
|
data = malloc(swidth * sheight * sizeof(unsigned char));
|
|
|
|
if (!data) return 0;
|
|
|
|
|
|
|
|
ximage = XCreateImage(ps->dpy, ps->vis, 8,
|
|
|
|
ZPixmap, 0, (char *) data, swidth, sheight, 8, swidth * sizeof(char));
|
|
|
|
|
|
|
|
if (!ximage) {
|
|
|
|
free(data);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Build the gaussian in sections
|
|
|
|
*/
|
|
|
|
|
|
|
|
/*
|
|
|
|
* center (fill the complete data array)
|
|
|
|
*/
|
|
|
|
|
|
|
|
// If clear_shadow is enabled and the border & corner shadow (which
|
|
|
|
// later will be filled) could entirely cover the area of the shadow
|
|
|
|
// that will be displayed, do not bother filling other pixels. If it
|
|
|
|
// can't, we must fill the other pixels here.
|
|
|
|
/* if (!(clear_shadow && ps->o.shadow_offset_x <= 0 && ps->o.shadow_offset_x >= -ps->cgsize
|
|
|
|
&& ps->o.shadow_offset_y <= 0 && ps->o.shadow_offset_y >= -ps->cgsize)) { */
|
|
|
|
if (ps->cgsize > 0) {
|
|
|
|
d = ps->shadow_top[opacity_int * (ps->cgsize + 1) + ps->cgsize];
|
|
|
|
} else {
|
|
|
|
d = sum_gaussian(ps->gaussian_map,
|
|
|
|
opacity, center, center, width, height);
|
|
|
|
}
|
|
|
|
memset(data, d, sheight * swidth);
|
|
|
|
// }
|
|
|
|
|
|
|
|
/*
|
|
|
|
* corners
|
|
|
|
*/
|
|
|
|
|
|
|
|
ylimit = ps->cgsize;
|
|
|
|
if (ylimit > sheight / 2) ylimit = (sheight + 1) / 2;
|
|
|
|
|
|
|
|
xlimit = ps->cgsize;
|
|
|
|
if (xlimit > swidth / 2) xlimit = (swidth + 1) / 2;
|
|
|
|
|
|
|
|
for (y = 0; y < ylimit; y++) {
|
|
|
|
for (x = 0; x < xlimit; x++) {
|
|
|
|
if (xlimit == ps->cgsize && ylimit == ps->cgsize) {
|
|
|
|
d = ps->shadow_corner[opacity_int * (ps->cgsize + 1) * (ps->cgsize + 1)
|
|
|
|
+ y * (ps->cgsize + 1) + x];
|
|
|
|
} else {
|
|
|
|
d = sum_gaussian(ps->gaussian_map,
|
|
|
|
opacity, x - center, y - center, width, height);
|
|
|
|
}
|
|
|
|
data[y * swidth + x] = d;
|
|
|
|
data[(sheight - y - 1) * swidth + x] = d;
|
|
|
|
data[(sheight - y - 1) * swidth + (swidth - x - 1)] = d;
|
|
|
|
data[y * swidth + (swidth - x - 1)] = d;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* top/bottom
|
|
|
|
*/
|
|
|
|
|
|
|
|
x_diff = swidth - (ps->cgsize * 2);
|
|
|
|
if (x_diff > 0 && ylimit > 0) {
|
|
|
|
for (y = 0; y < ylimit; y++) {
|
|
|
|
if (ylimit == ps->cgsize) {
|
|
|
|
d = ps->shadow_top[opacity_int * (ps->cgsize + 1) + y];
|
|
|
|
} else {
|
|
|
|
d = sum_gaussian(ps->gaussian_map,
|
|
|
|
opacity, center, y - center, width, height);
|
|
|
|
}
|
|
|
|
memset(&data[y * swidth + ps->cgsize], d, x_diff);
|
|
|
|
memset(&data[(sheight - y - 1) * swidth + ps->cgsize], d, x_diff);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* sides
|
|
|
|
*/
|
|
|
|
|
|
|
|
for (x = 0; x < xlimit; x++) {
|
|
|
|
if (xlimit == ps->cgsize) {
|
|
|
|
d = ps->shadow_top[opacity_int * (ps->cgsize + 1) + x];
|
|
|
|
} else {
|
|
|
|
d = sum_gaussian(ps->gaussian_map,
|
|
|
|
opacity, x - center, center, width, height);
|
|
|
|
}
|
|
|
|
for (y = ps->cgsize; y < sheight - ps->cgsize; y++) {
|
|
|
|
data[y * swidth + x] = d;
|
|
|
|
data[y * swidth + (swidth - x - 1)] = d;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
if (clear_shadow) {
|
|
|
|
// Clear the region in the shadow that the window would cover based
|
|
|
|
// on shadow_offset_{x,y} user provides
|
|
|
|
int xstart = normalize_i_range(- (int) ps->o.shadow_offset_x, 0, swidth);
|
|
|
|
int xrange = normalize_i_range(width - (int) ps->o.shadow_offset_x,
|
|
|
|
0, swidth) - xstart;
|
|
|
|
int ystart = normalize_i_range(- (int) ps->o.shadow_offset_y, 0, sheight);
|
|
|
|
int yend = normalize_i_range(height - (int) ps->o.shadow_offset_y,
|
|
|
|
0, sheight);
|
|
|
|
int y;
|
|
|
|
|
|
|
|
for (y = ystart; y < yend; y++) {
|
|
|
|
memset(&data[y * swidth + xstart], 0, xrange);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
*/
|
|
|
|
|
|
|
|
return ximage;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Generate shadow <code>Picture</code> for a window.
|
|
|
|
*/
|
|
|
|
static Picture
|
|
|
|
shadow_picture(session_t *ps, double opacity, int width, int height) {
|
|
|
|
XImage *shadow_image = NULL;
|
|
|
|
Pixmap shadow_pixmap = None, shadow_pixmap_argb = None;
|
|
|
|
Picture shadow_picture = None, shadow_picture_argb = None;
|
|
|
|
GC gc = None;
|
|
|
|
|
|
|
|
shadow_image = make_shadow(ps, opacity, width, height);
|
|
|
|
if (!shadow_image)
|
|
|
|
return None;
|
|
|
|
|
|
|
|
shadow_pixmap = XCreatePixmap(ps->dpy, ps->root,
|
|
|
|
shadow_image->width, shadow_image->height, 8);
|
|
|
|
shadow_pixmap_argb = XCreatePixmap(ps->dpy, ps->root,
|
|
|
|
shadow_image->width, shadow_image->height, 32);
|
|
|
|
|
|
|
|
if (!shadow_pixmap || !shadow_pixmap_argb)
|
|
|
|
goto shadow_picture_err;
|
|
|
|
|
|
|
|
shadow_picture = XRenderCreatePicture(ps->dpy, shadow_pixmap,
|
|
|
|
XRenderFindStandardFormat(ps->dpy, PictStandardA8), 0, 0);
|
|
|
|
shadow_picture_argb = XRenderCreatePicture(ps->dpy, shadow_pixmap_argb,
|
|
|
|
XRenderFindStandardFormat(ps->dpy, PictStandardARGB32), 0, 0);
|
|
|
|
if (!shadow_picture || !shadow_picture_argb)
|
|
|
|
goto shadow_picture_err;
|
|
|
|
|
|
|
|
gc = XCreateGC(ps->dpy, shadow_pixmap, 0, 0);
|
|
|
|
if (!gc)
|
|
|
|
goto shadow_picture_err;
|
|
|
|
|
|
|
|
XPutImage(ps->dpy, shadow_pixmap, gc, shadow_image, 0, 0, 0, 0,
|
|
|
|
shadow_image->width, shadow_image->height);
|
|
|
|
XRenderComposite(ps->dpy, PictOpSrc, ps->cshadow_picture, shadow_picture,
|
|
|
|
shadow_picture_argb, 0, 0, 0, 0, 0, 0,
|
|
|
|
shadow_image->width, shadow_image->height);
|
|
|
|
|
|
|
|
XFreeGC(ps->dpy, gc);
|
|
|
|
XDestroyImage(shadow_image);
|
|
|
|
XFreePixmap(ps->dpy, shadow_pixmap);
|
|
|
|
XFreePixmap(ps->dpy, shadow_pixmap_argb);
|
|
|
|
XRenderFreePicture(ps->dpy, shadow_picture);
|
|
|
|
|
|
|
|
return shadow_picture_argb;
|
|
|
|
|
|
|
|
shadow_picture_err:
|
|
|
|
if (shadow_image)
|
|
|
|
XDestroyImage(shadow_image);
|
|
|
|
if (shadow_pixmap)
|
|
|
|
XFreePixmap(ps->dpy, shadow_pixmap);
|
|
|
|
if (shadow_pixmap_argb)
|
|
|
|
XFreePixmap(ps->dpy, shadow_pixmap_argb);
|
|
|
|
if (shadow_picture)
|
|
|
|
XRenderFreePicture(ps->dpy, shadow_picture);
|
|
|
|
if (shadow_picture_argb)
|
|
|
|
XRenderFreePicture(ps->dpy, shadow_picture_argb);
|
|
|
|
if (gc)
|
|
|
|
XFreeGC(ps->dpy, gc);
|
|
|
|
return None;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Generate a 1x1 <code>Picture</code> of a particular color.
|
|
|
|
*/
|
|
|
|
static Picture
|
|
|
|
solid_picture(session_t *ps, bool argb, double a,
|
|
|
|
double r, double g, double b) {
|
|
|
|
Pixmap pixmap;
|
|
|
|
Picture picture;
|
|
|
|
XRenderPictureAttributes pa;
|
|
|
|
XRenderColor c;
|
|
|
|
|
|
|
|
pixmap = XCreatePixmap(ps->dpy, ps->root, 1, 1, argb ? 32 : 8);
|
|
|
|
|
|
|
|
if (!pixmap) return None;
|
|
|
|
|
|
|
|
pa.repeat = True;
|
|
|
|
picture = XRenderCreatePicture(ps->dpy, pixmap,
|
|
|
|
XRenderFindStandardFormat(ps->dpy, argb
|
|
|
|
? PictStandardARGB32 : PictStandardA8),
|
|
|
|
CPRepeat,
|
|
|
|
&pa);
|
|
|
|
|
|
|
|
if (!picture) {
|
|
|
|
XFreePixmap(ps->dpy, pixmap);
|
|
|
|
return None;
|
|
|
|
}
|
|
|
|
|
|
|
|
c.alpha = a * 0xffff;
|
|
|
|
c.red = r * 0xffff;
|
|
|
|
c.green = g * 0xffff;
|
|
|
|
c.blue = b * 0xffff;
|
|
|
|
|
|
|
|
XRenderFillRectangle(ps->dpy, PictOpSrc, picture, &c, 0, 0, 1, 1);
|
|
|
|
XFreePixmap(ps->dpy, pixmap);
|
|
|
|
|
|
|
|
return picture;
|
|
|
|
}
|
|
|
|
|
|
|
|
// === Error handling ===
|
|
|
|
|
|
|
|
static void
|
|
|
|
discard_ignore(session_t *ps, unsigned long sequence) {
|
|
|
|
while (ps->ignore_head) {
|
|
|
|
if ((long) (sequence - ps->ignore_head->sequence) > 0) {
|
|
|
|
ignore_t *next = ps->ignore_head->next;
|
|
|
|
free(ps->ignore_head);
|
|
|
|
ps->ignore_head = next;
|
|
|
|
if (!ps->ignore_head) {
|
|
|
|
ps->ignore_tail = &ps->ignore_head;
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
set_ignore(session_t *ps, unsigned long sequence) {
|
|
|
|
ignore_t *i = malloc(sizeof(ignore_t));
|
|
|
|
if (!i) return;
|
|
|
|
|
|
|
|
i->sequence = sequence;
|
|
|
|
i->next = 0;
|
|
|
|
*ps->ignore_tail = i;
|
|
|
|
ps->ignore_tail = &i->next;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int
|
|
|
|
should_ignore(session_t *ps, unsigned long sequence) {
|
|
|
|
discard_ignore(ps, sequence);
|
|
|
|
return ps->ignore_head && ps->ignore_head->sequence == sequence;
|
|
|
|
}
|
|
|
|
|
|
|
|
// === Windows ===
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Check if a window has rounded corners.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
win_rounded_corners(session_t *ps, win *w) {
|
|
|
|
if (!w->bounding_shaped)
|
|
|
|
return;
|
|
|
|
|
|
|
|
// Fetch its bounding region
|
|
|
|
if (!w->border_size)
|
|
|
|
w->border_size = border_size(ps, w, true);
|
|
|
|
|
|
|
|
// Quit if border_size() returns None
|
|
|
|
if (!w->border_size)
|
|
|
|
return;
|
|
|
|
|
|
|
|
// Determine the minimum width/height of a rectangle that could mark
|
|
|
|
// a window as having rounded corners
|
|
|
|
unsigned short minwidth = max_i(w->widthb * (1 - ROUNDED_PERCENT),
|
|
|
|
w->widthb - ROUNDED_PIXELS);
|
|
|
|
unsigned short minheight = max_i(w->heightb * (1 - ROUNDED_PERCENT),
|
|
|
|
w->heightb - ROUNDED_PIXELS);
|
|
|
|
|
|
|
|
// Get the rectangles in the bounding region
|
|
|
|
int nrects = 0, i;
|
|
|
|
XRectangle *rects = XFixesFetchRegion(ps->dpy, w->border_size, &nrects);
|
|
|
|
if (!rects)
|
|
|
|
return;
|
|
|
|
|
|
|
|
// Look for a rectangle large enough for this window be considered
|
|
|
|
// having rounded corners
|
|
|
|
for (i = 0; i < nrects; ++i)
|
|
|
|
if (rects[i].width >= minwidth && rects[i].height >= minheight) {
|
|
|
|
w->rounded_corners = true;
|
|
|
|
XFree(rects);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
w->rounded_corners = false;
|
|
|
|
XFree(rects);
|
|
|
|
}
|
|
|
|
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
/**
|
|
|
|
* Match a window against a single window condition.
|
|
|
|
*
|
|
|
|
* @return true if matched, false otherwise.
|
|
|
|
*/
|
|
|
|
static bool
|
|
|
|
win_match_once(win *w, const wincond_t *cond) {
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
const char *target;
|
|
|
|
bool matched = false;
|
|
|
|
|
|
|
|
#ifdef DEBUG_WINMATCH
|
|
|
|
printf("win_match_once(%#010lx \"%s\"): cond = %p", w->id, w->name,
|
|
|
|
cond);
|
|
|
|
#endif
|
|
|
|
|
|
|
|
if (InputOnly == w->a.class) {
|
|
|
|
#ifdef DEBUG_WINMATCH
|
|
|
|
printf(": InputOnly\n");
|
|
|
|
#endif
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
// Determine the target
|
|
|
|
target = NULL;
|
|
|
|
switch (cond->target) {
|
|
|
|
case CONDTGT_NAME:
|
|
|
|
target = w->name;
|
|
|
|
break;
|
|
|
|
case CONDTGT_CLASSI:
|
|
|
|
target = w->class_instance;
|
|
|
|
break;
|
|
|
|
case CONDTGT_CLASSG:
|
|
|
|
target = w->class_general;
|
|
|
|
break;
|
|
|
|
case CONDTGT_ROLE:
|
|
|
|
target = w->role;
|
|
|
|
break;
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
}
|
|
|
|
|
|
|
|
if (!target) {
|
|
|
|
#ifdef DEBUG_WINMATCH
|
|
|
|
printf(": Target not found\n");
|
|
|
|
#endif
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
|
|
|
// Determine pattern type and match
|
|
|
|
switch (cond->type) {
|
|
|
|
case CONDTP_EXACT:
|
|
|
|
if (cond->flags & CONDF_IGNORECASE)
|
|
|
|
matched = !strcasecmp(target, cond->pattern);
|
|
|
|
else
|
|
|
|
matched = !strcmp(target, cond->pattern);
|
|
|
|
break;
|
|
|
|
case CONDTP_ANYWHERE:
|
|
|
|
if (cond->flags & CONDF_IGNORECASE)
|
|
|
|
matched = strcasestr(target, cond->pattern);
|
|
|
|
else
|
|
|
|
matched = strstr(target, cond->pattern);
|
|
|
|
break;
|
|
|
|
case CONDTP_FROMSTART:
|
|
|
|
if (cond->flags & CONDF_IGNORECASE)
|
|
|
|
matched = !strncasecmp(target, cond->pattern,
|
|
|
|
strlen(cond->pattern));
|
|
|
|
else
|
|
|
|
matched = !strncmp(target, cond->pattern,
|
|
|
|
strlen(cond->pattern));
|
|
|
|
break;
|
|
|
|
case CONDTP_WILDCARD:
|
|
|
|
{
|
|
|
|
int flags = 0;
|
|
|
|
if (cond->flags & CONDF_IGNORECASE)
|
|
|
|
flags = FNM_CASEFOLD;
|
|
|
|
matched = !fnmatch(cond->pattern, target, flags);
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
case CONDTP_REGEX_PCRE:
|
|
|
|
#ifdef CONFIG_REGEX_PCRE
|
|
|
|
matched = (pcre_exec(cond->regex_pcre, cond->regex_pcre_extra,
|
|
|
|
target, strlen(target), 0, 0, NULL, 0) >= 0);
|
|
|
|
#endif
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
#ifdef DEBUG_WINMATCH
|
|
|
|
printf(", matched = %d\n", matched);
|
|
|
|
#endif
|
|
|
|
|
|
|
|
return matched;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Match a window against a condition linked list.
|
|
|
|
*
|
|
|
|
* @param cache a place to cache the last matched condition
|
|
|
|
* @return true if matched, false otherwise.
|
|
|
|
*/
|
|
|
|
static bool
|
|
|
|
win_match(win *w, wincond_t *condlst, wincond_t **cache) {
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
// Check if the cached entry matches firstly
|
|
|
|
if (cache && *cache && win_match_once(w, *cache))
|
|
|
|
return true;
|
|
|
|
|
|
|
|
// Then go through the whole linked list
|
|
|
|
for (; condlst; condlst = condlst->next) {
|
|
|
|
if (win_match_once(w, condlst)) {
|
|
|
|
if (cache)
|
|
|
|
*cache = condlst;
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
return true;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Add a pattern to a condition linked list.
|
|
|
|
*/
|
|
|
|
static bool
|
|
|
|
condlst_add(wincond_t **pcondlst, const char *pattern) {
|
|
|
|
if (!pattern)
|
|
|
|
return false;
|
|
|
|
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
unsigned plen = strlen(pattern);
|
|
|
|
wincond_t *cond;
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
const char *pos;
|
|
|
|
|
|
|
|
if (plen < 4 || ':' != pattern[1] || !strchr(pattern + 2, ':')) {
|
|
|
|
printf("Pattern \"%s\": Format invalid.\n", pattern);
|
|
|
|
return false;
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
}
|
|
|
|
|
|
|
|
// Allocate memory for the new condition
|
|
|
|
cond = malloc(sizeof(wincond_t));
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
|
|
|
|
// Determine the pattern target
|
|
|
|
switch (pattern[0]) {
|
|
|
|
case 'n':
|
|
|
|
cond->target = CONDTGT_NAME;
|
|
|
|
break;
|
|
|
|
case 'i':
|
|
|
|
cond->target = CONDTGT_CLASSI;
|
|
|
|
break;
|
|
|
|
case 'g':
|
|
|
|
cond->target = CONDTGT_CLASSG;
|
|
|
|
break;
|
|
|
|
case 'r':
|
|
|
|
cond->target = CONDTGT_ROLE;
|
|
|
|
break;
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
default:
|
|
|
|
printf("Pattern \"%s\": Target \"%c\" invalid.\n",
|
|
|
|
pattern, pattern[0]);
|
|
|
|
free(cond);
|
|
|
|
return false;
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
}
|
|
|
|
|
|
|
|
// Determine the pattern type
|
|
|
|
switch (pattern[2]) {
|
|
|
|
case 'e':
|
|
|
|
cond->type = CONDTP_EXACT;
|
|
|
|
break;
|
|
|
|
case 'a':
|
|
|
|
cond->type = CONDTP_ANYWHERE;
|
|
|
|
break;
|
|
|
|
case 's':
|
|
|
|
cond->type = CONDTP_FROMSTART;
|
|
|
|
break;
|
|
|
|
case 'w':
|
|
|
|
cond->type = CONDTP_WILDCARD;
|
|
|
|
break;
|
|
|
|
#ifdef CONFIG_REGEX_PCRE
|
|
|
|
case 'p':
|
|
|
|
cond->type = CONDTP_REGEX_PCRE;
|
|
|
|
break;
|
|
|
|
#endif
|
|
|
|
default:
|
|
|
|
printf("Pattern \"%s\": Type \"%c\" invalid.\n",
|
|
|
|
pattern, pattern[2]);
|
|
|
|
free(cond);
|
|
|
|
return false;
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
}
|
|
|
|
|
|
|
|
// Determine the pattern flags
|
|
|
|
pos = &pattern[3];
|
|
|
|
cond->flags = 0;
|
|
|
|
while (':' != *pos) {
|
|
|
|
switch (*pos) {
|
|
|
|
case 'i':
|
|
|
|
cond->flags |= CONDF_IGNORECASE;
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
printf("Pattern \"%s\": Flag \"%c\" invalid.\n",
|
|
|
|
pattern, *pos);
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
++pos;
|
|
|
|
}
|
|
|
|
|
|
|
|
// Copy the pattern
|
|
|
|
++pos;
|
|
|
|
cond->pattern = NULL;
|
|
|
|
#ifdef CONFIG_REGEX_PCRE
|
|
|
|
cond->regex_pcre = NULL;
|
|
|
|
cond->regex_pcre_extra = NULL;
|
|
|
|
#endif
|
|
|
|
if (CONDTP_REGEX_PCRE == cond->type) {
|
|
|
|
#ifdef CONFIG_REGEX_PCRE
|
|
|
|
const char *error = NULL;
|
|
|
|
int erroffset = 0;
|
|
|
|
int options = 0;
|
|
|
|
|
|
|
|
if (cond->flags & CONDF_IGNORECASE)
|
|
|
|
options |= PCRE_CASELESS;
|
|
|
|
|
|
|
|
cond->regex_pcre = pcre_compile(pos, options, &error, &erroffset,
|
|
|
|
NULL);
|
|
|
|
if (!cond->regex_pcre) {
|
|
|
|
printf("Pattern \"%s\": PCRE regular expression parsing failed on "
|
|
|
|
"offset %d: %s\n", pattern, erroffset, error);
|
|
|
|
free(cond);
|
|
|
|
return false;
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
}
|
|
|
|
#ifdef CONFIG_REGEX_PCRE_JIT
|
|
|
|
cond->regex_pcre_extra = pcre_study(cond->regex_pcre, PCRE_STUDY_JIT_COMPILE, &error);
|
|
|
|
if (!cond->regex_pcre_extra) {
|
|
|
|
printf("Pattern \"%s\": PCRE regular expression study failed: %s",
|
|
|
|
pattern, error);
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
#endif
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
cond->pattern = mstrcpy(pos);
|
|
|
|
}
|
|
|
|
|
|
|
|
// Insert it into the linked list
|
|
|
|
cond->next = *pcondlst;
|
|
|
|
*pcondlst = cond;
|
|
|
|
|
|
|
|
return true;
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Determine the event mask for a window.
|
|
|
|
*/
|
|
|
|
static long
|
|
|
|
determine_evmask(session_t *ps, Window wid, win_evmode_t mode) {
|
|
|
|
long evmask = NoEventMask;
|
|
|
|
win *w = NULL;
|
|
|
|
|
|
|
|
// Check if it's a mapped frame window
|
|
|
|
if (WIN_EVMODE_FRAME == mode
|
|
|
|
|| ((w = find_win(ps, wid)) && IsViewable == w->a.map_state)) {
|
|
|
|
evmask |= PropertyChangeMask;
|
|
|
|
if (ps->o.track_focus && !ps->o.use_ewmh_active_win)
|
|
|
|
evmask |= FocusChangeMask;
|
|
|
|
}
|
|
|
|
|
|
|
|
// Check if it's a mapped client window
|
|
|
|
if (WIN_EVMODE_CLIENT == mode
|
|
|
|
|| ((w = find_toplevel(ps, wid)) && IsViewable == w->a.map_state)) {
|
|
|
|
if (ps->o.frame_opacity || ps->o.track_wdata
|
|
|
|
|| ps->o.detect_client_opacity)
|
|
|
|
evmask |= PropertyChangeMask;
|
|
|
|
}
|
|
|
|
|
|
|
|
return evmask;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Find out the WM frame of a client window by querying X.
|
|
|
|
*
|
|
|
|
* @param ps current session
|
|
|
|
* @param w window ID
|
|
|
|
* @return struct _win object of the found window, NULL if not found
|
|
|
|
*/
|
|
|
|
static win *
|
|
|
|
find_toplevel2(session_t *ps, Window wid) {
|
|
|
|
win *w = NULL;
|
|
|
|
|
|
|
|
// We traverse through its ancestors to find out the frame
|
|
|
|
while (wid && wid != ps->root && !(w = find_win(ps, wid))) {
|
|
|
|
Window troot;
|
|
|
|
Window parent;
|
|
|
|
Window *tchildren;
|
|
|
|
unsigned tnchildren;
|
|
|
|
|
|
|
|
// XQueryTree probably fails if you run compton when X is somehow
|
|
|
|
// initializing (like add it in .xinitrc). In this case
|
|
|
|
// just leave it alone.
|
|
|
|
if (!XQueryTree(ps->dpy, wid, &troot, &parent, &tchildren,
|
|
|
|
&tnchildren)) {
|
|
|
|
parent = 0;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (tchildren) XFree(tchildren);
|
|
|
|
|
|
|
|
wid = parent;
|
|
|
|
}
|
|
|
|
|
|
|
|
return w;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Recheck currently focused window and set its <code>w->focused</code>
|
|
|
|
* to true.
|
|
|
|
*
|
|
|
|
* @param ps current session
|
|
|
|
* @return struct _win of currently focused window, NULL if not found
|
|
|
|
*/
|
|
|
|
static win *
|
|
|
|
recheck_focus(session_t *ps) {
|
|
|
|
// Use EWMH _NET_ACTIVE_WINDOW if enabled
|
|
|
|
if (ps->o.use_ewmh_active_win) {
|
|
|
|
update_ewmh_active_win(ps);
|
|
|
|
return ps->active_win;
|
|
|
|
}
|
|
|
|
|
|
|
|
// Determine the currently focused window so we can apply appropriate
|
|
|
|
// opacity on it
|
|
|
|
Window wid = 0;
|
|
|
|
int revert_to;
|
|
|
|
win *w = NULL;
|
|
|
|
|
|
|
|
XGetInputFocus(ps->dpy, &wid, &revert_to);
|
|
|
|
|
|
|
|
if (!wid || PointerRoot == wid)
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
// Fallback to the old method if find_toplevel() fails
|
|
|
|
if (!(w = find_toplevel(ps, wid))) {
|
|
|
|
w = find_toplevel2(ps, wid);
|
|
|
|
}
|
|
|
|
|
|
|
|
// And we set the focus state and opacity here
|
|
|
|
if (w) {
|
|
|
|
win_set_focused(ps, w, true);
|
|
|
|
return w;
|
|
|
|
}
|
|
|
|
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
static Picture
|
|
|
|
root_tile_f(session_t *ps) {
|
|
|
|
/*
|
|
|
|
if (ps->o.paint_on_overlay) {
|
|
|
|
return ps->root_picture;
|
|
|
|
} */
|
|
|
|
|
|
|
|
Picture picture;
|
|
|
|
Pixmap pixmap;
|
|
|
|
bool fill = false;
|
|
|
|
XRenderPictureAttributes pa;
|
|
|
|
int p;
|
|
|
|
|
|
|
|
pixmap = None;
|
|
|
|
|
|
|
|
// Get the values of background attributes
|
|
|
|
for (p = 0; background_props_str[p]; p++) {
|
|
|
|
winprop_t prop = wid_get_prop(ps, ps->root,
|
|
|
|
get_atom(ps, background_props_str[p]),
|
|
|
|
1L, XA_PIXMAP, 32);
|
|
|
|
if (prop.nitems) {
|
|
|
|
pixmap = *prop.data.p32;
|
|
|
|
fill = false;
|
|
|
|
free_winprop(&prop);
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
free_winprop(&prop);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (!pixmap) {
|
|
|
|
pixmap = XCreatePixmap(ps->dpy, ps->root, 1, 1, ps->depth);
|
|
|
|
fill = true;
|
|
|
|
}
|
|
|
|
|
|
|
|
pa.repeat = True;
|
|
|
|
picture = XRenderCreatePicture(
|
|
|
|
ps->dpy, pixmap, XRenderFindVisualFormat(ps->dpy, ps->vis),
|
|
|
|
CPRepeat, &pa);
|
|
|
|
|
|
|
|
if (fill) {
|
|
|
|
XRenderColor c;
|
|
|
|
|
|
|
|
c.red = c.green = c.blue = 0x8080;
|
|
|
|
c.alpha = 0xffff;
|
|
|
|
XRenderFillRectangle(
|
|
|
|
ps->dpy, PictOpSrc, picture, &c, 0, 0, 1, 1);
|
|
|
|
}
|
|
|
|
|
|
|
|
return picture;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Paint root window content.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
paint_root(session_t *ps, Picture tgt_buffer) {
|
|
|
|
if (!ps->root_tile) {
|
|
|
|
ps->root_tile = root_tile_f(ps);
|
|
|
|
}
|
|
|
|
|
|
|
|
XRenderComposite(
|
|
|
|
ps->dpy, PictOpSrc, ps->root_tile, None,
|
|
|
|
tgt_buffer, 0, 0, 0, 0, 0, 0,
|
|
|
|
ps->root_width, ps->root_height);
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Get a rectangular region a window occupies, excluding shadow.
|
|
|
|
*/
|
|
|
|
static XserverRegion
|
|
|
|
win_get_region(session_t *ps, win *w, bool use_offset) {
|
|
|
|
XRectangle r;
|
|
|
|
|
|
|
|
r.x = (use_offset ? w->a.x: 0);
|
|
|
|
r.y = (use_offset ? w->a.y: 0);
|
|
|
|
r.width = w->widthb;
|
|
|
|
r.height = w->heightb;
|
|
|
|
|
|
|
|
return XFixesCreateRegion(ps->dpy, &r, 1);
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Get a rectangular region a window occupies, excluding frame and shadow.
|
|
|
|
*/
|
|
|
|
static XserverRegion
|
|
|
|
win_get_region_noframe(session_t *ps, win *w, bool use_offset) {
|
|
|
|
XRectangle r;
|
|
|
|
|
|
|
|
r.x = (use_offset ? w->a.x: 0) + w->a.border_width + w->left_width;
|
|
|
|
r.y = (use_offset ? w->a.y: 0) + w->a.border_width + w->top_width;
|
|
|
|
r.width = max_i(w->a.width - w->left_width - w->right_width, 0);
|
|
|
|
r.height = max_i(w->a.height - w->top_width - w->bottom_width, 0);
|
|
|
|
|
|
|
|
if (r.width > 0 && r.height > 0)
|
|
|
|
return XFixesCreateRegion(ps->dpy, &r, 1);
|
|
|
|
else
|
|
|
|
return XFixesCreateRegion(ps->dpy, NULL, 0);
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Get a rectangular region a window (and possibly its shadow) occupies.
|
|
|
|
*
|
|
|
|
* Note w->shadow and shadow geometry must be correct before calling this
|
|
|
|
* function.
|
|
|
|
*/
|
|
|
|
static XserverRegion
|
|
|
|
win_extents(session_t *ps, win *w) {
|
|
|
|
XRectangle r;
|
|
|
|
|
|
|
|
r.x = w->a.x;
|
|
|
|
r.y = w->a.y;
|
|
|
|
r.width = w->widthb;
|
|
|
|
r.height = w->heightb;
|
|
|
|
|
|
|
|
if (w->shadow) {
|
|
|
|
XRectangle sr;
|
|
|
|
|
|
|
|
sr.x = w->a.x + w->shadow_dx;
|
|
|
|
sr.y = w->a.y + w->shadow_dy;
|
|
|
|
sr.width = w->shadow_width;
|
|
|
|
sr.height = w->shadow_height;
|
|
|
|
|
|
|
|
if (sr.x < r.x) {
|
|
|
|
r.width = (r.x + r.width) - sr.x;
|
|
|
|
r.x = sr.x;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (sr.y < r.y) {
|
|
|
|
r.height = (r.y + r.height) - sr.y;
|
|
|
|
r.y = sr.y;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (sr.x + sr.width > r.x + r.width) {
|
|
|
|
r.width = sr.x + sr.width - r.x;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (sr.y + sr.height > r.y + r.height) {
|
|
|
|
r.height = sr.y + sr.height - r.y;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return XFixesCreateRegion(ps->dpy, &r, 1);
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Retrieve the bounding shape of a window.
|
|
|
|
*/
|
|
|
|
static XserverRegion
|
|
|
|
border_size(session_t *ps, win *w, bool use_offset) {
|
|
|
|
// Start with the window rectangular region
|
|
|
|
XserverRegion fin = win_get_region(ps, w, use_offset);
|
|
|
|
|
|
|
|
// Only request for a bounding region if the window is shaped
|
|
|
|
if (w->bounding_shaped) {
|
|
|
|
/*
|
|
|
|
* if window doesn't exist anymore, this will generate an error
|
|
|
|
* as well as not generate a region. Perhaps a better XFixes
|
|
|
|
* architecture would be to have a request that copies instead
|
|
|
|
* of creates, that way you'd just end up with an empty region
|
|
|
|
* instead of an invalid XID.
|
|
|
|
*/
|
|
|
|
|
|
|
|
XserverRegion border = XFixesCreateRegionFromWindow(
|
|
|
|
ps->dpy, w->id, WindowRegionBounding);
|
|
|
|
|
|
|
|
if (!border)
|
|
|
|
return fin;
|
|
|
|
|
|
|
|
if (use_offset) {
|
|
|
|
// Translate the region to the correct place
|
|
|
|
XFixesTranslateRegion(ps->dpy, border,
|
|
|
|
w->a.x + w->a.border_width,
|
|
|
|
w->a.y + w->a.border_width);
|
|
|
|
}
|
|
|
|
|
|
|
|
// Intersect the bounding region we got with the window rectangle, to
|
|
|
|
// make sure the bounding region is not bigger than the window
|
|
|
|
// rectangle
|
|
|
|
XFixesIntersectRegion(ps->dpy, fin, fin, border);
|
|
|
|
XFixesDestroyRegion(ps->dpy, border);
|
|
|
|
}
|
|
|
|
|
|
|
|
return fin;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Look for the client window of a particular window.
|
|
|
|
*/
|
|
|
|
static Window
|
|
|
|
find_client_win(session_t *ps, Window w) {
|
|
|
|
if (wid_has_prop(ps, w, ps->atom_client)) {
|
|
|
|
return w;
|
|
|
|
}
|
|
|
|
|
|
|
|
Window *children;
|
|
|
|
unsigned int nchildren;
|
|
|
|
unsigned int i;
|
|
|
|
Window ret = 0;
|
|
|
|
|
|
|
|
if (!wid_get_children(ps, w, &children, &nchildren)) {
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
for (i = 0; i < nchildren; ++i) {
|
|
|
|
if ((ret = find_client_win(ps, children[i])))
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
XFree(children);
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Retrieve frame extents from a window.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
get_frame_extents(session_t *ps, win *w, Window client) {
|
|
|
|
w->left_width = 0;
|
|
|
|
w->right_width = 0;
|
|
|
|
w->top_width = 0;
|
|
|
|
w->bottom_width = 0;
|
|
|
|
|
|
|
|
winprop_t prop = wid_get_prop(ps, client, ps->atom_frame_extents,
|
|
|
|
4L, XA_CARDINAL, 32);
|
|
|
|
|
|
|
|
if (4 == prop.nitems) {
|
|
|
|
const long * const extents = prop.data.p32;
|
|
|
|
w->left_width = extents[0];
|
|
|
|
w->right_width = extents[1];
|
|
|
|
w->top_width = extents[2];
|
|
|
|
w->bottom_width = extents[3];
|
|
|
|
|
|
|
|
if (ps->o.frame_opacity)
|
|
|
|
update_reg_ignore_expire(ps, w);
|
|
|
|
}
|
|
|
|
|
|
|
|
#ifdef DEBUG_FRAME
|
|
|
|
printf_dbgf("(%#010lx): %d, %d, %d, %d\n", w->id,
|
|
|
|
w->left_width, w->right_width, w->top_width, w->bottom_width);
|
|
|
|
#endif
|
|
|
|
|
|
|
|
free_winprop(&prop);
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Get alpha <code>Picture</code> for an opacity in <code>double</code>.
|
|
|
|
*/
|
|
|
|
static inline Picture
|
|
|
|
get_alpha_pict_d(session_t *ps, double o) {
|
|
|
|
assert((round(normalize_d(o) / ps->o.alpha_step)) <= round(1.0 / ps->o.alpha_step));
|
|
|
|
return ps->alpha_picts[(int) (round(normalize_d(o)
|
|
|
|
/ ps->o.alpha_step))];
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Get alpha <code>Picture</code> for an opacity in
|
|
|
|
* <code>opacity_t</code>.
|
|
|
|
*/
|
|
|
|
static inline Picture
|
|
|
|
get_alpha_pict_o(session_t *ps, opacity_t o) {
|
|
|
|
return get_alpha_pict_d(ps, (double) o / OPAQUE);
|
|
|
|
}
|
|
|
|
|
|
|
|
static win *
|
|
|
|
paint_preprocess(session_t *ps, win *list) {
|
|
|
|
// Initialize unredir_possible
|
|
|
|
ps->unredir_possible = false;
|
|
|
|
|
|
|
|
win *w;
|
|
|
|
win *t = NULL, *next = NULL;
|
|
|
|
// Trace whether it's the highest window to paint
|
|
|
|
bool is_highest = true;
|
|
|
|
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
// Fading step calculation
|
|
|
|
time_ms_t steps = 0L;
|
|
|
|
if (ps->fade_time) {
|
|
|
|
steps = ((get_time_ms() - ps->fade_time) + FADE_DELTA_TOLERANCE * ps->o.fade_delta) / ps->o.fade_delta;
|
|
|
|
}
|
|
|
|
// Reset fade_time if unset, or there appears to be a time disorder
|
|
|
|
if (!ps->fade_time || steps < 0L) {
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
ps->fade_time = get_time_ms();
|
|
|
|
steps = 0L;
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
}
|
|
|
|
ps->fade_time += steps * ps->o.fade_delta;
|
|
|
|
|
|
|
|
XserverRegion last_reg_ignore = None;
|
|
|
|
|
|
|
|
for (w = list; w; w = next) {
|
|
|
|
bool to_paint = true;
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
const winmode_t mode_old = w->mode;
|
|
|
|
|
|
|
|
// In case calling the fade callback function destroys this window
|
|
|
|
next = w->next;
|
|
|
|
opacity_t opacity_old = w->opacity;
|
|
|
|
|
|
|
|
// Destroy reg_ignore on all windows if they should expire
|
|
|
|
if (ps->reg_ignore_expire)
|
|
|
|
free_region(ps, &w->reg_ignore);
|
|
|
|
|
|
|
|
// Update window opacity target and dim state if asked
|
|
|
|
if (WFLAG_OPCT_CHANGE & w->flags) {
|
|
|
|
calc_opacity(ps, w);
|
|
|
|
calc_dim(ps, w);
|
|
|
|
}
|
|
|
|
|
|
|
|
// Run fading
|
|
|
|
run_fade(ps, w, steps);
|
|
|
|
|
|
|
|
// Give up if it's not damaged or invisible, or it's unmapped and its
|
|
|
|
// picture is gone (for example due to a ConfigureNotify)
|
|
|
|
if (!w->damaged
|
|
|
|
|| w->a.x + w->a.width < 1 || w->a.y + w->a.height < 1
|
|
|
|
|| w->a.x >= ps->root_width || w->a.y >= ps->root_height
|
|
|
|
|| ((IsUnmapped == w->a.map_state || w->destroyed)
|
|
|
|
&& !w->picture)) {
|
|
|
|
to_paint = false;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (to_paint) {
|
|
|
|
// If opacity changes
|
|
|
|
if (w->opacity != opacity_old) {
|
|
|
|
determine_mode(ps, w);
|
|
|
|
add_damage_win(ps, w);
|
|
|
|
}
|
|
|
|
|
|
|
|
w->alpha_pict = get_alpha_pict_o(ps, w->opacity);
|
|
|
|
|
|
|
|
// End the game if we are using the 0 opacity alpha_pict
|
|
|
|
if (w->alpha_pict == ps->alpha_picts[0]) {
|
|
|
|
to_paint = false;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
if (to_paint) {
|
|
|
|
// Fetch bounding region
|
|
|
|
if (!w->border_size) {
|
|
|
|
w->border_size = border_size(ps, w, true);
|
|
|
|
}
|
|
|
|
|
|
|
|
// Fetch window extents
|
|
|
|
if (!w->extents) {
|
|
|
|
w->extents = win_extents(ps, w);
|
|
|
|
// If w->extents does not exist, the previous add_damage_win()
|
|
|
|
// call when opacity changes has no effect, so redo it here.
|
|
|
|
if (w->opacity != opacity_old)
|
|
|
|
add_damage_win(ps, w);
|
|
|
|
}
|
|
|
|
|
|
|
|
// Calculate frame_opacity
|
|
|
|
{
|
|
|
|
double frame_opacity_old = w->frame_opacity;
|
|
|
|
|
|
|
|
if (ps->o.frame_opacity && 1.0 != ps->o.frame_opacity
|
|
|
|
&& win_has_frame(w))
|
|
|
|
w->frame_opacity = get_opacity_percent(w) *
|
|
|
|
ps->o.frame_opacity;
|
|
|
|
else
|
|
|
|
w->frame_opacity = 0.0;
|
|
|
|
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
if (w->to_paint && WMODE_SOLID == mode_old
|
|
|
|
&& (0.0 == frame_opacity_old) != (0.0 == w->frame_opacity))
|
|
|
|
ps->reg_ignore_expire = true;
|
|
|
|
}
|
|
|
|
|
|
|
|
w->frame_alpha_pict = get_alpha_pict_d(ps, w->frame_opacity);
|
|
|
|
|
|
|
|
// Calculate shadow opacity
|
|
|
|
if (w->frame_opacity)
|
|
|
|
w->shadow_opacity = ps->o.shadow_opacity * w->frame_opacity;
|
|
|
|
else
|
|
|
|
w->shadow_opacity = ps->o.shadow_opacity * get_opacity_percent(w);
|
|
|
|
|
|
|
|
// Rebuild shadow_pict if necessary
|
|
|
|
if (w->flags & WFLAG_SIZE_CHANGE)
|
|
|
|
free_picture(ps, &w->shadow_pict);
|
|
|
|
|
|
|
|
if (w->shadow && !w->shadow_pict) {
|
|
|
|
w->shadow_pict = shadow_picture(ps, 1, w->widthb, w->heightb);
|
|
|
|
}
|
|
|
|
|
|
|
|
w->shadow_alpha_pict = get_alpha_pict_d(ps, w->shadow_opacity);
|
|
|
|
|
|
|
|
// Dimming preprocessing
|
|
|
|
if (w->dim) {
|
|
|
|
double dim_opacity = ps->o.inactive_dim;
|
|
|
|
if (!ps->o.inactive_dim_fixed)
|
|
|
|
dim_opacity *= get_opacity_percent(w);
|
|
|
|
|
|
|
|
w->dim_alpha_pict = get_alpha_pict_d(ps, dim_opacity);
|
|
|
|
}
|
|
|
|
|
|
|
|
}
|
|
|
|
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
if ((to_paint && WMODE_SOLID == w->mode)
|
|
|
|
!= (w->to_paint && WMODE_SOLID == mode_old))
|
|
|
|
ps->reg_ignore_expire = true;
|
|
|
|
|
|
|
|
// Add window to damaged area if its painting status changes
|
|
|
|
if (to_paint != w->to_paint)
|
|
|
|
add_damage_win(ps, w);
|
|
|
|
|
|
|
|
if (to_paint) {
|
|
|
|
// Generate ignore region for painting to reduce GPU load
|
|
|
|
if (ps->reg_ignore_expire || !w->to_paint) {
|
|
|
|
free_region(ps, &w->reg_ignore);
|
|
|
|
|
|
|
|
// If the window is solid, we add the window region to the
|
|
|
|
// ignored region
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
if (WMODE_SOLID == w->mode) {
|
|
|
|
if (!w->frame_opacity) {
|
|
|
|
if (w->border_size)
|
|
|
|
w->reg_ignore = copy_region(ps, w->border_size);
|
|
|
|
else
|
|
|
|
w->reg_ignore = win_get_region(ps, w, true);
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
w->reg_ignore = win_get_region_noframe(ps, w, true);
|
|
|
|
if (w->border_size)
|
|
|
|
XFixesIntersectRegion(ps->dpy, w->reg_ignore, w->reg_ignore,
|
|
|
|
w->border_size);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (last_reg_ignore)
|
|
|
|
XFixesUnionRegion(ps->dpy, w->reg_ignore, w->reg_ignore,
|
|
|
|
last_reg_ignore);
|
|
|
|
}
|
|
|
|
// Otherwise we copy the last region over
|
|
|
|
else if (last_reg_ignore)
|
|
|
|
w->reg_ignore = copy_region(ps, last_reg_ignore);
|
|
|
|
else
|
|
|
|
w->reg_ignore = None;
|
|
|
|
}
|
|
|
|
|
|
|
|
last_reg_ignore = w->reg_ignore;
|
|
|
|
|
|
|
|
if (is_highest && to_paint) {
|
|
|
|
is_highest = false;
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
if (WMODE_SOLID == w->mode
|
|
|
|
&& (!w->frame_opacity || !win_has_frame(w))
|
|
|
|
&& win_is_fullscreen(ps, w))
|
|
|
|
ps->unredir_possible = true;
|
|
|
|
}
|
|
|
|
|
|
|
|
// Reset flags
|
|
|
|
w->flags = 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
// Avoid setting w->to_paint if w is to be freed
|
|
|
|
bool destroyed = (w->opacity_tgt == w->opacity && w->destroyed);
|
|
|
|
|
|
|
|
if (to_paint) {
|
|
|
|
w->prev_trans = t;
|
|
|
|
t = w;
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
check_fade_fin(ps, w);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (!destroyed)
|
|
|
|
w->to_paint = to_paint;
|
|
|
|
}
|
|
|
|
|
|
|
|
// If possible, unredirect all windows and stop painting
|
|
|
|
if (ps->o.unredir_if_possible && ps->unredir_possible) {
|
|
|
|
redir_stop(ps);
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
redir_start(ps);
|
|
|
|
}
|
|
|
|
|
|
|
|
// Fetch pictures only if windows are redirected
|
|
|
|
if (ps->redirected) {
|
|
|
|
for (w = t; w; w = w->prev_trans) {
|
|
|
|
// Fetch the picture and pixmap if needed
|
|
|
|
if (!w->picture) {
|
|
|
|
XRenderPictureAttributes pa;
|
|
|
|
XRenderPictFormat *format;
|
|
|
|
Drawable draw = w->id;
|
|
|
|
|
|
|
|
if (ps->has_name_pixmap && !w->pixmap) {
|
|
|
|
set_ignore_next(ps);
|
|
|
|
w->pixmap = XCompositeNameWindowPixmap(ps->dpy, w->id);
|
|
|
|
}
|
|
|
|
if (w->pixmap) draw = w->pixmap;
|
|
|
|
|
|
|
|
format = XRenderFindVisualFormat(ps->dpy, w->a.visual);
|
|
|
|
pa.subwindow_mode = IncludeInferiors;
|
|
|
|
w->picture = XRenderCreatePicture(
|
|
|
|
ps->dpy, draw, format, CPSubwindowMode, &pa);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return t;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Paint the shadow of a window.
|
|
|
|
*/
|
|
|
|
static inline void
|
|
|
|
win_paint_shadow(session_t *ps, win *w, Picture tgt_buffer) {
|
|
|
|
XRenderComposite(
|
|
|
|
ps->dpy, PictOpOver, w->shadow_pict, w->shadow_alpha_pict,
|
|
|
|
tgt_buffer, 0, 0, 0, 0,
|
|
|
|
w->a.x + w->shadow_dx, w->a.y + w->shadow_dy,
|
|
|
|
w->shadow_width, w->shadow_height);
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Blur the background of a window.
|
|
|
|
*/
|
|
|
|
static inline void
|
|
|
|
win_blur_background(session_t *ps, win *w, Picture tgt_buffer,
|
|
|
|
XserverRegion reg_paint) {
|
|
|
|
const static int convolution_blur_size = 3;
|
|
|
|
// Convolution filter parameter (box blur)
|
|
|
|
// gaussian or binomial filters are definitely superior, yet looks
|
|
|
|
// like they aren't supported as of xorg-server-1.13.0
|
|
|
|
XFixed convolution_blur[] = {
|
|
|
|
// Must convert to XFixed with XDoubleToFixed()
|
|
|
|
// Matrix size
|
|
|
|
XDoubleToFixed(convolution_blur_size),
|
|
|
|
XDoubleToFixed(convolution_blur_size),
|
|
|
|
// Matrix
|
|
|
|
XDoubleToFixed(1), XDoubleToFixed(1), XDoubleToFixed(1),
|
|
|
|
XDoubleToFixed(1), XDoubleToFixed(1), XDoubleToFixed(1),
|
|
|
|
XDoubleToFixed(1), XDoubleToFixed(1), XDoubleToFixed(1),
|
|
|
|
};
|
|
|
|
|
|
|
|
Pixmap tmp_pixmap = None;
|
|
|
|
Picture tmp_picture = None;
|
|
|
|
|
|
|
|
int x = w->a.x;
|
|
|
|
int y = w->a.y;
|
|
|
|
int wid = w->widthb;
|
|
|
|
int hei = w->heightb;
|
|
|
|
|
|
|
|
// Directly copying from tgt_buffer does not work, so we create a
|
|
|
|
// Picture in the middle.
|
|
|
|
tmp_pixmap = XCreatePixmap(ps->dpy, ps->root, wid, hei, ps->depth);
|
|
|
|
if (!tmp_pixmap)
|
|
|
|
goto win_blur_background_err;
|
|
|
|
|
|
|
|
tmp_picture = XRenderCreatePicture(ps->dpy, tmp_pixmap,
|
|
|
|
XRenderFindVisualFormat(ps->dpy, ps->vis), 0, 0);
|
|
|
|
if (!tmp_picture)
|
|
|
|
goto win_blur_background_err;
|
|
|
|
|
|
|
|
// Adjust blur strength according to window opacity, to make it appear
|
|
|
|
// better during fading
|
|
|
|
if (!ps->o.blur_background_fixed) {
|
|
|
|
double pct = 1.0 - get_opacity_percent(w) * (1.0 - 1.0 / 9.0);
|
|
|
|
convolution_blur[2 + convolution_blur_size + ((convolution_blur_size - 1) / 2)] = XDoubleToFixed(pct * 8.0 / (1.1 - pct));
|
|
|
|
}
|
|
|
|
|
|
|
|
// Minimize the region we try to blur, if the window itself is not
|
|
|
|
// opaque, only the frame is.
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
if (WMODE_SOLID == w->mode && w->frame_opacity) {
|
|
|
|
XserverRegion reg_all = border_size(ps, w, false);
|
|
|
|
XserverRegion reg_noframe = win_get_region_noframe(ps, w, false);
|
|
|
|
XFixesSubtractRegion(ps->dpy, reg_noframe, reg_all, reg_noframe);
|
|
|
|
XFixesSetPictureClipRegion(ps->dpy, tmp_picture, reg_noframe, 0, 0);
|
|
|
|
free_region(ps, ®_all);
|
|
|
|
free_region(ps, ®_noframe);
|
|
|
|
}
|
|
|
|
|
|
|
|
// Copy the content to tmp_picture, then copy back. The filter must
|
|
|
|
// be applied on tgt_buffer, to get the nearby pixels outside the
|
|
|
|
// window.
|
|
|
|
XRenderSetPictureFilter(ps->dpy, tgt_buffer, XRFILTER_CONVOLUTION, (XFixed *) convolution_blur, sizeof(convolution_blur) / sizeof(XFixed));
|
|
|
|
XRenderComposite(ps->dpy, PictOpSrc, tgt_buffer, None, tmp_picture, x, y, 0, 0, 0, 0, wid, hei);
|
|
|
|
xrfilter_reset(ps, tgt_buffer);
|
|
|
|
XRenderComposite(ps->dpy, PictOpSrc, tmp_picture, None, tgt_buffer, 0, 0, 0, 0, x, y, wid, hei);
|
|
|
|
|
|
|
|
win_blur_background_err:
|
|
|
|
free_pixmap(ps, &tmp_pixmap);
|
|
|
|
free_picture(ps, &tmp_picture);
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Paint a window itself and dim it if asked.
|
|
|
|
*/
|
|
|
|
static inline void
|
|
|
|
win_paint_win(session_t *ps, win *w, Picture tgt_buffer) {
|
|
|
|
int x = w->a.x;
|
|
|
|
int y = w->a.y;
|
|
|
|
int wid = w->widthb;
|
|
|
|
int hei = w->heightb;
|
|
|
|
|
|
|
|
Picture alpha_mask = (OPAQUE == w->opacity ? None: w->alpha_pict);
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
int op = (w->mode == WMODE_SOLID ? PictOpSrc: PictOpOver);
|
|
|
|
|
|
|
|
if (!w->frame_opacity) {
|
|
|
|
XRenderComposite(ps->dpy, op, w->picture, alpha_mask,
|
|
|
|
tgt_buffer, 0, 0, 0, 0, x, y, wid, hei);
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
// Painting parameters
|
|
|
|
const int t = w->a.border_width + w->top_width;
|
|
|
|
const int l = w->a.border_width + w->left_width;
|
|
|
|
const int b = w->a.border_width + w->bottom_width;
|
|
|
|
const int r = w->a.border_width + w->right_width;
|
|
|
|
|
|
|
|
#define COMP_BDR(cx, cy, cwid, chei) \
|
|
|
|
XRenderComposite(ps->dpy, PictOpOver, w->picture, w->frame_alpha_pict, \
|
|
|
|
tgt_buffer, (cx), (cy), 0, 0, x + (cx), y + (cy), (cwid), (chei))
|
|
|
|
|
|
|
|
// The following complicated logic is required because some broken
|
|
|
|
// window managers (I'm talking about you, Openbox!) that makes
|
|
|
|
// top_width + bottom_width > height in some cases.
|
|
|
|
|
|
|
|
// top
|
|
|
|
int phei = min_i(t, hei);
|
|
|
|
if (phei > 0)
|
|
|
|
COMP_BDR(0, 0, wid, phei);
|
|
|
|
|
|
|
|
if (hei > t) {
|
|
|
|
phei = min_i(hei - t, b);
|
|
|
|
|
|
|
|
// bottom
|
|
|
|
if (phei > 0)
|
|
|
|
COMP_BDR(0, hei - phei, wid, phei);
|
|
|
|
|
|
|
|
phei = hei - t - phei;
|
|
|
|
if (phei > 0) {
|
|
|
|
int pwid = min_i(l, wid);
|
|
|
|
// left
|
|
|
|
if (pwid > 0)
|
|
|
|
COMP_BDR(0, t, pwid, phei);
|
|
|
|
|
|
|
|
if (wid > l) {
|
|
|
|
pwid = min_i(wid - l, r);
|
|
|
|
|
|
|
|
// right
|
|
|
|
if (pwid > 0)
|
|
|
|
COMP_BDR(wid - pwid, t, pwid, phei);
|
|
|
|
|
|
|
|
pwid = wid - l - pwid;
|
|
|
|
if (pwid > 0) {
|
|
|
|
// body
|
|
|
|
XRenderComposite(ps->dpy, op, w->picture, alpha_mask,
|
|
|
|
tgt_buffer, l, t, 0, 0, x + l, y + t, pwid, phei);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
#undef COMP_BDR
|
|
|
|
|
|
|
|
// Dimming the window if needed
|
|
|
|
if (w->dim && w->dim_alpha_pict != ps->alpha_picts[0]) {
|
|
|
|
XRenderComposite(ps->dpy, PictOpOver, ps->black_picture,
|
|
|
|
w->dim_alpha_pict, tgt_buffer, 0, 0, 0, 0, x, y, wid, hei);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Rebuild cached <code>screen_reg</code>.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
rebuild_screen_reg(session_t *ps) {
|
|
|
|
if (ps->screen_reg)
|
|
|
|
XFixesDestroyRegion(ps->dpy, ps->screen_reg);
|
|
|
|
ps->screen_reg = get_screen_region(ps);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
paint_all(session_t *ps, XserverRegion region, win *t) {
|
|
|
|
#ifdef DEBUG_REPAINT
|
|
|
|
static struct timespec last_paint = { 0 };
|
|
|
|
#endif
|
|
|
|
|
|
|
|
win *w;
|
|
|
|
XserverRegion reg_paint = None, reg_tmp = None, reg_tmp2 = None;
|
|
|
|
|
|
|
|
if (!region) {
|
|
|
|
region = get_screen_region(ps);
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
// Remove the damaged area out of screen
|
|
|
|
XFixesIntersectRegion(ps->dpy, region, region, ps->screen_reg);
|
|
|
|
}
|
|
|
|
|
|
|
|
#ifdef MONITOR_REPAINT
|
|
|
|
// Note: MONITOR_REPAINT cannot work with DBE right now.
|
|
|
|
ps->tgt_buffer = ps->tgt_picture;
|
|
|
|
#else
|
|
|
|
if (!ps->tgt_buffer) {
|
|
|
|
// DBE painting mode: Directly paint to a Picture of the back buffer
|
|
|
|
if (ps->o.dbe) {
|
|
|
|
ps->tgt_buffer = XRenderCreatePicture(ps->dpy, ps->root_dbe,
|
|
|
|
XRenderFindVisualFormat(ps->dpy, ps->vis),
|
|
|
|
0, 0);
|
|
|
|
}
|
|
|
|
// No-DBE painting mode: Paint to an intermediate Picture then paint
|
|
|
|
// the Picture to root window
|
|
|
|
else {
|
|
|
|
Pixmap root_pixmap = XCreatePixmap(
|
|
|
|
ps->dpy, ps->root, ps->root_width, ps->root_height,
|
|
|
|
ps->depth);
|
|
|
|
|
|
|
|
ps->tgt_buffer = XRenderCreatePicture(ps->dpy, root_pixmap,
|
|
|
|
XRenderFindVisualFormat(ps->dpy, ps->vis),
|
|
|
|
0, 0);
|
|
|
|
|
|
|
|
XFreePixmap(ps->dpy, root_pixmap);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
|
|
|
XFixesSetPictureClipRegion(ps->dpy, ps->tgt_picture, 0, 0, region);
|
|
|
|
|
|
|
|
#ifdef MONITOR_REPAINT
|
|
|
|
XRenderComposite(
|
|
|
|
ps->dpy, PictOpSrc, ps->black_picture, None,
|
|
|
|
ps->tgt_picture, 0, 0, 0, 0, 0, 0,
|
|
|
|
ps->root_width, ps->root_height);
|
|
|
|
#endif
|
|
|
|
|
|
|
|
if (t && t->reg_ignore) {
|
|
|
|
// Calculate the region upon which the root window is to be painted
|
|
|
|
// based on the ignore region of the lowest window, if available
|
|
|
|
reg_paint = reg_tmp = XFixesCreateRegion(ps->dpy, NULL, 0);
|
|
|
|
XFixesSubtractRegion(ps->dpy, reg_paint, region, t->reg_ignore);
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
reg_paint = region;
|
|
|
|
}
|
|
|
|
|
|
|
|
XFixesSetPictureClipRegion(ps->dpy, ps->tgt_buffer, 0, 0, reg_paint);
|
|
|
|
|
|
|
|
paint_root(ps, ps->tgt_buffer);
|
|
|
|
|
|
|
|
// Create temporary regions for use during painting
|
|
|
|
if (!reg_tmp)
|
|
|
|
reg_tmp = XFixesCreateRegion(ps->dpy, NULL, 0);
|
|
|
|
reg_tmp2 = XFixesCreateRegion(ps->dpy, NULL, 0);
|
|
|
|
|
|
|
|
for (w = t; w; w = w->prev_trans) {
|
|
|
|
// Painting shadow
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
if (w->shadow) {
|
|
|
|
// Shadow is to be painted based on the ignore region of current
|
|
|
|
// window
|
|
|
|
if (w->reg_ignore) {
|
|
|
|
if (w == t) {
|
|
|
|
// If it's the first cycle and reg_tmp2 is not ready, calculate
|
|
|
|
// the paint region here
|
|
|
|
reg_paint = reg_tmp;
|
|
|
|
XFixesSubtractRegion(ps->dpy, reg_paint, region, w->reg_ignore);
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
// Otherwise, used the cached region during last cycle
|
|
|
|
reg_paint = reg_tmp2;
|
|
|
|
}
|
|
|
|
XFixesIntersectRegion(ps->dpy, reg_paint, reg_paint, w->extents);
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
reg_paint = reg_tmp;
|
|
|
|
XFixesIntersectRegion(ps->dpy, reg_paint, region, w->extents);
|
|
|
|
}
|
|
|
|
// Clear the shadow here instead of in make_shadow() for saving GPU
|
|
|
|
// power and handling shaped windows
|
|
|
|
if (ps->o.clear_shadow && w->border_size)
|
|
|
|
XFixesSubtractRegion(ps->dpy, reg_paint, reg_paint, w->border_size);
|
|
|
|
|
|
|
|
// Detect if the region is empty before painting
|
|
|
|
if (region == reg_paint || !is_region_empty(ps, reg_paint)) {
|
|
|
|
XFixesSetPictureClipRegion(ps->dpy, ps->tgt_buffer, 0, 0,
|
|
|
|
reg_paint);
|
|
|
|
|
|
|
|
win_paint_shadow(ps, w, ps->tgt_buffer);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
// Calculate the region based on the reg_ignore of the next (higher)
|
|
|
|
// window and the bounding region
|
|
|
|
reg_paint = reg_tmp;
|
|
|
|
if (w->prev_trans && w->prev_trans->reg_ignore) {
|
|
|
|
XFixesSubtractRegion(ps->dpy, reg_paint, region,
|
|
|
|
w->prev_trans->reg_ignore);
|
|
|
|
// Copy the subtracted region to be used for shadow painting in next
|
|
|
|
// cycle
|
|
|
|
XFixesCopyRegion(ps->dpy, reg_tmp2, reg_paint);
|
|
|
|
|
|
|
|
if (w->border_size)
|
|
|
|
XFixesIntersectRegion(ps->dpy, reg_paint, reg_paint, w->border_size);
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
if (w->border_size)
|
|
|
|
XFixesIntersectRegion(ps->dpy, reg_paint, region, w->border_size);
|
|
|
|
else
|
|
|
|
reg_paint = region;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (!is_region_empty(ps, reg_paint)) {
|
|
|
|
XFixesSetPictureClipRegion(ps->dpy, ps->tgt_buffer, 0, 0, reg_paint);
|
|
|
|
// Blur window background
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
if ((ps->o.blur_background && WMODE_SOLID != w->mode)
|
|
|
|
|| (ps->o.blur_background_frame && w->frame_opacity)) {
|
|
|
|
win_blur_background(ps, w, ps->tgt_buffer, reg_paint);
|
|
|
|
}
|
|
|
|
|
|
|
|
// Painting the window
|
|
|
|
win_paint_win(ps, w, ps->tgt_buffer);
|
|
|
|
}
|
|
|
|
|
|
|
|
check_fade_fin(ps, w);
|
|
|
|
}
|
|
|
|
|
|
|
|
// Free up all temporary regions
|
|
|
|
XFixesDestroyRegion(ps->dpy, region);
|
|
|
|
XFixesDestroyRegion(ps->dpy, reg_tmp);
|
|
|
|
XFixesDestroyRegion(ps->dpy, reg_tmp2);
|
|
|
|
|
|
|
|
// Do this as early as possible
|
|
|
|
if (!ps->o.dbe)
|
|
|
|
XFixesSetPictureClipRegion(ps->dpy, ps->tgt_buffer, 0, 0, None);
|
|
|
|
|
|
|
|
if (VSYNC_NONE != ps->o.vsync) {
|
|
|
|
// Make sure all previous requests are processed to achieve best
|
|
|
|
// effect
|
|
|
|
XSync(ps->dpy, False);
|
|
|
|
}
|
|
|
|
|
|
|
|
// Wait for VBlank. We could do it aggressively (send the painting
|
|
|
|
// request and XFlush() on VBlank) or conservatively (send the request
|
|
|
|
// only on VBlank).
|
|
|
|
if (!ps->o.vsync_aggressive)
|
|
|
|
vsync_wait(ps);
|
|
|
|
|
|
|
|
// DBE painting mode, only need to swap the buffer
|
|
|
|
if (ps->o.dbe) {
|
|
|
|
XdbeSwapInfo swap_info = {
|
|
|
|
.swap_window = get_tgt_window(ps),
|
|
|
|
// Is it safe to use XdbeUndefined?
|
|
|
|
.swap_action = XdbeCopied
|
|
|
|
};
|
|
|
|
XdbeSwapBuffers(ps->dpy, &swap_info, 1);
|
|
|
|
}
|
|
|
|
// No-DBE painting mode
|
|
|
|
else if (ps->tgt_buffer != ps->tgt_picture) {
|
|
|
|
XRenderComposite(
|
|
|
|
ps->dpy, PictOpSrc, ps->tgt_buffer, None,
|
|
|
|
ps->tgt_picture, 0, 0, 0, 0,
|
|
|
|
0, 0, ps->root_width, ps->root_height);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (ps->o.vsync_aggressive)
|
|
|
|
vsync_wait(ps);
|
|
|
|
|
|
|
|
XFlush(ps->dpy);
|
|
|
|
|
|
|
|
#ifdef DEBUG_REPAINT
|
|
|
|
print_timestamp(ps);
|
|
|
|
struct timespec now = get_time_timespec();
|
|
|
|
struct timespec diff = { 0 };
|
|
|
|
timespec_subtract(&diff, &now, &last_paint);
|
|
|
|
printf("[ %5ld:%09ld ] ", diff.tv_sec, diff.tv_nsec);
|
|
|
|
last_paint = now;
|
|
|
|
printf("paint:");
|
|
|
|
for (w = t; w; w = w->prev_trans)
|
|
|
|
printf(" %#010lx", w->id);
|
|
|
|
putchar('\n');
|
|
|
|
fflush(stdout);
|
|
|
|
#endif
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
add_damage(session_t *ps, XserverRegion damage) {
|
|
|
|
if (ps->all_damage) {
|
|
|
|
XFixesUnionRegion(ps->dpy, ps->all_damage, ps->all_damage, damage);
|
|
|
|
XFixesDestroyRegion(ps->dpy, damage);
|
|
|
|
} else {
|
|
|
|
ps->all_damage = damage;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
repair_win(session_t *ps, win *w) {
|
|
|
|
XserverRegion parts;
|
|
|
|
|
|
|
|
if (!w->damaged) {
|
|
|
|
parts = win_extents(ps, w);
|
|
|
|
set_ignore_next(ps);
|
|
|
|
XDamageSubtract(ps->dpy, w->damage, None, None);
|
|
|
|
} else {
|
|
|
|
parts = XFixesCreateRegion(ps->dpy, 0, 0);
|
|
|
|
set_ignore_next(ps);
|
|
|
|
XDamageSubtract(ps->dpy, w->damage, None, parts);
|
|
|
|
XFixesTranslateRegion(ps->dpy, parts,
|
|
|
|
w->a.x + w->a.border_width,
|
|
|
|
w->a.y + w->a.border_width);
|
|
|
|
}
|
|
|
|
|
|
|
|
// Remove the part in the damage area that could be ignored
|
|
|
|
if (!ps->reg_ignore_expire && w->prev_trans && w->prev_trans->reg_ignore)
|
|
|
|
XFixesSubtractRegion(ps->dpy, parts, parts, w->prev_trans->reg_ignore);
|
|
|
|
|
|
|
|
add_damage(ps, parts);
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
w->damaged = true;
|
|
|
|
}
|
|
|
|
|
|
|
|
static wintype_t
|
|
|
|
wid_get_prop_wintype(session_t *ps, Window wid) {
|
|
|
|
int i;
|
|
|
|
|
|
|
|
set_ignore_next(ps);
|
|
|
|
winprop_t prop = wid_get_prop(ps, wid, ps->atom_win_type, 32L, XA_ATOM, 32);
|
|
|
|
|
|
|
|
for (i = 0; i < prop.nitems; ++i) {
|
|
|
|
for (wintype_t j = 1; j < NUM_WINTYPES; ++j) {
|
|
|
|
if (ps->atoms_wintypes[j] == (Atom) prop.data.p32[i]) {
|
|
|
|
free_winprop(&prop);
|
|
|
|
return j;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
free_winprop(&prop);
|
|
|
|
|
|
|
|
return WINTYPE_UNKNOWN;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
map_win(session_t *ps, Window id) {
|
|
|
|
win *w = find_win(ps, id);
|
|
|
|
|
|
|
|
// Don't care about window mapping if it's an InputOnly window
|
|
|
|
// Try avoiding mapping a window twice
|
|
|
|
if (!w || InputOnly == w->a.class
|
|
|
|
|| IsViewable == w->a.map_state)
|
|
|
|
return;
|
|
|
|
|
|
|
|
assert(!w->focused_real);
|
|
|
|
|
|
|
|
w->a.map_state = IsViewable;
|
|
|
|
|
|
|
|
// Set focused to false
|
|
|
|
bool focused_real = false;
|
|
|
|
if (ps->o.track_focus && ps->o.use_ewmh_active_win
|
|
|
|
&& w == ps->active_win)
|
|
|
|
focused_real = true;
|
|
|
|
win_set_focused(ps, w, focused_real);
|
|
|
|
|
|
|
|
// Call XSelectInput() before reading properties so that no property
|
|
|
|
// changes are lost
|
|
|
|
XSelectInput(ps->dpy, id, determine_evmask(ps, id, WIN_EVMODE_FRAME));
|
|
|
|
|
|
|
|
// Notify compton when the shape of a window changes
|
|
|
|
if (ps->shape_exists) {
|
|
|
|
XShapeSelectInput(ps->dpy, id, ShapeNotifyMask);
|
|
|
|
}
|
|
|
|
|
|
|
|
// Make sure the XSelectInput() requests are sent
|
|
|
|
XSync(ps->dpy, False);
|
|
|
|
|
|
|
|
// Detect client window here instead of in add_win() as the client
|
|
|
|
// window should have been prepared at this point
|
|
|
|
if (!w->client_win) {
|
|
|
|
win_recheck_client(ps, w);
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
// Re-mark client window here
|
|
|
|
win_mark_client(ps, w, w->client_win);
|
|
|
|
}
|
|
|
|
|
|
|
|
assert(w->client_win);
|
|
|
|
|
|
|
|
#ifdef DEBUG_WINTYPE
|
|
|
|
printf_dbgf("(%#010lx): type %s\n", w->id, WINTYPES[w->window_type]);
|
|
|
|
#endif
|
|
|
|
|
|
|
|
// Detect if the window is shaped or has rounded corners
|
|
|
|
win_update_shape_raw(ps, w);
|
|
|
|
|
|
|
|
// Occasionally compton does not seem able to get a FocusIn event from
|
|
|
|
// a window just mapped. I suspect it's a timing issue again when the
|
|
|
|
// XSelectInput() is called too late. We have to recheck the focused
|
|
|
|
// window here. It makes no sense if we are using EWMH
|
|
|
|
// _NET_ACTIVE_WINDOW.
|
|
|
|
if (ps->o.track_focus && !ps->o.use_ewmh_active_win) {
|
|
|
|
recheck_focus(ps);
|
|
|
|
}
|
|
|
|
|
|
|
|
// Update window focus state
|
|
|
|
win_update_focused(ps, w);
|
|
|
|
|
|
|
|
// Update opacity and dim state
|
|
|
|
win_update_opacity_prop(ps, w);
|
|
|
|
w->flags |= WFLAG_OPCT_CHANGE;
|
|
|
|
|
|
|
|
// Check for _COMPTON_SHADOW
|
|
|
|
if (ps->o.respect_prop_shadow)
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
win_update_prop_shadow_raw(ps, w);
|
|
|
|
|
|
|
|
// Many things above could affect shadow
|
|
|
|
determine_shadow(ps, w);
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
|
|
|
|
// Set fading state
|
|
|
|
if (ps->o.no_fading_openclose) {
|
|
|
|
set_fade_callback(ps, w, finish_map_win, true);
|
|
|
|
// Must be set after we execute the old fade callback, in case we
|
|
|
|
// receive two continuous MapNotify for the same window
|
|
|
|
w->fade = false;
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
set_fade_callback(ps, w, NULL, true);
|
|
|
|
determine_fade(ps, w);
|
|
|
|
}
|
|
|
|
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
w->damaged = true;
|
|
|
|
|
|
|
|
|
|
|
|
/* if any configure events happened while
|
|
|
|
the window was unmapped, then configure
|
|
|
|
the window to its correct place */
|
|
|
|
if (w->need_configure) {
|
|
|
|
configure_win(ps, &w->queue_configure);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
finish_map_win(session_t *ps, win *w) {
|
|
|
|
if (ps->o.no_fading_openclose)
|
|
|
|
determine_fade(ps, w);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
finish_unmap_win(session_t *ps, win *w) {
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
w->damaged = false;
|
|
|
|
|
|
|
|
update_reg_ignore_expire(ps, w);
|
|
|
|
|
|
|
|
if (w->extents != None) {
|
|
|
|
/* destroys region */
|
|
|
|
add_damage(ps, w->extents);
|
|
|
|
w->extents = None;
|
|
|
|
}
|
|
|
|
|
|
|
|
free_pixmap(ps, &w->pixmap);
|
|
|
|
free_picture(ps, &w->picture);
|
|
|
|
free_region(ps, &w->border_size);
|
|
|
|
free_picture(ps, &w->shadow_pict);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
unmap_callback(session_t *ps, win *w) {
|
|
|
|
finish_unmap_win(ps, w);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
unmap_win(session_t *ps, Window id) {
|
|
|
|
win *w = find_win(ps, id);
|
|
|
|
|
|
|
|
if (!w || IsUnmapped == w->a.map_state) return;
|
|
|
|
|
|
|
|
// Set focus out
|
|
|
|
win_set_focused(ps, w, false);
|
|
|
|
|
|
|
|
w->a.map_state = IsUnmapped;
|
|
|
|
|
|
|
|
// Fading out
|
|
|
|
w->flags |= WFLAG_OPCT_CHANGE;
|
|
|
|
set_fade_callback(ps, w, unmap_callback, false);
|
|
|
|
if (ps->o.no_fading_openclose)
|
|
|
|
w->fade = false;
|
|
|
|
|
|
|
|
// don't care about properties anymore
|
|
|
|
win_ev_stop(ps, w);
|
|
|
|
}
|
|
|
|
|
|
|
|
static opacity_t
|
|
|
|
wid_get_opacity_prop(session_t *ps, Window wid, opacity_t def) {
|
|
|
|
opacity_t val = def;
|
|
|
|
|
|
|
|
winprop_t prop = wid_get_prop(ps, wid, ps->atom_opacity, 1L,
|
|
|
|
XA_CARDINAL, 32);
|
|
|
|
|
|
|
|
if (prop.nitems)
|
|
|
|
val = *prop.data.p32;
|
|
|
|
|
|
|
|
free_winprop(&prop);
|
|
|
|
|
|
|
|
return val;
|
|
|
|
}
|
|
|
|
|
|
|
|
static double
|
|
|
|
get_opacity_percent(win *w) {
|
|
|
|
return ((double) w->opacity) / OPAQUE;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
determine_mode(session_t *ps, win *w) {
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
winmode_t mode = WMODE_SOLID;
|
|
|
|
XRenderPictFormat *format;
|
|
|
|
|
|
|
|
/* if trans prop == -1 fall back on previous tests */
|
|
|
|
|
|
|
|
if (w->a.class == InputOnly) {
|
|
|
|
format = 0;
|
|
|
|
} else {
|
|
|
|
format = XRenderFindVisualFormat(ps->dpy, w->a.visual);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (format && format->type == PictTypeDirect
|
|
|
|
&& format->direct.alphaMask) {
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
mode = WMODE_ARGB;
|
|
|
|
} else if (w->opacity != OPAQUE) {
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
mode = WMODE_TRANS;
|
|
|
|
} else {
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
mode = WMODE_SOLID;
|
|
|
|
}
|
|
|
|
|
|
|
|
w->mode = mode;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Calculate and set the opacity of a window.
|
|
|
|
*
|
|
|
|
* If window is inactive and inactive_opacity_override is set, the
|
|
|
|
* priority is: (Simulates the old behavior)
|
|
|
|
*
|
|
|
|
* inactive_opacity > _NET_WM_WINDOW_OPACITY (if not opaque)
|
|
|
|
* > window type default opacity
|
|
|
|
*
|
|
|
|
* Otherwise:
|
|
|
|
*
|
|
|
|
* _NET_WM_WINDOW_OPACITY (if not opaque)
|
|
|
|
* > window type default opacity (if not opaque)
|
|
|
|
* > inactive_opacity
|
|
|
|
*
|
|
|
|
* @param dpy X display to use
|
|
|
|
* @param w struct _win object representing the window
|
|
|
|
* @param refetch_prop whether _NET_WM_OPACITY of the window needs to be
|
|
|
|
* refetched
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
calc_opacity(session_t *ps, win *w) {
|
|
|
|
opacity_t opacity = OPAQUE;
|
|
|
|
|
|
|
|
if (w->destroyed || IsViewable != w->a.map_state)
|
|
|
|
opacity = 0;
|
|
|
|
else {
|
|
|
|
// Try obeying opacity property and window type opacity firstly
|
|
|
|
if (OPAQUE == (opacity = w->opacity_prop)
|
|
|
|
&& OPAQUE == (opacity = w->opacity_prop_client)) {
|
|
|
|
opacity = ps->o.wintype_opacity[w->window_type] * OPAQUE;
|
|
|
|
}
|
|
|
|
|
|
|
|
// Respect inactive_opacity in some cases
|
|
|
|
if (ps->o.inactive_opacity && false == w->focused
|
|
|
|
&& (OPAQUE == opacity || ps->o.inactive_opacity_override)) {
|
|
|
|
opacity = ps->o.inactive_opacity;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
w->opacity_tgt = opacity;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Determine whether a window is to be dimmed.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
calc_dim(session_t *ps, win *w) {
|
|
|
|
bool dim;
|
|
|
|
|
|
|
|
// Make sure we do nothing if the window is unmapped / destroyed
|
|
|
|
if (w->destroyed || IsViewable != w->a.map_state)
|
|
|
|
return;
|
|
|
|
|
|
|
|
if (ps->o.inactive_dim && !(w->focused)) {
|
|
|
|
dim = true;
|
|
|
|
} else {
|
|
|
|
dim = false;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (dim != w->dim) {
|
|
|
|
w->dim = dim;
|
|
|
|
add_damage_win(ps, w);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Determine if a window should fade on opacity change.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
determine_fade(session_t *ps, win *w) {
|
|
|
|
w->fade = ps->o.wintype_fade[w->window_type];
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Update window-shape.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
win_update_shape_raw(session_t *ps, win *w) {
|
|
|
|
if (ps->shape_exists) {
|
|
|
|
w->bounding_shaped = wid_bounding_shaped(ps, w->id);
|
|
|
|
if (w->bounding_shaped && ps->o.detect_rounded_corners)
|
|
|
|
win_rounded_corners(ps, w);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Update window-shape related information.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
win_update_shape(session_t *ps, win *w) {
|
|
|
|
if (ps->shape_exists) {
|
|
|
|
// bool bounding_shaped_old = w->bounding_shaped;
|
|
|
|
|
|
|
|
win_update_shape_raw(ps, w);
|
|
|
|
|
|
|
|
// Shadow state could be changed
|
|
|
|
determine_shadow(ps, w);
|
|
|
|
|
|
|
|
/*
|
|
|
|
// If clear_shadow state on the window possibly changed, destroy the old
|
|
|
|
// shadow_pict
|
|
|
|
if (ps->o.clear_shadow && w->bounding_shaped != bounding_shaped_old)
|
|
|
|
free_picture(ps, &w->shadow_pict);
|
|
|
|
*/
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Reread _COMPTON_SHADOW property from a window.
|
|
|
|
*
|
|
|
|
* The property must be set on the outermost window, usually the WM frame.
|
|
|
|
*/
|
|
|
|
static void
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
win_update_prop_shadow_raw(session_t *ps, win *w) {
|
|
|
|
winprop_t prop = wid_get_prop(ps, w->id, ps->atom_compton_shadow, 1,
|
|
|
|
XA_CARDINAL, 32);
|
|
|
|
|
|
|
|
if (!prop.nitems) {
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
w->prop_shadow = -1;
|
|
|
|
}
|
|
|
|
else {
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
w->prop_shadow = *prop.data.p32;
|
|
|
|
}
|
|
|
|
|
|
|
|
free_winprop(&prop);
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Reread _COMPTON_SHADOW property from a window and update related
|
|
|
|
* things.
|
|
|
|
*/
|
|
|
|
static void
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
win_update_prop_shadow(session_t *ps, win *w) {
|
|
|
|
long attr_shadow_old = w->prop_shadow;
|
|
|
|
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
win_update_prop_shadow_raw(ps, w);
|
|
|
|
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
if (w->prop_shadow != attr_shadow_old)
|
|
|
|
determine_shadow(ps, w);
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Determine if a window should have shadow, and update things depending
|
|
|
|
* on shadow state.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
determine_shadow(session_t *ps, win *w) {
|
|
|
|
bool shadow_old = w->shadow;
|
|
|
|
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
w->shadow = (UNSET == w->shadow_force ?
|
|
|
|
(ps->o.wintype_shadow[w->window_type]
|
|
|
|
&& !win_match(w, ps->o.shadow_blacklist, &w->cache_sblst)
|
|
|
|
&& !(ps->o.shadow_ignore_shaped && w->bounding_shaped
|
|
|
|
&& !w->rounded_corners)
|
|
|
|
&& !(ps->o.respect_prop_shadow && 0 == w->prop_shadow))
|
|
|
|
: w->shadow_force);
|
|
|
|
|
|
|
|
// Window extents need update on shadow state change
|
|
|
|
if (w->shadow != shadow_old) {
|
|
|
|
// Shadow geometry currently doesn't change on shadow state change
|
|
|
|
// calc_shadow_geometry(ps, w);
|
|
|
|
if (w->extents) {
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
// Mark the old extents as damaged if the shadow is removed
|
|
|
|
if (!w->shadow)
|
|
|
|
add_damage(ps, w->extents);
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
else
|
|
|
|
free_region(ps, &w->extents);
|
|
|
|
w->extents = win_extents(ps, w);
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
// Mark the new extents as damaged if the shadow is added
|
|
|
|
if (w->shadow)
|
|
|
|
add_damage_win(ps, w);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Update cache data in struct _win that depends on window size.
|
|
|
|
*/
|
|
|
|
|
|
|
|
static void
|
|
|
|
calc_win_size(session_t *ps, win *w) {
|
|
|
|
w->widthb = w->a.width + w->a.border_width * 2;
|
|
|
|
w->heightb = w->a.height + w->a.border_width * 2;
|
|
|
|
calc_shadow_geometry(ps, w);
|
|
|
|
w->flags |= WFLAG_SIZE_CHANGE;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Calculate and update geometry of the shadow of a window.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
calc_shadow_geometry(session_t *ps, win *w) {
|
|
|
|
w->shadow_dx = ps->o.shadow_offset_x;
|
|
|
|
w->shadow_dy = ps->o.shadow_offset_y;
|
|
|
|
w->shadow_width = w->widthb + ps->gaussian_map->size;
|
|
|
|
w->shadow_height = w->heightb + ps->gaussian_map->size;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Mark a window as the client window of another.
|
|
|
|
*
|
|
|
|
* @param ps current session
|
|
|
|
* @param w struct _win of the parent window
|
|
|
|
* @param client window ID of the client window
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
win_mark_client(session_t *ps, win *w, Window client) {
|
|
|
|
w->client_win = client;
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
|
|
|
|
// If the window isn't mapped yet, stop here, as the function will be
|
|
|
|
// called in map_win()
|
|
|
|
if (IsViewable != w->a.map_state)
|
|
|
|
return;
|
|
|
|
|
|
|
|
XSelectInput(ps->dpy, client, determine_evmask(ps, client, WIN_EVMODE_CLIENT));
|
|
|
|
|
|
|
|
// Make sure the XSelectInput() requests are sent
|
|
|
|
XSync(ps->dpy, False);
|
|
|
|
|
|
|
|
// Get frame widths if needed
|
|
|
|
if (ps->o.frame_opacity) {
|
|
|
|
get_frame_extents(ps, w, client);
|
|
|
|
}
|
|
|
|
|
|
|
|
// Detect window type here
|
|
|
|
if (WINTYPE_UNKNOWN == w->window_type)
|
|
|
|
w->window_type = wid_get_prop_wintype(ps, w->client_win);
|
|
|
|
|
|
|
|
// Conform to EWMH standard, if _NET_WM_WINDOW_TYPE is not present, take
|
|
|
|
// override-redirect windows or windows without WM_TRANSIENT_FOR as
|
|
|
|
// _NET_WM_WINDOW_TYPE_NORMAL, otherwise as _NET_WM_WINDOW_TYPE_DIALOG.
|
|
|
|
if (WINTYPE_UNKNOWN == w->window_type) {
|
|
|
|
if (w->a.override_redirect
|
|
|
|
|| !wid_has_prop(ps, client, ps->atom_transient))
|
|
|
|
w->window_type = WINTYPE_NORMAL;
|
|
|
|
else
|
|
|
|
w->window_type = WINTYPE_DIALOG;
|
|
|
|
}
|
|
|
|
|
|
|
|
// Get window group
|
|
|
|
if (ps->o.track_leader)
|
|
|
|
win_update_leader(ps, w);
|
|
|
|
|
|
|
|
// Get window name and class if we are tracking them
|
|
|
|
if (ps->o.track_wdata) {
|
|
|
|
win_get_name(ps, w);
|
|
|
|
win_get_class(ps, w);
|
|
|
|
win_get_role(ps, w);
|
|
|
|
}
|
|
|
|
|
|
|
|
// Update window focus state
|
|
|
|
win_update_focused(ps, w);
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Unmark current client window of a window.
|
|
|
|
*
|
|
|
|
* @param ps current session
|
|
|
|
* @param w struct _win of the parent window
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
win_unmark_client(session_t *ps, win *w) {
|
|
|
|
Window client = w->client_win;
|
|
|
|
|
|
|
|
w->client_win = None;
|
|
|
|
|
|
|
|
// Recheck event mask
|
|
|
|
XSelectInput(ps->dpy, client,
|
|
|
|
determine_evmask(ps, client, WIN_EVMODE_UNKNOWN));
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Recheck client window of a window.
|
|
|
|
*
|
|
|
|
* @param ps current session
|
|
|
|
* @param w struct _win of the parent window
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
win_recheck_client(session_t *ps, win *w) {
|
|
|
|
// Initialize wmwin to false
|
|
|
|
w->wmwin = false;
|
|
|
|
|
|
|
|
// Look for the client window
|
|
|
|
|
|
|
|
// Always recursively look for a window with WM_STATE, as Fluxbox
|
|
|
|
// sets override-redirect flags on all frame windows.
|
|
|
|
Window cw = find_client_win(ps, w->id);
|
|
|
|
#ifdef DEBUG_CLIENTWIN
|
|
|
|
if (cw)
|
|
|
|
printf_dbgf("(%#010lx): client %#010lx\n", w->id, cw);
|
|
|
|
#endif
|
|
|
|
// Set a window's client window to itself if we couldn't find a
|
|
|
|
// client window
|
|
|
|
if (!cw) {
|
|
|
|
cw = w->id;
|
|
|
|
w->wmwin = !w->a.override_redirect;
|
|
|
|
#ifdef DEBUG_CLIENTWIN
|
|
|
|
printf_dbgf("(%#010lx): client self (%s)\n", w->id,
|
|
|
|
(w->wmwin ? "wmwin": "override-redirected"));
|
|
|
|
#endif
|
|
|
|
}
|
|
|
|
|
|
|
|
// Unmark the old one
|
|
|
|
if (w->client_win && w->client_win != cw)
|
|
|
|
win_unmark_client(ps, w);
|
|
|
|
|
|
|
|
// Mark the new one
|
|
|
|
win_mark_client(ps, w, cw);
|
|
|
|
}
|
|
|
|
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
static bool
|
|
|
|
add_win(session_t *ps, Window id, Window prev) {
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
const static win win_def = {
|
|
|
|
.next = NULL,
|
|
|
|
.prev_trans = NULL,
|
|
|
|
|
|
|
|
.id = None,
|
|
|
|
.a = { },
|
|
|
|
.mode = WMODE_TRANS,
|
|
|
|
.damaged = false,
|
|
|
|
.damage = None,
|
|
|
|
.pixmap = None,
|
|
|
|
.picture = None,
|
|
|
|
.border_size = None,
|
|
|
|
.extents = None,
|
|
|
|
.flags = 0,
|
|
|
|
.need_configure = false,
|
|
|
|
.queue_configure = { },
|
|
|
|
.reg_ignore = None,
|
|
|
|
.destroyed = false,
|
|
|
|
.widthb = 0,
|
|
|
|
.heightb = 0,
|
|
|
|
.bounding_shaped = false,
|
|
|
|
.rounded_corners = false,
|
|
|
|
.to_paint = false,
|
|
|
|
|
|
|
|
.client_win = None,
|
|
|
|
.window_type = WINTYPE_UNKNOWN,
|
|
|
|
.wmwin = false,
|
|
|
|
.leader = None,
|
|
|
|
.cache_leader = None,
|
|
|
|
|
|
|
|
.focused = false,
|
|
|
|
.focused_force = UNSET,
|
|
|
|
.focused_real = false,
|
|
|
|
|
|
|
|
.name = NULL,
|
|
|
|
.class_instance = NULL,
|
|
|
|
.class_general = NULL,
|
|
|
|
.role = NULL,
|
|
|
|
.cache_sblst = NULL,
|
|
|
|
.cache_fblst = NULL,
|
|
|
|
.cache_fcblst = NULL,
|
|
|
|
|
|
|
|
.opacity = 0,
|
|
|
|
.opacity_tgt = 0,
|
|
|
|
.opacity_prop = OPAQUE,
|
|
|
|
.opacity_prop_client = OPAQUE,
|
|
|
|
.alpha_pict = None,
|
|
|
|
|
|
|
|
.fade = false,
|
|
|
|
.fade_callback = NULL,
|
|
|
|
|
|
|
|
.frame_opacity = 0.0,
|
|
|
|
.frame_alpha_pict = None,
|
|
|
|
.left_width = 0,
|
|
|
|
.right_width = 0,
|
|
|
|
.top_width = 0,
|
|
|
|
.bottom_width = 0,
|
|
|
|
|
|
|
|
.shadow = false,
|
|
|
|
.shadow_force = UNSET,
|
|
|
|
.shadow_opacity = 0.0,
|
|
|
|
.shadow_dx = 0,
|
|
|
|
.shadow_dy = 0,
|
|
|
|
.shadow_width = 0,
|
|
|
|
.shadow_height = 0,
|
|
|
|
.shadow_pict = None,
|
|
|
|
.shadow_alpha_pict = None,
|
|
|
|
.prop_shadow = -1,
|
|
|
|
|
|
|
|
.dim = false,
|
|
|
|
.dim_alpha_pict = None,
|
|
|
|
};
|
|
|
|
|
|
|
|
// Reject overlay window and already added windows
|
|
|
|
if (id == ps->overlay || find_win(ps, id)) {
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
// Allocate and initialize the new win structure
|
|
|
|
win *new = malloc(sizeof(win));
|
|
|
|
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
if (!new) {
|
|
|
|
printf_errf("(%#010lx): Failed to allocate memory for the new window.", id);
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
|
|
|
memcpy(new, &win_def, sizeof(win));
|
|
|
|
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
// Find window insertion point
|
|
|
|
win **p = NULL;
|
|
|
|
if (prev) {
|
|
|
|
for (p = &ps->list; *p; p = &(*p)->next) {
|
|
|
|
if ((*p)->id == prev && !(*p)->destroyed)
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
p = &ps->list;
|
|
|
|
}
|
|
|
|
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
// Fill structure
|
|
|
|
new->id = id;
|
|
|
|
set_ignore_next(ps);
|
|
|
|
|
|
|
|
if (!XGetWindowAttributes(ps->dpy, id, &new->a)) {
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
// Failed to get window attributes. Which probably means, the window
|
|
|
|
// is gone already.
|
|
|
|
free(new);
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
|
|
|
// Delay window mapping
|
|
|
|
int map_state = new->a.map_state;
|
|
|
|
assert(IsViewable == map_state || IsUnmapped == map_state);
|
|
|
|
new->a.map_state = IsUnmapped;
|
|
|
|
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
// Create Damage for window
|
|
|
|
if (InputOutput == new->a.class) {
|
|
|
|
set_ignore_next(ps);
|
|
|
|
new->damage = XDamageCreate(ps->dpy, id, XDamageReportNonEmpty);
|
|
|
|
}
|
|
|
|
|
|
|
|
calc_win_size(ps, new);
|
|
|
|
|
|
|
|
new->next = *p;
|
|
|
|
*p = new;
|
|
|
|
|
|
|
|
if (map_state == IsViewable) {
|
|
|
|
map_win(ps, id);
|
|
|
|
}
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
restack_win(session_t *ps, win *w, Window new_above) {
|
|
|
|
Window old_above;
|
|
|
|
|
|
|
|
update_reg_ignore_expire(ps, w);
|
|
|
|
|
|
|
|
if (w->next) {
|
|
|
|
old_above = w->next->id;
|
|
|
|
} else {
|
|
|
|
old_above = None;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (old_above != new_above) {
|
|
|
|
win **prev;
|
|
|
|
|
|
|
|
/* unhook */
|
|
|
|
for (prev = &ps->list; *prev; prev = &(*prev)->next) {
|
|
|
|
if ((*prev) == w) break;
|
|
|
|
}
|
|
|
|
|
|
|
|
*prev = w->next;
|
|
|
|
|
|
|
|
/* rehook */
|
|
|
|
for (prev = &ps->list; *prev; prev = &(*prev)->next) {
|
|
|
|
if ((*prev)->id == new_above && !(*prev)->destroyed)
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
w->next = *prev;
|
|
|
|
*prev = w;
|
|
|
|
|
|
|
|
#ifdef DEBUG_RESTACK
|
|
|
|
{
|
|
|
|
const char *desc;
|
|
|
|
char *window_name = NULL;
|
|
|
|
bool to_free;
|
|
|
|
win* c = ps->list;
|
|
|
|
|
|
|
|
printf_dbgf("(%#010lx, %#010lx): "
|
|
|
|
"Window stack modified. Current stack:\n", w->id, new_above);
|
|
|
|
|
|
|
|
for (; c; c = c->next) {
|
|
|
|
window_name = "(Failed to get title)";
|
|
|
|
|
|
|
|
to_free = ev_window_name(ps, c->id, &window_name);
|
|
|
|
|
|
|
|
desc = "";
|
|
|
|
if (c->destroyed) desc = "(D) ";
|
|
|
|
printf("%#010lx \"%s\" %s-> ", c->id, window_name, desc);
|
|
|
|
|
|
|
|
if (to_free) {
|
|
|
|
XFree(window_name);
|
|
|
|
window_name = NULL;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
fputs("\n", stdout);
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
configure_win(session_t *ps, XConfigureEvent *ce) {
|
|
|
|
if (ce->window == ps->root) {
|
|
|
|
if (ps->tgt_buffer) {
|
|
|
|
XRenderFreePicture(ps->dpy, ps->tgt_buffer);
|
|
|
|
ps->tgt_buffer = None;
|
|
|
|
}
|
|
|
|
ps->root_width = ce->width;
|
|
|
|
ps->root_height = ce->height;
|
|
|
|
|
|
|
|
rebuild_screen_reg(ps);
|
|
|
|
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
win *w = find_win(ps, ce->window);
|
|
|
|
XserverRegion damage = None;
|
|
|
|
|
|
|
|
if (!w)
|
|
|
|
return;
|
|
|
|
|
|
|
|
if (w->a.map_state == IsUnmapped) {
|
|
|
|
/* save the configure event for when the window maps */
|
|
|
|
w->need_configure = true;
|
|
|
|
w->queue_configure = *ce;
|
|
|
|
restack_win(ps, w, ce->above);
|
|
|
|
} else {
|
|
|
|
if (!(w->need_configure)) {
|
|
|
|
restack_win(ps, w, ce->above);
|
|
|
|
}
|
|
|
|
|
|
|
|
// Windows restack (including window restacks happened when this
|
|
|
|
// window is not mapped) could mess up all reg_ignore
|
|
|
|
ps->reg_ignore_expire = true;
|
|
|
|
|
|
|
|
w->need_configure = false;
|
|
|
|
|
|
|
|
damage = XFixesCreateRegion(ps->dpy, 0, 0);
|
|
|
|
if (w->extents != None) {
|
|
|
|
XFixesCopyRegion(ps->dpy, damage, w->extents);
|
|
|
|
}
|
|
|
|
|
|
|
|
// If window geometry did not change, don't free extents here
|
|
|
|
if (w->a.x != ce->x || w->a.y != ce->y
|
|
|
|
|| w->a.width != ce->width || w->a.height != ce->height
|
|
|
|
|| w->a.border_width != ce->border_width) {
|
|
|
|
free_region(ps, &w->extents);
|
|
|
|
free_region(ps, &w->border_size);
|
|
|
|
}
|
|
|
|
|
|
|
|
w->a.x = ce->x;
|
|
|
|
w->a.y = ce->y;
|
|
|
|
|
|
|
|
if (w->a.width != ce->width || w->a.height != ce->height
|
|
|
|
|| w->a.border_width != ce->border_width) {
|
|
|
|
free_pixmap(ps, &w->pixmap);
|
|
|
|
free_picture(ps, &w->picture);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (w->a.width != ce->width || w->a.height != ce->height
|
|
|
|
|| w->a.border_width != ce->border_width) {
|
|
|
|
w->a.width = ce->width;
|
|
|
|
w->a.height = ce->height;
|
|
|
|
w->a.border_width = ce->border_width;
|
|
|
|
calc_win_size(ps, w);
|
|
|
|
|
|
|
|
// Rounded corner detection is affected by window size
|
|
|
|
if (ps->shape_exists && ps->o.shadow_ignore_shaped
|
|
|
|
&& ps->o.detect_rounded_corners && w->bounding_shaped)
|
|
|
|
win_update_shape(ps, w);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (damage) {
|
|
|
|
XserverRegion extents = win_extents(ps, w);
|
|
|
|
XFixesUnionRegion(ps->dpy, damage, damage, extents);
|
|
|
|
XFixesDestroyRegion(ps->dpy, extents);
|
|
|
|
add_damage(ps, damage);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
w->a.override_redirect = ce->override_redirect;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
circulate_win(session_t *ps, XCirculateEvent *ce) {
|
|
|
|
win *w = find_win(ps, ce->window);
|
|
|
|
Window new_above;
|
|
|
|
|
|
|
|
if (!w) return;
|
|
|
|
|
|
|
|
if (ce->place == PlaceOnTop) {
|
|
|
|
new_above = ps->list->id;
|
|
|
|
} else {
|
|
|
|
new_above = None;
|
|
|
|
}
|
|
|
|
|
|
|
|
restack_win(ps, w, new_above);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
finish_destroy_win(session_t *ps, Window id) {
|
|
|
|
win **prev, *w;
|
|
|
|
|
|
|
|
for (prev = &ps->list; (w = *prev); prev = &w->next) {
|
|
|
|
if (w->id == id && w->destroyed) {
|
|
|
|
finish_unmap_win(ps, w);
|
|
|
|
*prev = w->next;
|
|
|
|
|
|
|
|
// Clear active_win if it's pointing to the destroyed window
|
|
|
|
if (w == ps->active_win)
|
|
|
|
ps->active_win = NULL;
|
|
|
|
|
|
|
|
free_win_res(ps, w);
|
|
|
|
|
|
|
|
free(w);
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
destroy_callback(session_t *ps, win *w) {
|
|
|
|
finish_destroy_win(ps, w->id);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
destroy_win(session_t *ps, Window id) {
|
|
|
|
win *w = find_win(ps, id);
|
|
|
|
|
|
|
|
if (w) {
|
|
|
|
w->destroyed = true;
|
|
|
|
|
|
|
|
// Fading out the window
|
|
|
|
w->flags |= WFLAG_OPCT_CHANGE;
|
|
|
|
set_fade_callback(ps, w, destroy_callback, false);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void
|
|
|
|
root_damaged(session_t *ps) {
|
|
|
|
if (ps->root_tile) {
|
|
|
|
XClearArea(ps->dpy, ps->root, 0, 0, 0, 0, true);
|
|
|
|
// if (ps->root_picture != ps->root_tile) {
|
|
|
|
XRenderFreePicture(ps->dpy, ps->root_tile);
|
|
|
|
ps->root_tile = None;
|
|
|
|
/* }
|
|
|
|
if (root_damage) {
|
|
|
|
XserverRegion parts = XFixesCreateRegion(ps->dpy, 0, 0);
|
|
|
|
XDamageSubtract(ps->dpy, root_damage, None, parts);
|
|
|
|
add_damage(ps, parts);
|
|
|
|
} */
|
|
|
|
}
|
|
|
|
// Mark screen damaged if we are painting on overlay
|
|
|
|
if (ps->o.paint_on_overlay)
|
|
|
|
add_damage(ps, get_screen_region(ps));
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
damage_win(session_t *ps, XDamageNotifyEvent *de) {
|
|
|
|
/*
|
|
|
|
if (ps->root == de->drawable) {
|
|
|
|
root_damaged();
|
|
|
|
return;
|
|
|
|
} */
|
|
|
|
|
|
|
|
win *w = find_win(ps, de->drawable);
|
|
|
|
|
|
|
|
if (!w) return;
|
|
|
|
|
|
|
|
repair_win(ps, w);
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Xlib error handler function.
|
|
|
|
*/
|
|
|
|
static int
|
|
|
|
error(Display __attribute__((unused)) *dpy, XErrorEvent *ev) {
|
|
|
|
session_t * const ps = ps_g;
|
|
|
|
|
|
|
|
int o;
|
|
|
|
const char *name = "Unknown";
|
|
|
|
|
|
|
|
if (should_ignore(ps, ev->serial)) {
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (ev->request_code == ps->composite_opcode
|
|
|
|
&& ev->minor_code == X_CompositeRedirectSubwindows) {
|
|
|
|
fprintf(stderr, "Another composite manager is already running\n");
|
|
|
|
exit(1);
|
|
|
|
}
|
|
|
|
|
|
|
|
#define CASESTRRET2(s) case s: name = #s; break
|
|
|
|
|
|
|
|
o = ev->error_code - ps->xfixes_error;
|
|
|
|
switch (o) {
|
|
|
|
CASESTRRET2(BadRegion);
|
|
|
|
}
|
|
|
|
|
|
|
|
o = ev->error_code - ps->damage_error;
|
|
|
|
switch (o) {
|
|
|
|
CASESTRRET2(BadDamage);
|
|
|
|
}
|
|
|
|
|
|
|
|
o = ev->error_code - ps->render_error;
|
|
|
|
switch (o) {
|
|
|
|
CASESTRRET2(BadPictFormat);
|
|
|
|
CASESTRRET2(BadPicture);
|
|
|
|
CASESTRRET2(BadPictOp);
|
|
|
|
CASESTRRET2(BadGlyphSet);
|
|
|
|
CASESTRRET2(BadGlyph);
|
|
|
|
}
|
|
|
|
|
|
|
|
switch (ev->error_code) {
|
|
|
|
CASESTRRET2(BadAccess);
|
|
|
|
CASESTRRET2(BadAlloc);
|
|
|
|
CASESTRRET2(BadAtom);
|
|
|
|
CASESTRRET2(BadColor);
|
|
|
|
CASESTRRET2(BadCursor);
|
|
|
|
CASESTRRET2(BadDrawable);
|
|
|
|
CASESTRRET2(BadFont);
|
|
|
|
CASESTRRET2(BadGC);
|
|
|
|
CASESTRRET2(BadIDChoice);
|
|
|
|
CASESTRRET2(BadImplementation);
|
|
|
|
CASESTRRET2(BadLength);
|
|
|
|
CASESTRRET2(BadMatch);
|
|
|
|
CASESTRRET2(BadName);
|
|
|
|
CASESTRRET2(BadPixmap);
|
|
|
|
CASESTRRET2(BadRequest);
|
|
|
|
CASESTRRET2(BadValue);
|
|
|
|
CASESTRRET2(BadWindow);
|
|
|
|
}
|
|
|
|
|
|
|
|
#undef CASESTRRET2
|
|
|
|
|
|
|
|
print_timestamp(ps);
|
|
|
|
printf("error %d (%s) request %d minor %d serial %lu\n",
|
|
|
|
ev->error_code, name, ev->request_code,
|
|
|
|
ev->minor_code, ev->serial);
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
expose_root(session_t *ps, XRectangle *rects, int nrects) {
|
|
|
|
XserverRegion region = XFixesCreateRegion(ps->dpy, rects, nrects);
|
|
|
|
add_damage(ps, region);
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Get the value of a type-<code>Window</code> property of a window.
|
|
|
|
*
|
|
|
|
* @return the value if successful, 0 otherwise
|
|
|
|
*/
|
|
|
|
static Window
|
|
|
|
wid_get_prop_window(session_t *ps, Window wid, Atom aprop) {
|
|
|
|
// Get the attribute
|
|
|
|
Window p = None;
|
|
|
|
winprop_t prop = wid_get_prop(ps, wid, aprop, 1L, XA_WINDOW, 32);
|
|
|
|
|
|
|
|
// Return it
|
|
|
|
if (prop.nitems) {
|
|
|
|
p = *prop.data.p32;
|
|
|
|
}
|
|
|
|
|
|
|
|
free_winprop(&prop);
|
|
|
|
|
|
|
|
return p;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Update focused state of a window.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
win_update_focused(session_t *ps, win *w) {
|
|
|
|
bool focused_old = w->focused;
|
|
|
|
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
if (UNSET != w->focused_force) {
|
|
|
|
w->focused = w->focused_force;
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
w->focused = w->focused_real;
|
|
|
|
|
|
|
|
// Use wintype_focus, and treat WM windows and override-redirected
|
|
|
|
// windows specially
|
|
|
|
if (ps->o.wintype_focus[w->window_type]
|
|
|
|
|| (ps->o.mark_wmwin_focused && w->wmwin)
|
|
|
|
|| (ps->o.mark_ovredir_focused
|
|
|
|
&& w->id == w->client_win && !w->wmwin)
|
|
|
|
|| win_match(w, ps->o.focus_blacklist, &w->cache_fcblst))
|
|
|
|
w->focused = true;
|
|
|
|
|
|
|
|
// If window grouping detection is enabled, mark the window active if
|
|
|
|
// its group is
|
|
|
|
if (ps->o.track_leader && ps->active_leader
|
|
|
|
&& win_get_leader(ps, w) == ps->active_leader) {
|
|
|
|
w->focused = true;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
if (w->focused != focused_old)
|
|
|
|
w->flags |= WFLAG_OPCT_CHANGE;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Set real focused state of a window.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
win_set_focused(session_t *ps, win *w, bool focused) {
|
|
|
|
// Unmapped windows will have their focused state reset on map
|
|
|
|
if (IsUnmapped == w->a.map_state)
|
|
|
|
return;
|
|
|
|
|
|
|
|
if (w->focused_real != focused) {
|
|
|
|
w->focused_real = focused;
|
|
|
|
|
|
|
|
// If window grouping detection is enabled
|
|
|
|
if (ps->o.track_leader) {
|
|
|
|
Window leader = win_get_leader(ps, w);
|
|
|
|
|
|
|
|
// If the window gets focused, replace the old active_leader
|
|
|
|
if (w->focused_real && leader != ps->active_leader) {
|
|
|
|
Window active_leader_old = ps->active_leader;
|
|
|
|
|
|
|
|
ps->active_leader = leader;
|
|
|
|
|
|
|
|
group_update_focused(ps, active_leader_old);
|
|
|
|
group_update_focused(ps, leader);
|
|
|
|
}
|
|
|
|
// If the group get unfocused, remove it from active_leader
|
|
|
|
else if (!w->focused_real && leader && leader == ps->active_leader
|
|
|
|
&& !group_is_focused(ps, leader)) {
|
|
|
|
ps->active_leader = None;
|
|
|
|
group_update_focused(ps, leader);
|
|
|
|
}
|
|
|
|
|
|
|
|
// The window itself must be updated anyway
|
|
|
|
win_update_focused(ps, w);
|
|
|
|
}
|
|
|
|
// Otherwise, only update the window itself
|
|
|
|
else {
|
|
|
|
win_update_focused(ps, w);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
/**
|
|
|
|
* Update leader of a window.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
win_update_leader(session_t *ps, win *w) {
|
|
|
|
Window leader = None;
|
|
|
|
|
|
|
|
// Read the leader properties
|
|
|
|
if (ps->o.detect_transient && !leader)
|
|
|
|
leader = wid_get_prop_window(ps, w->client_win, ps->atom_transient);
|
|
|
|
|
|
|
|
if (ps->o.detect_client_leader && !leader)
|
|
|
|
leader = wid_get_prop_window(ps, w->client_win, ps->atom_client_leader);
|
|
|
|
|
|
|
|
win_set_leader(ps, w, leader);
|
|
|
|
|
|
|
|
#ifdef DEBUG_LEADER
|
|
|
|
printf_dbgf("(%#010lx): client %#010lx, leader %#010lx, cache %#010lx\n", w->id, w->client_win, w->leader, win_get_leader(ps, w));
|
|
|
|
#endif
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Set leader of a window.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
win_set_leader(session_t *ps, win *w, Window nleader) {
|
|
|
|
// If the leader changes
|
|
|
|
if (w->leader != nleader) {
|
|
|
|
Window cache_leader_old = win_get_leader(ps, w);
|
|
|
|
|
|
|
|
w->leader = nleader;
|
|
|
|
|
|
|
|
// Forcefully do this to deal with the case when a child window
|
|
|
|
// gets mapped before parent, or when the window is a waypoint
|
|
|
|
clear_cache_win_leaders(ps);
|
|
|
|
|
|
|
|
// Update the old and new window group and active_leader if the window
|
|
|
|
// could affect their state.
|
|
|
|
Window cache_leader = win_get_leader(ps, w);
|
|
|
|
if (w->focused_real && cache_leader_old != cache_leader) {
|
|
|
|
ps->active_leader = cache_leader;
|
|
|
|
|
|
|
|
group_update_focused(ps, cache_leader_old);
|
|
|
|
group_update_focused(ps, cache_leader);
|
|
|
|
}
|
|
|
|
// Otherwise, at most the window itself is affected
|
|
|
|
else {
|
|
|
|
win_update_focused(ps, w);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Internal function of win_get_leader().
|
|
|
|
*/
|
|
|
|
static Window
|
|
|
|
win_get_leader_raw(session_t *ps, win *w, int recursions) {
|
|
|
|
// Rebuild the cache if needed
|
|
|
|
if (!w->cache_leader && (w->client_win || w->leader)) {
|
|
|
|
// Leader defaults to client window
|
|
|
|
if (!(w->cache_leader = w->leader))
|
|
|
|
w->cache_leader = w->client_win;
|
|
|
|
|
|
|
|
// If the leader of this window isn't itself, look for its ancestors
|
|
|
|
if (w->cache_leader && w->cache_leader != w->client_win) {
|
|
|
|
win *wp = find_toplevel(ps, w->cache_leader);
|
|
|
|
if (wp) {
|
|
|
|
// Dead loop?
|
|
|
|
if (recursions > WIN_GET_LEADER_MAX_RECURSION)
|
|
|
|
return None;
|
|
|
|
|
|
|
|
w->cache_leader = win_get_leader_raw(ps, wp, recursions + 1);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return w->cache_leader;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Get the value of a text property of a window.
|
|
|
|
*/
|
|
|
|
static bool
|
|
|
|
wid_get_text_prop(session_t *ps, Window wid, Atom prop,
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
char ***pstrlst, int *pnstr) {
|
|
|
|
XTextProperty text_prop = { NULL, None, 0, 0 };
|
|
|
|
|
|
|
|
if (!(XGetTextProperty(ps->dpy, wid, &text_prop, prop) && text_prop.value))
|
|
|
|
return false;
|
|
|
|
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
if (Success !=
|
|
|
|
XmbTextPropertyToTextList(ps->dpy, &text_prop, pstrlst, pnstr)
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
|| !*pnstr) {
|
|
|
|
*pnstr = 0;
|
|
|
|
if (*pstrlst)
|
|
|
|
XFreeStringList(*pstrlst);
|
|
|
|
XFree(text_prop.value);
|
|
|
|
return false;
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
}
|
|
|
|
|
|
|
|
XFree(text_prop.value);
|
|
|
|
return true;
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Get the name of a window from window ID.
|
|
|
|
*/
|
|
|
|
static bool
|
|
|
|
wid_get_name(session_t *ps, Window wid, char **name) {
|
|
|
|
XTextProperty text_prop = { NULL, None, 0, 0 };
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
char **strlst = NULL;
|
|
|
|
int nstr = 0;
|
|
|
|
|
|
|
|
if (!(wid_get_text_prop(ps, wid, ps->atom_name_ewmh, &strlst, &nstr))) {
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
#ifdef DEBUG_WINDATA
|
|
|
|
printf_dbgf("(%#010lx): _NET_WM_NAME unset, falling back to WM_NAME.\n", wid);
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
#endif
|
|
|
|
|
|
|
|
if (!(XGetWMName(ps->dpy, wid, &text_prop) && text_prop.value)) {
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
if (Success !=
|
|
|
|
XmbTextPropertyToTextList(ps->dpy, &text_prop, &strlst, &nstr)
|
|
|
|
|| !nstr || !strlst) {
|
|
|
|
if (strlst)
|
|
|
|
XFreeStringList(strlst);
|
|
|
|
XFree(text_prop.value);
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
XFree(text_prop.value);
|
|
|
|
}
|
|
|
|
|
|
|
|
*name = mstrcpy(strlst[0]);
|
|
|
|
|
|
|
|
XFreeStringList(strlst);
|
|
|
|
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Get the role of a window from window ID.
|
|
|
|
*/
|
|
|
|
static bool
|
|
|
|
wid_get_role(session_t *ps, Window wid, char **role) {
|
|
|
|
char **strlst = NULL;
|
|
|
|
int nstr = 0;
|
|
|
|
|
|
|
|
if (!wid_get_text_prop(ps, wid, ps->atom_role, &strlst, &nstr)) {
|
|
|
|
return false;
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
}
|
|
|
|
|
|
|
|
*role = mstrcpy(strlst[0]);
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
|
|
|
|
XFreeStringList(strlst);
|
|
|
|
|
|
|
|
return true;
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Retrieve a string property of a window and update its <code>win</code>
|
|
|
|
* structure.
|
|
|
|
*/
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
static int
|
|
|
|
win_get_prop_str(session_t *ps, win *w, char **tgt,
|
|
|
|
bool (*func_wid_get_prop_str)(session_t *ps, Window wid, char **tgt)) {
|
|
|
|
int ret = -1;
|
|
|
|
char *prop_old = *tgt;
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
|
|
|
|
// Can't do anything if there's no client window
|
|
|
|
if (!w->client_win)
|
|
|
|
return false;
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
|
|
|
|
// Get the property
|
|
|
|
ret = func_wid_get_prop_str(ps, w->client_win, tgt);
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
|
|
|
|
// Return -1 if func_wid_get_prop_str() failed, 0 if the property
|
|
|
|
// doesn't change, 1 if it changes
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
if (!ret)
|
|
|
|
ret = -1;
|
|
|
|
else if (prop_old && !strcmp(*tgt, prop_old))
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
ret = 0;
|
|
|
|
else
|
|
|
|
ret = 1;
|
|
|
|
|
|
|
|
// Keep the old property if there's no new one
|
|
|
|
if (*tgt != prop_old)
|
|
|
|
free(prop_old);
|
|
|
|
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
return ret;
|
|
|
|
}
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
|
|
|
|
/**
|
|
|
|
* Retrieve the <code>WM_CLASS</code> of a window and update its
|
|
|
|
* <code>win</code> structure.
|
|
|
|
*/
|
|
|
|
static bool
|
|
|
|
win_get_class(session_t *ps, win *w) {
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
char **strlst = NULL;
|
|
|
|
int nstr = 0;
|
|
|
|
|
|
|
|
// Can't do anything if there's no client window
|
|
|
|
if (!w->client_win)
|
|
|
|
return false;
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
|
|
|
|
// Free and reset old strings
|
|
|
|
free(w->class_instance);
|
|
|
|
free(w->class_general);
|
|
|
|
w->class_instance = NULL;
|
|
|
|
w->class_general = NULL;
|
|
|
|
|
|
|
|
// Retrieve the property string list
|
|
|
|
if (!wid_get_text_prop(ps, w->client_win, ps->atom_class, &strlst, &nstr))
|
|
|
|
return false;
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
|
|
|
|
// Copy the strings if successful
|
|
|
|
w->class_instance = mstrcpy(strlst[0]);
|
|
|
|
|
|
|
|
if (nstr > 1)
|
|
|
|
w->class_general = mstrcpy(strlst[1]);
|
|
|
|
|
|
|
|
XFreeStringList(strlst);
|
|
|
|
|
|
|
|
#ifdef DEBUG_WINDATA
|
|
|
|
printf_dbgf("(%#010lx): client = %#010lx, "
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
"instance = \"%s\", general = \"%s\"\n",
|
|
|
|
w->id, w->client_win, w->class_instance, w->class_general);
|
|
|
|
#endif
|
|
|
|
|
|
|
|
return true;
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
}
|
|
|
|
|
|
|
|
#ifdef DEBUG_EVENTS
|
|
|
|
static int
|
|
|
|
ev_serial(XEvent *ev) {
|
|
|
|
if ((ev->type & 0x7f) != KeymapNotify) {
|
|
|
|
return ev->xany.serial;
|
|
|
|
}
|
|
|
|
return NextRequest(ev->xany.display);
|
|
|
|
}
|
|
|
|
|
|
|
|
static const char *
|
|
|
|
ev_name(session_t *ps, XEvent *ev) {
|
|
|
|
static char buf[128];
|
|
|
|
switch (ev->type & 0x7f) {
|
|
|
|
CASESTRRET(FocusIn);
|
|
|
|
CASESTRRET(FocusOut);
|
|
|
|
CASESTRRET(CreateNotify);
|
|
|
|
CASESTRRET(ConfigureNotify);
|
|
|
|
CASESTRRET(DestroyNotify);
|
|
|
|
CASESTRRET(MapNotify);
|
|
|
|
CASESTRRET(UnmapNotify);
|
|
|
|
CASESTRRET(ReparentNotify);
|
|
|
|
CASESTRRET(CirculateNotify);
|
|
|
|
CASESTRRET(Expose);
|
|
|
|
CASESTRRET(PropertyNotify);
|
|
|
|
CASESTRRET(ClientMessage);
|
|
|
|
default:
|
|
|
|
if (ev->type == ps->damage_event + XDamageNotify) {
|
|
|
|
return "Damage";
|
|
|
|
}
|
|
|
|
|
|
|
|
if (ps->shape_exists && ev->type == ps->shape_event) {
|
|
|
|
return "ShapeNotify";
|
|
|
|
}
|
|
|
|
|
|
|
|
sprintf(buf, "Event %d", ev->type);
|
|
|
|
|
|
|
|
return buf;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static Window
|
|
|
|
ev_window(session_t *ps, XEvent *ev) {
|
|
|
|
switch (ev->type) {
|
|
|
|
case FocusIn:
|
|
|
|
case FocusOut:
|
|
|
|
return ev->xfocus.window;
|
|
|
|
case CreateNotify:
|
|
|
|
return ev->xcreatewindow.window;
|
|
|
|
case ConfigureNotify:
|
|
|
|
return ev->xconfigure.window;
|
|
|
|
case DestroyNotify:
|
|
|
|
return ev->xdestroywindow.window;
|
|
|
|
case MapNotify:
|
|
|
|
return ev->xmap.window;
|
|
|
|
case UnmapNotify:
|
|
|
|
return ev->xunmap.window;
|
|
|
|
case ReparentNotify:
|
|
|
|
return ev->xreparent.window;
|
|
|
|
case CirculateNotify:
|
|
|
|
return ev->xcirculate.window;
|
|
|
|
case Expose:
|
|
|
|
return ev->xexpose.window;
|
|
|
|
case PropertyNotify:
|
|
|
|
return ev->xproperty.window;
|
|
|
|
case ClientMessage:
|
|
|
|
return ev->xclient.window;
|
|
|
|
default:
|
|
|
|
if (ev->type == ps->damage_event + XDamageNotify) {
|
|
|
|
return ((XDamageNotifyEvent *)ev)->drawable;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (ps->shape_exists && ev->type == ps->shape_event) {
|
|
|
|
return ((XShapeEvent *) ev)->window;
|
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline const char *
|
|
|
|
ev_focus_mode_name(XFocusChangeEvent* ev) {
|
|
|
|
switch (ev->mode) {
|
|
|
|
CASESTRRET(NotifyNormal);
|
|
|
|
CASESTRRET(NotifyWhileGrabbed);
|
|
|
|
CASESTRRET(NotifyGrab);
|
|
|
|
CASESTRRET(NotifyUngrab);
|
|
|
|
}
|
|
|
|
|
|
|
|
return "Unknown";
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline const char *
|
|
|
|
ev_focus_detail_name(XFocusChangeEvent* ev) {
|
|
|
|
switch (ev->detail) {
|
|
|
|
CASESTRRET(NotifyAncestor);
|
|
|
|
CASESTRRET(NotifyVirtual);
|
|
|
|
CASESTRRET(NotifyInferior);
|
|
|
|
CASESTRRET(NotifyNonlinear);
|
|
|
|
CASESTRRET(NotifyNonlinearVirtual);
|
|
|
|
CASESTRRET(NotifyPointer);
|
|
|
|
CASESTRRET(NotifyPointerRoot);
|
|
|
|
CASESTRRET(NotifyDetailNone);
|
|
|
|
}
|
|
|
|
|
|
|
|
return "Unknown";
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void
|
|
|
|
ev_focus_report(XFocusChangeEvent* ev) {
|
|
|
|
printf(" { mode: %s, detail: %s }\n", ev_focus_mode_name(ev),
|
|
|
|
ev_focus_detail_name(ev));
|
|
|
|
}
|
|
|
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
// === Events ===
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Determine whether we should respond to a <code>FocusIn/Out</code>
|
|
|
|
* event.
|
|
|
|
*/
|
|
|
|
inline static bool
|
|
|
|
ev_focus_accept(XFocusChangeEvent *ev) {
|
|
|
|
return ev->detail == NotifyNonlinear
|
|
|
|
|| ev->detail == NotifyNonlinearVirtual;
|
|
|
|
}
|
|
|
|
|
|
|
|
inline static void
|
|
|
|
ev_focus_in(session_t *ps, XFocusChangeEvent *ev) {
|
|
|
|
#ifdef DEBUG_EVENTS
|
|
|
|
ev_focus_report(ev);
|
|
|
|
#endif
|
|
|
|
|
|
|
|
if (!ev_focus_accept(ev))
|
|
|
|
return;
|
|
|
|
|
|
|
|
win *w = find_win(ps, ev->window);
|
|
|
|
|
|
|
|
// To deal with events sent from windows just destroyed
|
|
|
|
if (!w) return;
|
|
|
|
|
|
|
|
win_set_focused(ps, w, true);
|
|
|
|
}
|
|
|
|
|
|
|
|
inline static void
|
|
|
|
ev_focus_out(session_t *ps, XFocusChangeEvent *ev) {
|
|
|
|
#ifdef DEBUG_EVENTS
|
|
|
|
ev_focus_report(ev);
|
|
|
|
#endif
|
|
|
|
|
|
|
|
if (!ev_focus_accept(ev))
|
|
|
|
return;
|
|
|
|
|
|
|
|
win *w = find_win(ps, ev->window);
|
|
|
|
|
|
|
|
// To deal with events sent from windows just destroyed
|
|
|
|
if (!w) return;
|
|
|
|
|
|
|
|
win_set_focused(ps, w, false);
|
|
|
|
}
|
|
|
|
|
|
|
|
inline static void
|
|
|
|
ev_create_notify(session_t *ps, XCreateWindowEvent *ev) {
|
|
|
|
assert(ev->parent == ps->root);
|
|
|
|
add_win(ps, ev->window, 0);
|
|
|
|
}
|
|
|
|
|
|
|
|
inline static void
|
|
|
|
ev_configure_notify(session_t *ps, XConfigureEvent *ev) {
|
|
|
|
#ifdef DEBUG_EVENTS
|
|
|
|
printf(" { send_event: %d, "
|
|
|
|
" above: %#010lx, "
|
|
|
|
" override_redirect: %d }\n",
|
|
|
|
ev->send_event, ev->above, ev->override_redirect);
|
|
|
|
#endif
|
|
|
|
configure_win(ps, ev);
|
|
|
|
}
|
|
|
|
|
|
|
|
inline static void
|
|
|
|
ev_destroy_notify(session_t *ps, XDestroyWindowEvent *ev) {
|
|
|
|
destroy_win(ps, ev->window);
|
|
|
|
}
|
|
|
|
|
|
|
|
inline static void
|
|
|
|
ev_map_notify(session_t *ps, XMapEvent *ev) {
|
|
|
|
map_win(ps, ev->window);
|
|
|
|
}
|
|
|
|
|
|
|
|
inline static void
|
|
|
|
ev_unmap_notify(session_t *ps, XUnmapEvent *ev) {
|
|
|
|
unmap_win(ps, ev->window);
|
|
|
|
}
|
|
|
|
|
|
|
|
inline static void
|
|
|
|
ev_reparent_notify(session_t *ps, XReparentEvent *ev) {
|
|
|
|
if (ev->parent == ps->root) {
|
|
|
|
add_win(ps, ev->window, 0);
|
|
|
|
} else {
|
|
|
|
destroy_win(ps, ev->window);
|
|
|
|
|
|
|
|
// Reset event mask in case something wrong happens
|
|
|
|
XSelectInput(ps->dpy, ev->window,
|
|
|
|
determine_evmask(ps, ev->window, WIN_EVMODE_UNKNOWN));
|
|
|
|
|
|
|
|
// Check if the window is an undetected client window
|
|
|
|
// Firstly, check if it's a known client window
|
|
|
|
if (!find_toplevel(ps, ev->window)) {
|
|
|
|
// If not, look for its frame window
|
|
|
|
win *w_top = find_toplevel2(ps, ev->parent);
|
|
|
|
// If found, and the client window has not been determined, or its
|
|
|
|
// frame may not have a correct client, continue
|
|
|
|
if (w_top && (!w_top->client_win
|
|
|
|
|| w_top->client_win == w_top->id)) {
|
|
|
|
// If it has WM_STATE, mark it the client window
|
|
|
|
if (wid_has_prop(ps, ev->window, ps->atom_client)) {
|
|
|
|
w_top->wmwin = false;
|
|
|
|
win_unmark_client(ps, w_top);
|
|
|
|
win_mark_client(ps, w_top, ev->window);
|
|
|
|
}
|
|
|
|
// Otherwise, watch for WM_STATE on it
|
|
|
|
else {
|
|
|
|
XSelectInput(ps->dpy, ev->window,
|
|
|
|
determine_evmask(ps, ev->window, WIN_EVMODE_UNKNOWN)
|
|
|
|
| PropertyChangeMask);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
inline static void
|
|
|
|
ev_circulate_notify(session_t *ps, XCirculateEvent *ev) {
|
|
|
|
circulate_win(ps, ev);
|
|
|
|
}
|
|
|
|
|
|
|
|
inline static void
|
|
|
|
ev_expose(session_t *ps, XExposeEvent *ev) {
|
|
|
|
if (ev->window == ps->root || (ps->overlay && ev->window == ps->overlay)) {
|
|
|
|
int more = ev->count + 1;
|
|
|
|
if (ps->n_expose == ps->size_expose) {
|
|
|
|
if (ps->expose_rects) {
|
|
|
|
ps->expose_rects = realloc(ps->expose_rects,
|
|
|
|
(ps->size_expose + more) * sizeof(XRectangle));
|
|
|
|
ps->size_expose += more;
|
|
|
|
} else {
|
|
|
|
ps->expose_rects = malloc(more * sizeof(XRectangle));
|
|
|
|
ps->size_expose = more;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
ps->expose_rects[ps->n_expose].x = ev->x;
|
|
|
|
ps->expose_rects[ps->n_expose].y = ev->y;
|
|
|
|
ps->expose_rects[ps->n_expose].width = ev->width;
|
|
|
|
ps->expose_rects[ps->n_expose].height = ev->height;
|
|
|
|
ps->n_expose++;
|
|
|
|
|
|
|
|
if (ev->count == 0) {
|
|
|
|
expose_root(ps, ps->expose_rects, ps->n_expose);
|
|
|
|
ps->n_expose = 0;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Update current active window based on EWMH _NET_ACTIVE_WIN.
|
|
|
|
*
|
|
|
|
* Does not change anything if we fail to get the attribute or the window
|
|
|
|
* returned could not be found.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
update_ewmh_active_win(session_t *ps) {
|
|
|
|
// Search for the window
|
|
|
|
Window wid =
|
|
|
|
wid_get_prop_window(ps, ps->root, ps->atom_ewmh_active_win);
|
|
|
|
win *w = NULL;
|
|
|
|
|
|
|
|
if (wid && !(w = find_toplevel(ps, wid)))
|
|
|
|
if (!(w = find_win(ps, wid)))
|
|
|
|
w = find_toplevel2(ps, wid);
|
|
|
|
|
|
|
|
// Mark the window focused
|
|
|
|
if (w) {
|
|
|
|
if (ps->active_win && w != ps->active_win)
|
|
|
|
win_set_focused(ps, ps->active_win, false);
|
|
|
|
ps->active_win = w;
|
|
|
|
win_set_focused(ps, w, true);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
inline static void
|
|
|
|
ev_property_notify(session_t *ps, XPropertyEvent *ev) {
|
|
|
|
if (ps->root == ev->window) {
|
|
|
|
if (ps->o.track_focus && ps->o.use_ewmh_active_win
|
|
|
|
&& ps->atom_ewmh_active_win == ev->atom) {
|
|
|
|
update_ewmh_active_win(ps);
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
// Destroy the root "image" if the wallpaper probably changed
|
|
|
|
for (int p = 0; background_props_str[p]; p++) {
|
|
|
|
if (ev->atom == get_atom(ps, background_props_str[p])) {
|
|
|
|
root_damaged(ps);
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
// Unconcerned about any other proprties on root window
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
// If WM_STATE changes
|
|
|
|
if (ev->atom == ps->atom_client) {
|
|
|
|
// Check whether it could be a client window
|
|
|
|
if (!find_toplevel(ps, ev->window)) {
|
|
|
|
// Reset event mask anyway
|
|
|
|
XSelectInput(ps->dpy, ev->window,
|
|
|
|
determine_evmask(ps, ev->window, WIN_EVMODE_UNKNOWN));
|
|
|
|
|
|
|
|
win *w_top = find_toplevel2(ps, ev->window);
|
|
|
|
// Initialize client_win as early as possible
|
|
|
|
if (w_top && (!w_top->client_win || w_top->client_win == w_top->id)
|
|
|
|
&& wid_has_prop(ps, ev->window, ps->atom_client)) {
|
|
|
|
w_top->wmwin = false;
|
|
|
|
win_unmark_client(ps, w_top);
|
|
|
|
win_mark_client(ps, w_top, ev->window);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
// If _NET_WM_OPACITY changes
|
|
|
|
if (ev->atom == ps->atom_opacity) {
|
|
|
|
win *w = NULL;
|
|
|
|
if ((w = find_win(ps, ev->window)))
|
|
|
|
w->opacity_prop = wid_get_opacity_prop(ps, w->id, OPAQUE);
|
|
|
|
else if (ps->o.detect_client_opacity
|
|
|
|
&& (w = find_toplevel(ps, ev->window)))
|
|
|
|
w->opacity_prop_client = wid_get_opacity_prop(ps, w->client_win,
|
|
|
|
OPAQUE);
|
|
|
|
if (w) {
|
|
|
|
w->flags |= WFLAG_OPCT_CHANGE;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
// If frame extents property changes
|
|
|
|
if (ps->o.frame_opacity && ev->atom == ps->atom_frame_extents) {
|
|
|
|
win *w = find_toplevel(ps, ev->window);
|
|
|
|
if (w) {
|
|
|
|
get_frame_extents(ps, w, ev->window);
|
|
|
|
// If frame extents change, the window needs repaint
|
|
|
|
add_damage_win(ps, w);
|
|
|
|
}
|
|
|
|
}
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
|
|
|
|
// If name changes
|
|
|
|
if (ps->o.track_wdata
|
|
|
|
&& (ps->atom_name == ev->atom || ps->atom_name_ewmh == ev->atom)) {
|
|
|
|
win *w = find_toplevel(ps, ev->window);
|
|
|
|
if (w && 1 == win_get_name(ps, w)) {
|
|
|
|
determine_shadow(ps, w);
|
|
|
|
win_update_focused(ps, w);
|
|
|
|
}
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
}
|
|
|
|
|
|
|
|
// If class changes
|
|
|
|
if (ps->o.track_wdata && ps->atom_class == ev->atom) {
|
|
|
|
win *w = find_toplevel(ps, ev->window);
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
if (w) {
|
|
|
|
win_get_class(ps, w);
|
|
|
|
determine_shadow(ps, w);
|
|
|
|
win_update_focused(ps, w);
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
// If role changes
|
|
|
|
if (ps->o.track_wdata && ps->atom_role == ev->atom) {
|
|
|
|
win *w = find_toplevel(ps, ev->window);
|
|
|
|
if (w && 1 == win_get_role(ps, w)) {
|
|
|
|
determine_shadow(ps, w);
|
|
|
|
win_update_focused(ps, w);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
// If _COMPTON_SHADOW changes
|
|
|
|
if (ps->o.respect_prop_shadow && ps->atom_compton_shadow == ev->atom) {
|
|
|
|
win *w = find_win(ps, ev->window);
|
|
|
|
if (w)
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
win_update_prop_shadow(ps, w);
|
|
|
|
}
|
|
|
|
|
|
|
|
// If a leader property changes
|
|
|
|
if ((ps->o.detect_transient && ps->atom_transient == ev->atom)
|
|
|
|
|| (ps->o.detect_client_leader && ps->atom_client_leader == ev->atom)) {
|
|
|
|
win *w = find_toplevel(ps, ev->window);
|
|
|
|
if (w) {
|
|
|
|
win_update_leader(ps, w);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
inline static void
|
|
|
|
ev_damage_notify(session_t *ps, XDamageNotifyEvent *ev) {
|
|
|
|
damage_win(ps, ev);
|
|
|
|
}
|
|
|
|
|
|
|
|
inline static void
|
|
|
|
ev_shape_notify(session_t *ps, XShapeEvent *ev) {
|
|
|
|
win *w = find_win(ps, ev->window);
|
|
|
|
if (!w || IsUnmapped == w->a.map_state) return;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Empty border_size may indicated an
|
|
|
|
* unmapped/destroyed window, in which case
|
|
|
|
* seemingly BadRegion errors would be triggered
|
|
|
|
* if we attempt to rebuild border_size
|
|
|
|
*/
|
|
|
|
if (w->border_size) {
|
|
|
|
// Mark the old border_size as damaged
|
|
|
|
add_damage(ps, w->border_size);
|
|
|
|
|
|
|
|
w->border_size = border_size(ps, w, true);
|
|
|
|
|
|
|
|
// Mark the new border_size as damaged
|
|
|
|
add_damage(ps, copy_region(ps, w->border_size));
|
|
|
|
}
|
|
|
|
|
|
|
|
// Redo bounding shape detection and rounded corner detection
|
|
|
|
win_update_shape(ps, w);
|
|
|
|
|
|
|
|
update_reg_ignore_expire(ps, w);
|
|
|
|
}
|
|
|
|
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
/**
|
|
|
|
* Handle ScreenChangeNotify events from X RandR extension.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
ev_screen_change_notify(session_t *ps,
|
|
|
|
XRRScreenChangeNotifyEvent __attribute__((unused)) *ev) {
|
|
|
|
if (!ps->o.refresh_rate) {
|
|
|
|
update_refresh_rate(ps);
|
|
|
|
if (!ps->refresh_rate) {
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
fprintf(stderr, "ev_screen_change_notify(): Refresh rate detection "
|
|
|
|
"failed, software VSync disabled.");
|
|
|
|
ps->o.vsync = VSYNC_NONE;
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
#if defined(DEBUG_EVENTS) || defined(DEBUG_RESTACK)
|
|
|
|
/**
|
|
|
|
* Get a window's name from window ID.
|
|
|
|
*/
|
|
|
|
static bool
|
|
|
|
ev_window_name(session_t *ps, Window wid, char **name) {
|
|
|
|
bool to_free = false;
|
|
|
|
|
|
|
|
*name = "";
|
|
|
|
if (wid) {
|
|
|
|
*name = "(Failed to get title)";
|
|
|
|
if (ps->root == wid)
|
|
|
|
*name = "(Root window)";
|
|
|
|
else if (ps->overlay == wid)
|
|
|
|
*name = "(Overlay)";
|
|
|
|
else {
|
|
|
|
win *w = find_win(ps, wid);
|
|
|
|
if (!w)
|
|
|
|
w = find_toplevel(ps, wid);
|
|
|
|
|
|
|
|
if (w && w->name)
|
|
|
|
*name = w->name;
|
|
|
|
else if (!(w && w->client_win
|
|
|
|
&& (to_free = wid_get_name(ps, w->client_win, name))))
|
|
|
|
to_free = wid_get_name(ps, wid, name);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return to_free;
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
static void
|
|
|
|
ev_handle(session_t *ps, XEvent *ev) {
|
|
|
|
if ((ev->type & 0x7f) != KeymapNotify) {
|
|
|
|
discard_ignore(ps, ev->xany.serial);
|
|
|
|
}
|
|
|
|
|
|
|
|
#ifdef DEBUG_EVENTS
|
|
|
|
if (ev->type != ps->damage_event + XDamageNotify) {
|
|
|
|
Window wid = ev_window(ps, ev);
|
|
|
|
char *window_name = NULL;
|
|
|
|
bool to_free = false;
|
|
|
|
|
|
|
|
to_free = ev_window_name(ps, wid, &window_name);
|
|
|
|
|
|
|
|
print_timestamp(ps);
|
|
|
|
printf("event %10.10s serial %#010x window %#010lx \"%s\"\n",
|
|
|
|
ev_name(ps, ev), ev_serial(ev), wid, window_name);
|
|
|
|
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
if (to_free) {
|
|
|
|
XFree(window_name);
|
|
|
|
window_name = NULL;
|
|
|
|
}
|
|
|
|
}
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
switch (ev->type) {
|
|
|
|
case FocusIn:
|
|
|
|
ev_focus_in(ps, (XFocusChangeEvent *)ev);
|
|
|
|
break;
|
|
|
|
case FocusOut:
|
|
|
|
ev_focus_out(ps, (XFocusChangeEvent *)ev);
|
|
|
|
break;
|
|
|
|
case CreateNotify:
|
|
|
|
ev_create_notify(ps, (XCreateWindowEvent *)ev);
|
|
|
|
break;
|
|
|
|
case ConfigureNotify:
|
|
|
|
ev_configure_notify(ps, (XConfigureEvent *)ev);
|
|
|
|
break;
|
|
|
|
case DestroyNotify:
|
|
|
|
ev_destroy_notify(ps, (XDestroyWindowEvent *)ev);
|
|
|
|
break;
|
|
|
|
case MapNotify:
|
|
|
|
ev_map_notify(ps, (XMapEvent *)ev);
|
|
|
|
break;
|
|
|
|
case UnmapNotify:
|
|
|
|
ev_unmap_notify(ps, (XUnmapEvent *)ev);
|
|
|
|
break;
|
|
|
|
case ReparentNotify:
|
|
|
|
ev_reparent_notify(ps, (XReparentEvent *)ev);
|
|
|
|
break;
|
|
|
|
case CirculateNotify:
|
|
|
|
ev_circulate_notify(ps, (XCirculateEvent *)ev);
|
|
|
|
break;
|
|
|
|
case Expose:
|
|
|
|
ev_expose(ps, (XExposeEvent *)ev);
|
|
|
|
break;
|
|
|
|
case PropertyNotify:
|
|
|
|
ev_property_notify(ps, (XPropertyEvent *)ev);
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
if (ps->shape_exists && ev->type == ps->shape_event) {
|
|
|
|
ev_shape_notify(ps, (XShapeEvent *) ev);
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
if (ps->randr_exists && ev->type == (ps->randr_event + RRScreenChangeNotify)) {
|
|
|
|
ev_screen_change_notify(ps, (XRRScreenChangeNotifyEvent *) ev);
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
break;
|
|
|
|
}
|
|
|
|
if (ev->type == ps->damage_event + XDamageNotify) {
|
|
|
|
ev_damage_notify(ps, (XDamageNotifyEvent *)ev);
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
// === Main ===
|
|
|
|
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
/**
|
|
|
|
* Print usage text and exit.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
usage(void) {
|
|
|
|
const static char *usage_text =
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
"compton (" COMPTON_VERSION ")\n"
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
"usage: compton [options]\n"
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
"Options:\n"
|
|
|
|
"\n"
|
|
|
|
"-d display\n"
|
|
|
|
" Which display should be managed.\n"
|
|
|
|
"-r radius\n"
|
|
|
|
" The blur radius for shadows. (default 12)\n"
|
|
|
|
"-o opacity\n"
|
|
|
|
" The translucency for shadows. (default .75)\n"
|
|
|
|
"-l left-offset\n"
|
|
|
|
" The left offset for shadows. (default -15)\n"
|
|
|
|
"-t top-offset\n"
|
|
|
|
" The top offset for shadows. (default -15)\n"
|
|
|
|
"-I fade-in-step\n"
|
|
|
|
" Opacity change between steps while fading in. (default 0.028)\n"
|
|
|
|
"-O fade-out-step\n"
|
|
|
|
" Opacity change between steps while fading out. (default 0.03)\n"
|
|
|
|
"-D fade-delta-time\n"
|
|
|
|
" The time between steps in a fade in milliseconds. (default 10)\n"
|
|
|
|
"-m opacity\n"
|
|
|
|
" The opacity for menus. (default 1.0)\n"
|
|
|
|
"-c\n"
|
|
|
|
" Enabled client-side shadows on windows.\n"
|
|
|
|
"-C\n"
|
|
|
|
" Avoid drawing shadows on dock/panel windows.\n"
|
|
|
|
"-z\n"
|
|
|
|
" Zero the part of the shadow's mask behind the window (experimental).\n"
|
|
|
|
"-f\n"
|
|
|
|
" Fade windows in/out when opening/closing and when opacity\n"
|
|
|
|
" changes, unless --no-fading-openclose is used.\n"
|
|
|
|
"-F\n"
|
|
|
|
" Equals -f. Deprecated.\n"
|
|
|
|
"-i opacity\n"
|
|
|
|
" Opacity of inactive windows. (0.1 - 1.0)\n"
|
|
|
|
"-e opacity\n"
|
|
|
|
" Opacity of window titlebars and borders. (0.1 - 1.0)\n"
|
|
|
|
"-G\n"
|
|
|
|
" Don't draw shadows on DND windows\n"
|
|
|
|
"-b\n"
|
|
|
|
" Daemonize process.\n"
|
|
|
|
"-S\n"
|
|
|
|
" Enable synchronous operation (for debugging).\n"
|
|
|
|
"--config path\n"
|
|
|
|
" Look for configuration file at the path.\n"
|
|
|
|
"--shadow-red value\n"
|
|
|
|
" Red color value of shadow (0.0 - 1.0, defaults to 0).\n"
|
|
|
|
"--shadow-green value\n"
|
|
|
|
" Green color value of shadow (0.0 - 1.0, defaults to 0).\n"
|
|
|
|
"--shadow-blue value\n"
|
|
|
|
" Blue color value of shadow (0.0 - 1.0, defaults to 0).\n"
|
|
|
|
"--inactive-opacity-override\n"
|
|
|
|
" Inactive opacity set by -i overrides value of _NET_WM_OPACITY.\n"
|
|
|
|
"--inactive-dim value\n"
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
" Dim inactive windows. (0.0 - 1.0, defaults to 0)\n"
|
|
|
|
"--mark-wmwin-focused\n"
|
|
|
|
" Try to detect WM windows and mark them as active.\n"
|
|
|
|
"--shadow-exclude condition\n"
|
|
|
|
" Exclude conditions for shadows.\n"
|
|
|
|
"--mark-ovredir-focused\n"
|
|
|
|
" Mark windows that have no WM frame as active.\n"
|
|
|
|
"--no-fading-openclose\n"
|
|
|
|
" Do not fade on window open/close.\n"
|
|
|
|
"--shadow-ignore-shaped\n"
|
|
|
|
" Do not paint shadows on shaped windows.\n"
|
|
|
|
"--detect-rounded-corners\n"
|
|
|
|
" Try to detect windows with rounded corners and don't consider\n"
|
|
|
|
" them shaped windows.\n"
|
|
|
|
"--detect-client-opacity\n"
|
|
|
|
" Detect _NET_WM_OPACITY on client windows, useful for window\n"
|
|
|
|
" managers not passing _NET_WM_OPACITY of client windows to frame\n"
|
|
|
|
" windows.\n"
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
"--refresh-rate val\n"
|
|
|
|
" Specify refresh rate of the screen. If not specified or 0, compton\n"
|
|
|
|
" will try detecting this with X RandR extension.\n"
|
|
|
|
"--vsync vsync-method\n"
|
|
|
|
" Set VSync method. There are 2 VSync methods currently available:\n"
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
" none = No VSync\n"
|
|
|
|
" drm = VSync with DRM_IOCTL_WAIT_VBLANK. May only work on some\n"
|
|
|
|
" drivers. Experimental.\n"
|
|
|
|
" opengl = Try to VSync with SGI_swap_control OpenGL extension. Only\n"
|
|
|
|
" work on some drivers. Experimental.\n"
|
|
|
|
" (Note some VSync methods may not be enabled at compile time.)\n"
|
|
|
|
"--alpha-step val\n"
|
|
|
|
" Step for pregenerating alpha pictures. 0.01 - 1.0. Defaults to\n"
|
|
|
|
" 0.03.\n"
|
|
|
|
"--dbe\n"
|
|
|
|
" Enable DBE painting mode, intended to use with VSync to\n"
|
|
|
|
" (hopefully) eliminate tearing.\n"
|
|
|
|
"--paint-on-overlay\n"
|
|
|
|
" Painting on X Composite overlay window.\n"
|
|
|
|
"--sw-opti\n"
|
|
|
|
" Limit compton to repaint at most once every 1 / refresh_rate\n"
|
|
|
|
" second to boost performance. Experimental.\n"
|
|
|
|
"--vsync-aggressive\n"
|
|
|
|
" Attempt to send painting request before VBlank and do XFlush()\n"
|
|
|
|
" during VBlank. This switch may be lifted out at any moment.\n"
|
|
|
|
"--use-ewmh-active-win\n"
|
|
|
|
" Use _NET_WM_ACTIVE_WINDOW on the root window to determine which\n"
|
|
|
|
" window is focused instead of using FocusIn/Out events.\n"
|
|
|
|
"--respect-prop-shadow\n"
|
|
|
|
" Respect _COMPTON_SHADOW. This a prototype-level feature, which\n"
|
|
|
|
" you must not rely on.\n"
|
|
|
|
"--unredir-if-possible\n"
|
|
|
|
" Unredirect all windows if a full-screen opaque window is\n"
|
|
|
|
" detected, to maximize performance for full-screen windows.\n"
|
|
|
|
" Experimental.\n"
|
|
|
|
"--focus-exclude condition\n"
|
|
|
|
" Specify a list of conditions of windows that should always be\n"
|
|
|
|
" considered focused.\n"
|
|
|
|
"--inactive-dim-fixed\n"
|
|
|
|
" Use fixed inactive dim value.\n"
|
|
|
|
"--detect-transient\n"
|
|
|
|
" Use WM_TRANSIENT_FOR to group windows, and consider windows in\n"
|
|
|
|
" the same group focused at the same time.\n"
|
|
|
|
"--detect-client-leader\n"
|
|
|
|
" Use WM_CLIENT_LEADER to group windows, and consider windows in\n"
|
|
|
|
" the same group focused at the same time. WM_TRANSIENT_FOR has\n"
|
|
|
|
" higher priority if --detect-transient is enabled, too.\n"
|
|
|
|
"--blur-background\n"
|
|
|
|
" Blur background of semi-transparent / ARGB windows. Bad in\n"
|
|
|
|
" performance. The switch name may change without prior\n"
|
|
|
|
" notifications.\n"
|
|
|
|
"--blur-background-frame\n"
|
|
|
|
" Blur background of windows when the window frame is not opaque.\n"
|
|
|
|
" Implies --blur-background. Bad in performance. The switch name\n"
|
|
|
|
" may change.\n"
|
|
|
|
"--blur-background-fixed\n"
|
|
|
|
" Use fixed blur strength instead of adjusting according to window\n"
|
|
|
|
" opacity.\n"
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
"\n"
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
"Format of a condition:\n"
|
|
|
|
"\n"
|
|
|
|
" condition = <target>:<type>[<flags>]:<pattern>\n"
|
|
|
|
"\n"
|
|
|
|
" <target> is one of \"n\" (window name), \"i\" (window class\n"
|
|
|
|
" instance), \"g\" (window general class), and \"r\"\n"
|
|
|
|
" (window role).\n"
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
"\n"
|
|
|
|
" <type> is one of \"e\" (exact match), \"a\" (match anywhere),\n"
|
|
|
|
" \"s\" (match from start), \"w\" (wildcard), and \"p\" (PCRE\n"
|
|
|
|
" regular expressions, if compiled with the support).\n"
|
|
|
|
"\n"
|
|
|
|
" <flags> could be a series of flags. Currently the only defined\n"
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
" flag is \"i\" (ignore case).\n"
|
|
|
|
"\n"
|
|
|
|
" <pattern> is the actual pattern string.\n";
|
|
|
|
fputs(usage_text , stderr);
|
|
|
|
|
|
|
|
exit(1);
|
|
|
|
}
|
|
|
|
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
/**
|
|
|
|
* Register a window as symbol, and initialize GLX context if wanted.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
register_cm(session_t *ps, bool want_glxct) {
|
|
|
|
Atom a;
|
|
|
|
char *buf;
|
|
|
|
int len, s;
|
|
|
|
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
#ifdef CONFIG_VSYNC_OPENGL
|
|
|
|
// Create a window with the wanted GLX visual
|
|
|
|
if (want_glxct) {
|
|
|
|
XVisualInfo *pvi = NULL;
|
|
|
|
bool ret = false;
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
// Get visual for the window
|
|
|
|
int attribs[] = { GLX_RGBA, GLX_RED_SIZE, 1, GLX_GREEN_SIZE, 1, GLX_BLUE_SIZE, 1, None };
|
|
|
|
pvi = glXChooseVisual(ps->dpy, ps->scr, attribs);
|
|
|
|
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
if (!pvi) {
|
|
|
|
fprintf(stderr, "register_cm(): Failed to choose visual required "
|
|
|
|
"by fake OpenGL VSync window. OpenGL VSync turned off.\n");
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
// Create the window
|
|
|
|
XSetWindowAttributes swa = {
|
|
|
|
.colormap = XCreateColormap(ps->dpy, ps->root, pvi->visual, AllocNone),
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
.border_pixel = 0,
|
|
|
|
};
|
|
|
|
|
|
|
|
pvi->screen = ps->scr;
|
|
|
|
ps->reg_win = XCreateWindow(ps->dpy, ps->root, 0, 0, 1, 1, 0, pvi->depth,
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
InputOutput, pvi->visual, CWBorderPixel | CWColormap, &swa);
|
|
|
|
|
|
|
|
if (!ps->reg_win)
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
fprintf(stderr, "register_cm(): Failed to create window required "
|
|
|
|
"by fake OpenGL VSync. OpenGL VSync turned off.\n");
|
|
|
|
else {
|
|
|
|
// Get GLX context
|
|
|
|
ps->glx_context = glXCreateContext(ps->dpy, pvi, None, GL_TRUE);
|
|
|
|
if (!ps->glx_context) {
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
fprintf(stderr, "register_cm(): Failed to get GLX context. "
|
|
|
|
"OpenGL VSync turned off.\n");
|
|
|
|
ps->o.vsync = VSYNC_NONE;
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
}
|
|
|
|
else {
|
|
|
|
// Attach GLX context
|
|
|
|
if (!(ret = glXMakeCurrent(ps->dpy, ps->reg_win, ps->glx_context)))
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
fprintf(stderr, "register_cm(): Failed to attach GLX context."
|
|
|
|
" OpenGL VSync turned off.\n");
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
if (pvi)
|
|
|
|
XFree(pvi);
|
|
|
|
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
if (!ret)
|
|
|
|
ps->o.vsync = VSYNC_NONE;
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
|
|
|
if (!ps->reg_win)
|
|
|
|
ps->reg_win = XCreateSimpleWindow(ps->dpy, ps->root, 0, 0, 1, 1, 0,
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
None, None);
|
|
|
|
|
|
|
|
Xutf8SetWMProperties(ps->dpy, ps->reg_win, "xcompmgr", "xcompmgr",
|
|
|
|
NULL, 0, NULL, NULL, NULL);
|
|
|
|
|
|
|
|
len = strlen(REGISTER_PROP) + 2;
|
|
|
|
s = ps->scr;
|
|
|
|
|
|
|
|
while (s >= 10) {
|
|
|
|
++len;
|
|
|
|
s /= 10;
|
|
|
|
}
|
|
|
|
|
|
|
|
buf = malloc(len);
|
|
|
|
snprintf(buf, len, REGISTER_PROP"%d", ps->scr);
|
|
|
|
|
|
|
|
a = get_atom(ps, buf);
|
|
|
|
free(buf);
|
|
|
|
|
|
|
|
XSetSelectionOwner(ps->dpy, a, ps->reg_win, 0);
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Reopen streams for logging.
|
|
|
|
*/
|
|
|
|
static bool
|
|
|
|
ostream_reopen(session_t *ps, const char *path) {
|
|
|
|
if (!path)
|
|
|
|
path = ps->o.logpath;
|
|
|
|
if (!path)
|
|
|
|
path = "/dev/null";
|
|
|
|
|
|
|
|
bool success = freopen(path, "a", stdout);
|
|
|
|
success = freopen(path, "a", stderr) && success;
|
|
|
|
if (!success)
|
|
|
|
printf_errfq(1, "(%s): freopen() failed.", path);
|
|
|
|
|
|
|
|
return success;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Fork program to background and disable all I/O streams.
|
|
|
|
*/
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
static bool
|
|
|
|
fork_after(session_t *ps) {
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
if (getppid() == 1)
|
|
|
|
return true;
|
|
|
|
|
|
|
|
int pid = fork();
|
|
|
|
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
if (-1 == pid) {
|
|
|
|
printf_errf("(): fork() failed.");
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (pid > 0) _exit(0);
|
|
|
|
|
|
|
|
setsid();
|
|
|
|
|
|
|
|
// Mainly to suppress the _FORTIFY_SOURCE warning
|
|
|
|
bool success = freopen("/dev/null", "r", stdin);
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
if (!success) {
|
|
|
|
printf_errf("(): freopen() failed.");
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
success = ostream_reopen(ps, NULL);
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
|
|
|
|
return success;
|
|
|
|
}
|
|
|
|
|
|
|
|
#ifdef CONFIG_LIBCONFIG
|
|
|
|
/**
|
|
|
|
* Get a file stream of the configuration file to read.
|
|
|
|
*
|
|
|
|
* Follows the XDG specification to search for the configuration file.
|
|
|
|
*/
|
|
|
|
static FILE *
|
|
|
|
open_config_file(char *cpath, char **ppath) {
|
|
|
|
const static char *config_filename = "/compton.conf";
|
|
|
|
const static char *config_filename_legacy = "/.compton.conf";
|
|
|
|
const static char *config_home_suffix = "/.config";
|
|
|
|
const static char *config_system_dir = "/etc/xdg";
|
|
|
|
|
|
|
|
char *dir = NULL, *home = NULL;
|
|
|
|
char *path = cpath;
|
|
|
|
FILE *f = NULL;
|
|
|
|
|
|
|
|
if (path) {
|
|
|
|
f = fopen(path, "r");
|
|
|
|
if (f && ppath)
|
|
|
|
*ppath = path;
|
|
|
|
else
|
|
|
|
free(path);
|
|
|
|
return f;
|
|
|
|
}
|
|
|
|
|
|
|
|
// Check user configuration file in $XDG_CONFIG_HOME firstly
|
|
|
|
if (!((dir = getenv("XDG_CONFIG_HOME")) && strlen(dir))) {
|
|
|
|
if (!((home = getenv("HOME")) && strlen(home)))
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
path = mstrjoin3(home, config_home_suffix, config_filename);
|
|
|
|
}
|
|
|
|
else
|
|
|
|
path = mstrjoin(dir, config_filename);
|
|
|
|
|
|
|
|
f = fopen(path, "r");
|
|
|
|
|
|
|
|
if (f && ppath)
|
|
|
|
*ppath = path;
|
|
|
|
else
|
|
|
|
free(path);
|
|
|
|
if (f)
|
|
|
|
return f;
|
|
|
|
|
|
|
|
// Then check user configuration file in $HOME
|
|
|
|
if ((home = getenv("HOME")) && strlen(home)) {
|
|
|
|
path = mstrjoin(home, config_filename_legacy);
|
|
|
|
f = fopen(path, "r");
|
|
|
|
if (f && ppath)
|
|
|
|
*ppath = path;
|
|
|
|
else
|
|
|
|
free(path);
|
|
|
|
if (f)
|
|
|
|
return f;
|
|
|
|
}
|
|
|
|
|
|
|
|
// Check system configuration file in $XDG_CONFIG_DIRS at last
|
|
|
|
if ((dir = getenv("XDG_CONFIG_DIRS")) && strlen(dir)) {
|
|
|
|
char *part = strtok(dir, ":");
|
|
|
|
while (part) {
|
|
|
|
path = mstrjoin(part, config_filename);
|
|
|
|
f = fopen(path, "r");
|
|
|
|
if (f && ppath)
|
|
|
|
*ppath = path;
|
|
|
|
else
|
|
|
|
free(path);
|
|
|
|
if (f)
|
|
|
|
return f;
|
|
|
|
part = strtok(NULL, ":");
|
|
|
|
}
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
path = mstrjoin(config_system_dir, config_filename);
|
|
|
|
f = fopen(path, "r");
|
|
|
|
if (f && ppath)
|
|
|
|
*ppath = path;
|
|
|
|
else
|
|
|
|
free(path);
|
|
|
|
if (f)
|
|
|
|
return f;
|
|
|
|
}
|
|
|
|
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Parse a VSync option argument.
|
|
|
|
*/
|
|
|
|
static inline void
|
|
|
|
parse_vsync(session_t *ps, const char *optarg) {
|
|
|
|
const static char * const vsync_str[] = {
|
|
|
|
"none", // VSYNC_NONE
|
|
|
|
"drm", // VSYNC_DRM
|
|
|
|
"opengl", // VSYNC_OPENGL
|
|
|
|
};
|
|
|
|
|
|
|
|
vsync_t i;
|
|
|
|
|
|
|
|
for (i = 0; i < (sizeof(vsync_str) / sizeof(vsync_str[0])); ++i)
|
|
|
|
if (!strcasecmp(optarg, vsync_str[i])) {
|
|
|
|
ps->o.vsync = i;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
if ((sizeof(vsync_str) / sizeof(vsync_str[0])) == i) {
|
|
|
|
fputs("Invalid --vsync argument. Ignored.\n", stderr);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Parse a condition list in configuration file.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
parse_cfg_condlst(const config_t *pcfg, wincond_t **pcondlst,
|
|
|
|
const char *name) {
|
|
|
|
config_setting_t *setting = config_lookup(pcfg, name);
|
|
|
|
if (setting) {
|
|
|
|
// Parse an array of options
|
|
|
|
if (config_setting_is_array(setting)) {
|
|
|
|
int i = config_setting_length(setting);
|
|
|
|
while (i--) {
|
|
|
|
condlst_add(pcondlst, config_setting_get_string_elem(setting, i));
|
|
|
|
}
|
|
|
|
}
|
|
|
|
// Treat it as a single pattern if it's a string
|
|
|
|
else if (CONFIG_TYPE_STRING == config_setting_type(setting)) {
|
|
|
|
condlst_add(pcondlst, config_setting_get_string(setting));
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Parse a configuration file from default location.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
parse_config(session_t *ps, char *cpath, struct options_tmp *pcfgtmp) {
|
|
|
|
char *path = NULL;
|
|
|
|
FILE *f;
|
|
|
|
config_t cfg;
|
|
|
|
int ival = 0;
|
|
|
|
double dval = 0.0;
|
|
|
|
const char *sval = NULL;
|
|
|
|
|
|
|
|
f = open_config_file(cpath, &path);
|
|
|
|
if (!f) {
|
|
|
|
if (cpath)
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
printf_errfq(1, "(): Failed to read the specified configuration file.");
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
config_init(&cfg);
|
|
|
|
#ifndef CONFIG_LIBCONFIG_LEGACY
|
|
|
|
{
|
|
|
|
// dirname() could modify the original string, thus we must pass a
|
|
|
|
// copy
|
|
|
|
char *path2 = mstrcpy(path);
|
|
|
|
char *parent = dirname(path2);
|
|
|
|
|
|
|
|
if (parent)
|
|
|
|
config_set_include_dir(&cfg, parent);
|
|
|
|
|
|
|
|
free(path2);
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
|
|
|
if (CONFIG_FALSE == config_read(&cfg, f)) {
|
|
|
|
printf("Error when reading configuration file \"%s\", line %d: %s\n",
|
|
|
|
path, config_error_line(&cfg), config_error_text(&cfg));
|
|
|
|
config_destroy(&cfg);
|
|
|
|
free(path);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
config_set_auto_convert(&cfg, 1);
|
|
|
|
|
|
|
|
free(path);
|
|
|
|
|
|
|
|
// Get options from the configuration file. We don't do range checking
|
|
|
|
// right now. It will be done later
|
|
|
|
|
|
|
|
// -D (fade_delta)
|
|
|
|
if (lcfg_lookup_int(&cfg, "fade-delta", &ival))
|
|
|
|
ps->o.fade_delta = ival;
|
|
|
|
// -I (fade_in_step)
|
|
|
|
if (config_lookup_float(&cfg, "fade-in-step", &dval))
|
|
|
|
ps->o.fade_in_step = normalize_d(dval) * OPAQUE;
|
|
|
|
// -O (fade_out_step)
|
|
|
|
if (config_lookup_float(&cfg, "fade-out-step", &dval))
|
|
|
|
ps->o.fade_out_step = normalize_d(dval) * OPAQUE;
|
|
|
|
// -r (shadow_radius)
|
|
|
|
lcfg_lookup_int(&cfg, "shadow-radius", &ps->o.shadow_radius);
|
|
|
|
// -o (shadow_opacity)
|
|
|
|
config_lookup_float(&cfg, "shadow-opacity", &ps->o.shadow_opacity);
|
|
|
|
// -l (shadow_offset_x)
|
|
|
|
lcfg_lookup_int(&cfg, "shadow-offset-x", &ps->o.shadow_offset_x);
|
|
|
|
// -t (shadow_offset_y)
|
|
|
|
lcfg_lookup_int(&cfg, "shadow-offset-y", &ps->o.shadow_offset_y);
|
|
|
|
// -i (inactive_opacity)
|
|
|
|
if (config_lookup_float(&cfg, "inactive-opacity", &dval))
|
|
|
|
ps->o.inactive_opacity = normalize_d(dval) * OPAQUE;
|
|
|
|
// -e (frame_opacity)
|
|
|
|
config_lookup_float(&cfg, "frame-opacity", &ps->o.frame_opacity);
|
|
|
|
// -z (clear_shadow)
|
|
|
|
lcfg_lookup_bool(&cfg, "clear-shadow", &ps->o.clear_shadow);
|
|
|
|
// -c (shadow_enable)
|
|
|
|
if (config_lookup_bool(&cfg, "shadow", &ival) && ival)
|
|
|
|
wintype_arr_enable(ps->o.wintype_shadow);
|
|
|
|
// -C (no_dock_shadow)
|
|
|
|
lcfg_lookup_bool(&cfg, "no-dock-shadow", &pcfgtmp->no_dock_shadow);
|
|
|
|
// -G (no_dnd_shadow)
|
|
|
|
lcfg_lookup_bool(&cfg, "no-dnd-shadow", &pcfgtmp->no_dnd_shadow);
|
|
|
|
// -m (menu_opacity)
|
|
|
|
config_lookup_float(&cfg, "menu-opacity", &pcfgtmp->menu_opacity);
|
|
|
|
// -f (fading_enable)
|
|
|
|
if (config_lookup_bool(&cfg, "fading", &ival) && ival)
|
|
|
|
wintype_arr_enable(ps->o.wintype_fade);
|
|
|
|
// --no-fading-open-close
|
|
|
|
lcfg_lookup_bool(&cfg, "no-fading-openclose", &ps->o.no_fading_openclose);
|
|
|
|
// --shadow-red
|
|
|
|
config_lookup_float(&cfg, "shadow-red", &ps->o.shadow_red);
|
|
|
|
// --shadow-green
|
|
|
|
config_lookup_float(&cfg, "shadow-green", &ps->o.shadow_green);
|
|
|
|
// --shadow-blue
|
|
|
|
config_lookup_float(&cfg, "shadow-blue", &ps->o.shadow_blue);
|
|
|
|
// --inactive-opacity-override
|
|
|
|
lcfg_lookup_bool(&cfg, "inactive-opacity-override",
|
|
|
|
&ps->o.inactive_opacity_override);
|
|
|
|
// --inactive-dim
|
|
|
|
config_lookup_float(&cfg, "inactive-dim", &ps->o.inactive_dim);
|
|
|
|
// --mark-wmwin-focused
|
|
|
|
lcfg_lookup_bool(&cfg, "mark-wmwin-focused", &ps->o.mark_wmwin_focused);
|
|
|
|
// --mark-ovredir-focused
|
|
|
|
lcfg_lookup_bool(&cfg, "mark-ovredir-focused",
|
|
|
|
&ps->o.mark_ovredir_focused);
|
|
|
|
// --shadow-ignore-shaped
|
|
|
|
lcfg_lookup_bool(&cfg, "shadow-ignore-shaped",
|
|
|
|
&ps->o.shadow_ignore_shaped);
|
|
|
|
// --detect-rounded-corners
|
|
|
|
lcfg_lookup_bool(&cfg, "detect-rounded-corners",
|
|
|
|
&ps->o.detect_rounded_corners);
|
|
|
|
// --detect-client-opacity
|
|
|
|
lcfg_lookup_bool(&cfg, "detect-client-opacity",
|
|
|
|
&ps->o.detect_client_opacity);
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
// --refresh-rate
|
|
|
|
lcfg_lookup_int(&cfg, "refresh-rate", &ps->o.refresh_rate);
|
|
|
|
// --vsync
|
|
|
|
if (config_lookup_string(&cfg, "vsync", &sval))
|
|
|
|
parse_vsync(ps, sval);
|
|
|
|
// --alpha-step
|
|
|
|
config_lookup_float(&cfg, "alpha-step", &ps->o.alpha_step);
|
|
|
|
// --dbe
|
|
|
|
lcfg_lookup_bool(&cfg, "dbe", &ps->o.dbe);
|
|
|
|
// --paint-on-overlay
|
|
|
|
lcfg_lookup_bool(&cfg, "paint-on-overlay", &ps->o.paint_on_overlay);
|
|
|
|
// --sw-opti
|
|
|
|
lcfg_lookup_bool(&cfg, "sw-opti", &ps->o.sw_opti);
|
|
|
|
// --use-ewmh-active-win
|
|
|
|
lcfg_lookup_bool(&cfg, "use-ewmh-active-win",
|
|
|
|
&ps->o.use_ewmh_active_win);
|
|
|
|
// --unredir-if-possible
|
|
|
|
lcfg_lookup_bool(&cfg, "unredir-if-possible",
|
|
|
|
&ps->o.unredir_if_possible);
|
|
|
|
// --inactive-dim-fixed
|
|
|
|
lcfg_lookup_bool(&cfg, "inactive-dim-fixed", &ps->o.inactive_dim_fixed);
|
|
|
|
// --detect-transient
|
|
|
|
lcfg_lookup_bool(&cfg, "detect-transient", &ps->o.detect_transient);
|
|
|
|
// --detect-client-leader
|
|
|
|
lcfg_lookup_bool(&cfg, "detect-client-leader",
|
|
|
|
&ps->o.detect_client_leader);
|
|
|
|
// --shadow-exclude
|
|
|
|
parse_cfg_condlst(&cfg, &ps->o.shadow_blacklist, "shadow-exclude");
|
|
|
|
// --focus-exclude
|
|
|
|
parse_cfg_condlst(&cfg, &ps->o.focus_blacklist, "focus-exclude");
|
|
|
|
// --blur-background
|
|
|
|
lcfg_lookup_bool(&cfg, "blur-background", &ps->o.blur_background);
|
|
|
|
// --blur-background-frame
|
|
|
|
lcfg_lookup_bool(&cfg, "blur-background-frame",
|
|
|
|
&ps->o.blur_background_frame);
|
|
|
|
// --blur-background-fixed
|
|
|
|
lcfg_lookup_bool(&cfg, "blur-background-fixed",
|
|
|
|
&ps->o.blur_background_fixed);
|
|
|
|
// Wintype settings
|
|
|
|
{
|
|
|
|
wintype_t i;
|
|
|
|
|
|
|
|
for (i = 0; i < NUM_WINTYPES; ++i) {
|
|
|
|
char *str = mstrjoin("wintypes.", WINTYPES[i]);
|
|
|
|
config_setting_t *setting = config_lookup(&cfg, str);
|
|
|
|
free(str);
|
|
|
|
if (setting) {
|
|
|
|
if (config_setting_lookup_bool(setting, "shadow", &ival))
|
|
|
|
ps->o.wintype_shadow[i] = (bool) ival;
|
|
|
|
if (config_setting_lookup_bool(setting, "fade", &ival))
|
|
|
|
ps->o.wintype_fade[i] = (bool) ival;
|
|
|
|
if (config_setting_lookup_bool(setting, "focus", &ival))
|
|
|
|
ps->o.wintype_focus[i] = (bool) ival;
|
|
|
|
config_setting_lookup_float(setting, "opacity",
|
|
|
|
&ps->o.wintype_opacity[i]);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
config_destroy(&cfg);
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Process arguments and configuration files.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
get_cfg(session_t *ps, int argc, char *const *argv) {
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
const static char *shortopts = "D:I:O:d:r:o:m:l:t:i:e:hscnfFCaSzGb";
|
|
|
|
const static struct option longopts[] = {
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
{ "help", no_argument, NULL, 'h' },
|
|
|
|
{ "config", required_argument, NULL, 256 },
|
|
|
|
{ "shadow-red", required_argument, NULL, 257 },
|
|
|
|
{ "shadow-green", required_argument, NULL, 258 },
|
|
|
|
{ "shadow-blue", required_argument, NULL, 259 },
|
|
|
|
{ "inactive-opacity-override", no_argument, NULL, 260 },
|
|
|
|
{ "inactive-dim", required_argument, NULL, 261 },
|
|
|
|
{ "mark-wmwin-focused", no_argument, NULL, 262 },
|
|
|
|
{ "shadow-exclude", required_argument, NULL, 263 },
|
|
|
|
{ "mark-ovredir-focused", no_argument, NULL, 264 },
|
|
|
|
{ "no-fading-openclose", no_argument, NULL, 265 },
|
|
|
|
{ "shadow-ignore-shaped", no_argument, NULL, 266 },
|
|
|
|
{ "detect-rounded-corners", no_argument, NULL, 267 },
|
|
|
|
{ "detect-client-opacity", no_argument, NULL, 268 },
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
{ "refresh-rate", required_argument, NULL, 269 },
|
|
|
|
{ "vsync", required_argument, NULL, 270 },
|
|
|
|
{ "alpha-step", required_argument, NULL, 271 },
|
|
|
|
{ "dbe", no_argument, NULL, 272 },
|
|
|
|
{ "paint-on-overlay", no_argument, NULL, 273 },
|
|
|
|
{ "sw-opti", no_argument, NULL, 274 },
|
|
|
|
{ "vsync-aggressive", no_argument, NULL, 275 },
|
|
|
|
{ "use-ewmh-active-win", no_argument, NULL, 276 },
|
|
|
|
{ "respect-prop-shadow", no_argument, NULL, 277 },
|
|
|
|
{ "unredir-if-possible", no_argument, NULL, 278 },
|
|
|
|
{ "focus-exclude", required_argument, NULL, 279 },
|
|
|
|
{ "inactive-dim-fixed", no_argument, NULL, 280 },
|
|
|
|
{ "detect-transient", no_argument, NULL, 281 },
|
|
|
|
{ "detect-client-leader", no_argument, NULL, 282 },
|
|
|
|
{ "blur-background", no_argument, NULL, 283 },
|
|
|
|
{ "blur-background-frame", no_argument, NULL, 284 },
|
|
|
|
{ "blur-background-fixed", no_argument, NULL, 285 },
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
{ "dbus", no_argument, NULL, 286 },
|
|
|
|
{ "logpath", required_argument, NULL, 287 },
|
|
|
|
// Must terminate with a NULL entry
|
|
|
|
{ NULL, 0, NULL, 0 },
|
|
|
|
};
|
|
|
|
|
|
|
|
struct options_tmp cfgtmp = {
|
|
|
|
.no_dock_shadow = false,
|
|
|
|
.no_dnd_shadow = false,
|
|
|
|
.menu_opacity = 1.0,
|
|
|
|
};
|
|
|
|
bool shadow_enable = false, fading_enable = false;
|
|
|
|
int o, longopt_idx, i;
|
|
|
|
char *config_file = NULL;
|
|
|
|
char *lc_numeric_old = mstrcpy(setlocale(LC_NUMERIC, NULL));
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
|
|
|
|
for (i = 0; i < NUM_WINTYPES; ++i) {
|
|
|
|
ps->o.wintype_fade[i] = false;
|
|
|
|
ps->o.wintype_shadow[i] = false;
|
|
|
|
ps->o.wintype_opacity[i] = 1.0;
|
|
|
|
}
|
|
|
|
|
|
|
|
// Pre-parse the commandline arguments to check for --config and invalid
|
|
|
|
// switches
|
|
|
|
// Must reset optind to 0 here in case we reread the commandline
|
|
|
|
// arguments
|
|
|
|
optind = 1;
|
|
|
|
while (-1 !=
|
|
|
|
(o = getopt_long(argc, argv, shortopts, longopts, &longopt_idx))) {
|
|
|
|
if (256 == o)
|
|
|
|
config_file = mstrcpy(optarg);
|
|
|
|
else if ('?' == o || ':' == o)
|
|
|
|
usage();
|
|
|
|
}
|
|
|
|
|
|
|
|
#ifdef CONFIG_LIBCONFIG
|
|
|
|
parse_config(ps, config_file, &cfgtmp);
|
|
|
|
#endif
|
|
|
|
|
|
|
|
// Parse commandline arguments. Range checking will be done later.
|
|
|
|
|
|
|
|
// Enforce LC_NUMERIC locale "C" here to make sure dots are recognized
|
|
|
|
// instead of commas in atof().
|
|
|
|
setlocale(LC_NUMERIC, "C");
|
|
|
|
|
|
|
|
optind = 1;
|
|
|
|
while (-1 !=
|
|
|
|
(o = getopt_long(argc, argv, shortopts, longopts, &longopt_idx))) {
|
|
|
|
switch (o) {
|
|
|
|
// Short options
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
case 'h':
|
|
|
|
usage();
|
|
|
|
break;
|
|
|
|
case 'd':
|
|
|
|
ps->o.display = mstrcpy(optarg);
|
|
|
|
break;
|
|
|
|
case 'D':
|
|
|
|
ps->o.fade_delta = atoi(optarg);
|
|
|
|
break;
|
|
|
|
case 'I':
|
|
|
|
ps->o.fade_in_step = normalize_d(atof(optarg)) * OPAQUE;
|
|
|
|
break;
|
|
|
|
case 'O':
|
|
|
|
ps->o.fade_out_step = normalize_d(atof(optarg)) * OPAQUE;
|
|
|
|
break;
|
|
|
|
case 'c':
|
|
|
|
shadow_enable = true;
|
|
|
|
break;
|
|
|
|
case 'C':
|
|
|
|
cfgtmp.no_dock_shadow = true;
|
|
|
|
break;
|
|
|
|
case 'G':
|
|
|
|
cfgtmp.no_dnd_shadow = true;
|
|
|
|
break;
|
|
|
|
case 'm':
|
|
|
|
cfgtmp.menu_opacity = atof(optarg);
|
|
|
|
break;
|
|
|
|
case 'f':
|
|
|
|
case 'F':
|
|
|
|
fading_enable = true;
|
|
|
|
break;
|
|
|
|
case 'S':
|
|
|
|
ps->o.synchronize = true;
|
|
|
|
break;
|
|
|
|
case 'r':
|
|
|
|
ps->o.shadow_radius = atoi(optarg);
|
|
|
|
break;
|
|
|
|
case 'o':
|
|
|
|
ps->o.shadow_opacity = atof(optarg);
|
|
|
|
break;
|
|
|
|
case 'l':
|
|
|
|
ps->o.shadow_offset_x = atoi(optarg);
|
|
|
|
break;
|
|
|
|
case 't':
|
|
|
|
ps->o.shadow_offset_y = atoi(optarg);
|
|
|
|
break;
|
|
|
|
case 'i':
|
|
|
|
ps->o.inactive_opacity = (normalize_d(atof(optarg)) * OPAQUE);
|
|
|
|
break;
|
|
|
|
case 'e':
|
|
|
|
ps->o.frame_opacity = atof(optarg);
|
|
|
|
break;
|
|
|
|
case 'z':
|
|
|
|
ps->o.clear_shadow = true;
|
|
|
|
break;
|
|
|
|
case 'n':
|
|
|
|
case 'a':
|
|
|
|
case 's':
|
|
|
|
printf_errfq(1, "(): -n, -a, and -s have been removed.");
|
|
|
|
break;
|
|
|
|
case 'b':
|
|
|
|
ps->o.fork_after_register = true;
|
|
|
|
break;
|
|
|
|
// Long options
|
|
|
|
case 256:
|
|
|
|
// --config
|
|
|
|
break;
|
|
|
|
case 257:
|
|
|
|
// --shadow-red
|
|
|
|
ps->o.shadow_red = atof(optarg);
|
|
|
|
break;
|
|
|
|
case 258:
|
|
|
|
// --shadow-green
|
|
|
|
ps->o.shadow_green = atof(optarg);
|
|
|
|
break;
|
|
|
|
case 259:
|
|
|
|
// --shadow-blue
|
|
|
|
ps->o.shadow_blue = atof(optarg);
|
|
|
|
break;
|
|
|
|
case 260:
|
|
|
|
// --inactive-opacity-override
|
|
|
|
ps->o.inactive_opacity_override = true;
|
|
|
|
break;
|
|
|
|
case 261:
|
|
|
|
// --inactive-dim
|
|
|
|
ps->o.inactive_dim = atof(optarg);
|
|
|
|
break;
|
|
|
|
case 262:
|
|
|
|
// --mark-wmwin-focused
|
|
|
|
ps->o.mark_wmwin_focused = true;
|
|
|
|
break;
|
|
|
|
case 263:
|
|
|
|
// --shadow-exclude
|
|
|
|
condlst_add(&ps->o.shadow_blacklist, optarg);
|
|
|
|
break;
|
|
|
|
case 264:
|
|
|
|
// --mark-ovredir-focused
|
|
|
|
ps->o.mark_ovredir_focused = true;
|
|
|
|
break;
|
|
|
|
case 265:
|
|
|
|
// --no-fading-openclose
|
|
|
|
ps->o.no_fading_openclose = true;
|
|
|
|
break;
|
|
|
|
case 266:
|
|
|
|
// --shadow-ignore-shaped
|
|
|
|
ps->o.shadow_ignore_shaped = true;
|
|
|
|
break;
|
|
|
|
case 267:
|
|
|
|
// --detect-rounded-corners
|
|
|
|
ps->o.detect_rounded_corners = true;
|
|
|
|
break;
|
|
|
|
case 268:
|
|
|
|
// --detect-client-opacity
|
|
|
|
ps->o.detect_client_opacity = true;
|
|
|
|
break;
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
case 269:
|
|
|
|
// --refresh-rate
|
|
|
|
ps->o.refresh_rate = atoi(optarg);
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
break;
|
|
|
|
case 270:
|
|
|
|
// --vsync
|
|
|
|
parse_vsync(ps, optarg);
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
break;
|
|
|
|
case 271:
|
|
|
|
// --alpha-step
|
|
|
|
ps->o.alpha_step = atof(optarg);
|
|
|
|
break;
|
|
|
|
case 272:
|
|
|
|
// --dbe
|
|
|
|
ps->o.dbe = true;
|
|
|
|
break;
|
|
|
|
case 273:
|
|
|
|
// --paint-on-overlay
|
|
|
|
ps->o.paint_on_overlay = true;
|
|
|
|
break;
|
|
|
|
case 274:
|
|
|
|
// --sw-opti
|
|
|
|
ps->o.sw_opti = true;
|
|
|
|
break;
|
|
|
|
case 275:
|
|
|
|
// --vsync-aggressive
|
|
|
|
ps->o.vsync_aggressive = true;
|
|
|
|
break;
|
|
|
|
case 276:
|
|
|
|
// --use-ewmh-active-win
|
|
|
|
ps->o.use_ewmh_active_win = true;
|
|
|
|
break;
|
|
|
|
case 277:
|
|
|
|
// --respect-prop-shadow
|
|
|
|
ps->o.respect_prop_shadow = true;
|
|
|
|
break;
|
|
|
|
case 278:
|
|
|
|
// --unredir-if-possible
|
|
|
|
ps->o.unredir_if_possible = true;
|
|
|
|
break;
|
|
|
|
case 279:
|
|
|
|
// --focus-exclude
|
|
|
|
condlst_add(&ps->o.focus_blacklist, optarg);
|
|
|
|
break;
|
|
|
|
case 280:
|
|
|
|
// --inactive-dim-fixed
|
|
|
|
ps->o.inactive_dim_fixed = true;
|
|
|
|
break;
|
|
|
|
case 281:
|
|
|
|
// --detect-transient
|
|
|
|
ps->o.detect_transient = true;
|
|
|
|
break;
|
|
|
|
case 282:
|
|
|
|
// --detect-client-leader
|
|
|
|
ps->o.detect_client_leader = true;
|
|
|
|
break;
|
|
|
|
case 283:
|
|
|
|
// --blur-background
|
|
|
|
ps->o.blur_background = true;
|
|
|
|
break;
|
|
|
|
case 284:
|
|
|
|
// --blur-background-frame
|
|
|
|
ps->o.blur_background_frame = true;
|
|
|
|
break;
|
|
|
|
case 285:
|
|
|
|
// --blur-background-fixed
|
|
|
|
ps->o.blur_background_fixed = true;
|
|
|
|
break;
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
case 286:
|
|
|
|
// --dbus
|
|
|
|
ps->o.dbus = true;
|
|
|
|
break;
|
|
|
|
case 287:
|
|
|
|
// --logpath
|
|
|
|
ps->o.logpath = mstrcpy(optarg);
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
usage();
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
// Restore LC_NUMERIC
|
|
|
|
setlocale(LC_NUMERIC, lc_numeric_old);
|
|
|
|
free(lc_numeric_old);
|
|
|
|
|
|
|
|
// Range checking and option assignments
|
|
|
|
ps->o.fade_delta = max_i(ps->o.fade_delta, 1);
|
|
|
|
ps->o.shadow_radius = max_i(ps->o.shadow_radius, 1);
|
|
|
|
ps->o.shadow_red = normalize_d(ps->o.shadow_red);
|
|
|
|
ps->o.shadow_green = normalize_d(ps->o.shadow_green);
|
|
|
|
ps->o.shadow_blue = normalize_d(ps->o.shadow_blue);
|
|
|
|
ps->o.inactive_dim = normalize_d(ps->o.inactive_dim);
|
|
|
|
ps->o.frame_opacity = normalize_d(ps->o.frame_opacity);
|
|
|
|
ps->o.shadow_opacity = normalize_d(ps->o.shadow_opacity);
|
|
|
|
cfgtmp.menu_opacity = normalize_d(cfgtmp.menu_opacity);
|
|
|
|
ps->o.refresh_rate = normalize_i_range(ps->o.refresh_rate, 0, 300);
|
|
|
|
ps->o.alpha_step = normalize_d_range(ps->o.alpha_step, 0.01, 1.0);
|
|
|
|
if (OPAQUE == ps->o.inactive_opacity) {
|
|
|
|
ps->o.inactive_opacity = 0;
|
|
|
|
}
|
|
|
|
if (shadow_enable)
|
|
|
|
wintype_arr_enable(ps->o.wintype_shadow);
|
|
|
|
ps->o.wintype_shadow[WINTYPE_DESKTOP] = false;
|
|
|
|
if (cfgtmp.no_dock_shadow)
|
|
|
|
ps->o.wintype_shadow[WINTYPE_DOCK] = false;
|
|
|
|
if (cfgtmp.no_dnd_shadow)
|
|
|
|
ps->o.wintype_shadow[WINTYPE_DND] = false;
|
|
|
|
if (fading_enable)
|
|
|
|
wintype_arr_enable(ps->o.wintype_fade);
|
|
|
|
if (1.0 != cfgtmp.menu_opacity) {
|
|
|
|
ps->o.wintype_opacity[WINTYPE_DROPDOWN_MENU] = cfgtmp.menu_opacity;
|
|
|
|
ps->o.wintype_opacity[WINTYPE_POPUP_MENU] = cfgtmp.menu_opacity;
|
|
|
|
}
|
|
|
|
|
|
|
|
// --blur-background-frame implies --blur-background
|
|
|
|
if (ps->o.blur_background_frame)
|
|
|
|
ps->o.blur_background = true;
|
|
|
|
|
|
|
|
// Other variables determined by options
|
|
|
|
|
|
|
|
// Determine whether we need to track focus changes
|
|
|
|
if (ps->o.inactive_opacity || ps->o.inactive_dim) {
|
|
|
|
ps->o.track_focus = true;
|
|
|
|
}
|
|
|
|
|
Feature: Issue #29: Alternative shadow blacklist implementation
- Add shadow blacklist feature, but a different implementation from
nicklan's. 5 matching modes (exact, starts-with, contains, wildcard,
PCRE) and 3 matching targets (window name, window class instance,
window general class). Not extensively tested, bugs to be expected.
It's slower for exact matching than nicklan's as it uses linear search
instead of hash table. Also, PCRE's JIT optimization may cause issues
on PaX kernels.
- Add dependency to libpcre. Could be made optional if we have a
graceful way to handle that in Makefile.
- Some matching functions are GNU extensions of glibc. So this version
may have troubles running on platforms not using glibc.
- Fix a bug that access freed memory blocks in set_fade_callcack() and
check_fade_fin(). valgrind found it out.
- Use WM_CLASS to detect client windows instead of WM_STATE. Some client
windows (like notification windows) have WM_CLASS but not WM_STATE.
- Mark the extents as damaged if shadow state changed in
determine_shadow().
- Rewrite wid_get_name(). Code clean-up.
- Two debugging options: DEBUG_WINDATA and DEBUG_WINMATCH.
- As the matching system is ready, it should be rather easy to add other
kinds of blacklists, like fading blacklist.
12 years ago
|
|
|
// Determine whether we need to track window name and class
|
|
|
|
if (ps->o.shadow_blacklist || ps->o.fade_blacklist
|
|
|
|
|| ps->o.focus_blacklist)
|
|
|
|
ps->o.track_wdata = true;
|
|
|
|
|
|
|
|
// Determine whether we track window grouping
|
|
|
|
if (ps->o.detect_transient || ps->o.detect_client_leader) {
|
|
|
|
ps->o.track_leader = true;
|
|
|
|
}
|
|
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Fetch all required atoms and save them to a session.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
init_atoms(session_t *ps) {
|
|
|
|
ps->atom_opacity = get_atom(ps, "_NET_WM_WINDOW_OPACITY");
|
|
|
|
ps->atom_frame_extents = get_atom(ps, "_NET_FRAME_EXTENTS");
|
|
|
|
ps->atom_client = get_atom(ps, "WM_STATE");
|
|
|
|
ps->atom_name = XA_WM_NAME;
|
|
|
|
ps->atom_name_ewmh = get_atom(ps, "_NET_WM_NAME");
|
|
|
|
ps->atom_class = XA_WM_CLASS;
|
|
|
|
ps->atom_role = get_atom(ps, "WM_WINDOW_ROLE");
|
|
|
|
ps->atom_transient = XA_WM_TRANSIENT_FOR;
|
|
|
|
ps->atom_client_leader = get_atom(ps, "WM_CLIENT_LEADER");
|
|
|
|
ps->atom_ewmh_active_win = get_atom(ps, "_NET_ACTIVE_WINDOW");
|
|
|
|
ps->atom_compton_shadow = get_atom(ps, "_COMPTON_SHADOW");
|
|
|
|
|
|
|
|
ps->atom_win_type = get_atom(ps, "_NET_WM_WINDOW_TYPE");
|
|
|
|
ps->atoms_wintypes[WINTYPE_UNKNOWN] = 0;
|
|
|
|
ps->atoms_wintypes[WINTYPE_DESKTOP] = get_atom(ps,
|
|
|
|
"_NET_WM_WINDOW_TYPE_DESKTOP");
|
|
|
|
ps->atoms_wintypes[WINTYPE_DOCK] = get_atom(ps,
|
|
|
|
"_NET_WM_WINDOW_TYPE_DOCK");
|
|
|
|
ps->atoms_wintypes[WINTYPE_TOOLBAR] = get_atom(ps,
|
|
|
|
"_NET_WM_WINDOW_TYPE_TOOLBAR");
|
|
|
|
ps->atoms_wintypes[WINTYPE_MENU] = get_atom(ps,
|
|
|
|
"_NET_WM_WINDOW_TYPE_MENU");
|
|
|
|
ps->atoms_wintypes[WINTYPE_UTILITY] = get_atom(ps,
|
|
|
|
"_NET_WM_WINDOW_TYPE_UTILITY");
|
|
|
|
ps->atoms_wintypes[WINTYPE_SPLASH] = get_atom(ps,
|
|
|
|
"_NET_WM_WINDOW_TYPE_SPLASH");
|
|
|
|
ps->atoms_wintypes[WINTYPE_DIALOG] = get_atom(ps,
|
|
|
|
"_NET_WM_WINDOW_TYPE_DIALOG");
|
|
|
|
ps->atoms_wintypes[WINTYPE_NORMAL] = get_atom(ps,
|
|
|
|
"_NET_WM_WINDOW_TYPE_NORMAL");
|
|
|
|
ps->atoms_wintypes[WINTYPE_DROPDOWN_MENU] = get_atom(ps,
|
|
|
|
"_NET_WM_WINDOW_TYPE_DROPDOWN_MENU");
|
|
|
|
ps->atoms_wintypes[WINTYPE_POPUP_MENU] = get_atom(ps,
|
|
|
|
"_NET_WM_WINDOW_TYPE_POPUP_MENU");
|
|
|
|
ps->atoms_wintypes[WINTYPE_TOOLTIP] = get_atom(ps,
|
|
|
|
"_NET_WM_WINDOW_TYPE_TOOLTIP");
|
|
|
|
ps->atoms_wintypes[WINTYPE_NOTIFY] = get_atom(ps,
|
|
|
|
"_NET_WM_WINDOW_TYPE_NOTIFICATION");
|
|
|
|
ps->atoms_wintypes[WINTYPE_COMBO] = get_atom(ps,
|
|
|
|
"_NET_WM_WINDOW_TYPE_COMBO");
|
|
|
|
ps->atoms_wintypes[WINTYPE_DND] = get_atom(ps,
|
|
|
|
"_NET_WM_WINDOW_TYPE_DND");
|
|
|
|
}
|
|
|
|
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
/**
|
|
|
|
* Update refresh rate info with X Randr extension.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
update_refresh_rate(session_t *ps) {
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
XRRScreenConfiguration* randr_info;
|
|
|
|
|
|
|
|
if (!(randr_info = XRRGetScreenInfo(ps->dpy, ps->root)))
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
return;
|
|
|
|
ps->refresh_rate = XRRConfigCurrentRate(randr_info);
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
|
|
|
|
XRRFreeScreenConfigInfo(randr_info);
|
|
|
|
|
|
|
|
if (ps->refresh_rate)
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
ps->refresh_intv = US_PER_SEC / ps->refresh_rate;
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
else
|
|
|
|
ps->refresh_intv = 0;
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Initialize refresh-rated based software optimization.
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
*
|
|
|
|
* @return true for success, false otherwise
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
*/
|
|
|
|
static bool
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
swopti_init(session_t *ps) {
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
// Prepare refresh rate
|
|
|
|
// Check if user provides one
|
|
|
|
ps->refresh_rate = ps->o.refresh_rate;
|
|
|
|
if (ps->refresh_rate)
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
ps->refresh_intv = US_PER_SEC / ps->refresh_rate;
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
|
|
|
|
// Auto-detect refresh rate otherwise
|
|
|
|
if (!ps->refresh_rate && ps->randr_exists) {
|
|
|
|
update_refresh_rate(ps);
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
}
|
|
|
|
|
|
|
|
// Turn off vsync_sw if we can't get the refresh rate
|
|
|
|
if (!ps->refresh_rate)
|
|
|
|
return false;
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
|
|
|
|
// Monitor screen changes only if vsync_sw is enabled and we are using
|
|
|
|
// an auto-detected refresh rate
|
|
|
|
if (ps->randr_exists && !ps->o.refresh_rate)
|
|
|
|
XRRSelectInput(ps->dpy, ps->root, RRScreenChangeNotify);
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
|
|
|
|
return true;
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
* Modify a struct timeval timeout value to render at a fixed pace.
|
|
|
|
*
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
* @param ps current session
|
|
|
|
* @param[in,out] ptv pointer to the timeout
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
*/
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
static void
|
|
|
|
swopti_handle_timeout(session_t *ps, struct timeval *ptv) {
|
|
|
|
if (!ptv)
|
|
|
|
return;
|
|
|
|
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
// Get the microsecond offset of the time when the we reach the timeout
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
// I don't think a 32-bit long could overflow here.
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
long offset = (ptv->tv_usec + get_time_timeval().tv_usec - ps->paint_tm_offset) % ps->refresh_intv;
|
|
|
|
if (offset < 0)
|
|
|
|
offset += ps->refresh_intv;
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
assert(offset >= 0 && offset < ps->refresh_intv);
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
|
|
|
|
// If the target time is sufficiently close to a refresh time, don't add
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
// an offset, to avoid certain blocking conditions.
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
if (offset < SWOPTI_TOLERANCE
|
|
|
|
|| offset > ps->refresh_intv - SWOPTI_TOLERANCE)
|
|
|
|
return;
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
|
|
|
|
// Add an offset so we wait until the next refresh after timeout
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
ptv->tv_usec += ps->refresh_intv - offset;
|
|
|
|
if (ptv->tv_usec > US_PER_SEC) {
|
|
|
|
ptv->tv_usec -= US_PER_SEC;
|
|
|
|
++ptv->tv_sec;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
/**
|
|
|
|
* Initialize DRM VSync.
|
|
|
|
*
|
|
|
|
* @return true for success, false otherwise
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
*/
|
|
|
|
static bool
|
|
|
|
vsync_drm_init(session_t *ps) {
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
#ifdef CONFIG_VSYNC_DRM
|
|
|
|
// Should we always open card0?
|
|
|
|
if ((ps->drm_fd = open("/dev/dri/card0", O_RDWR)) < 0) {
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
fprintf(stderr, "vsync_drm_init(): Failed to open device.\n");
|
|
|
|
return false;
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
}
|
|
|
|
|
|
|
|
if (vsync_drm_wait(ps))
|
|
|
|
return false;
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
|
|
|
|
return true;
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
#else
|
|
|
|
fprintf(stderr, "Program not compiled with DRM VSync support.\n");
|
|
|
|
return false;
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
#endif
|
|
|
|
}
|
|
|
|
|
|
|
|
#ifdef CONFIG_VSYNC_DRM
|
|
|
|
/**
|
|
|
|
* Wait for next VSync, DRM method.
|
|
|
|
*
|
|
|
|
* Stolen from: https://github.com/MythTV/mythtv/blob/master/mythtv/libs/libmythtv/vsync.cpp
|
|
|
|
*/
|
|
|
|
static int
|
|
|
|
vsync_drm_wait(session_t *ps) {
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
int ret = -1;
|
|
|
|
drm_wait_vblank_t vbl;
|
|
|
|
|
|
|
|
vbl.request.type = _DRM_VBLANK_RELATIVE,
|
|
|
|
vbl.request.sequence = 1;
|
|
|
|
|
|
|
|
do {
|
|
|
|
ret = ioctl(ps->drm_fd, DRM_IOCTL_WAIT_VBLANK, &vbl);
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
vbl.request.type &= ~_DRM_VBLANK_RELATIVE;
|
|
|
|
} while (ret && errno == EINTR);
|
|
|
|
|
|
|
|
if (ret)
|
|
|
|
fprintf(stderr, "vsync_drm_wait(): VBlank ioctl did not work, "
|
|
|
|
"unimplemented in this drmver?\n");
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Initialize OpenGL VSync.
|
|
|
|
*
|
|
|
|
* Stolen from: http://git.tuxfamily.org/?p=ccm/cairocompmgr.git;a=commitdiff;h=efa4ceb97da501e8630ca7f12c99b1dce853c73e
|
|
|
|
* Possible original source: http://www.inb.uni-luebeck.de/~boehme/xvideo_sync.html
|
|
|
|
*
|
|
|
|
* @return true for success, false otherwise
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
*/
|
|
|
|
static bool
|
|
|
|
vsync_opengl_init(session_t *ps) {
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
#ifdef CONFIG_VSYNC_OPENGL
|
|
|
|
// Get video sync functions
|
|
|
|
ps->glx_get_video_sync = (f_GetVideoSync)
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
glXGetProcAddress ((const GLubyte *) "glXGetVideoSyncSGI");
|
|
|
|
ps->glx_wait_video_sync = (f_WaitVideoSync)
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
glXGetProcAddress ((const GLubyte *) "glXWaitVideoSyncSGI");
|
|
|
|
if (!ps->glx_wait_video_sync || !ps->glx_get_video_sync) {
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
fprintf(stderr, "vsync_opengl_init(): "
|
|
|
|
"Failed to get glXWait/GetVideoSyncSGI function.\n");
|
|
|
|
return false;
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
}
|
|
|
|
|
|
|
|
return true;
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
#else
|
|
|
|
fprintf(stderr, "Program not compiled with OpenGL VSync support.\n");
|
|
|
|
return false;
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
#endif
|
|
|
|
}
|
|
|
|
|
|
|
|
#ifdef CONFIG_VSYNC_OPENGL
|
|
|
|
/**
|
|
|
|
* Wait for next VSync, OpenGL method.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
vsync_opengl_wait(session_t *ps) {
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
unsigned vblank_count;
|
|
|
|
|
|
|
|
ps->glx_get_video_sync(&vblank_count);
|
|
|
|
ps->glx_wait_video_sync(2, (vblank_count + 1) % 2, &vblank_count);
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
// I see some code calling glXSwapIntervalSGI(1) afterwards, is it required?
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Wait for next VSync.
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
*/
|
|
|
|
static void
|
|
|
|
vsync_wait(session_t *ps) {
|
|
|
|
if (VSYNC_NONE == ps->o.vsync)
|
|
|
|
return;
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
|
|
|
|
#ifdef CONFIG_VSYNC_DRM
|
|
|
|
if (VSYNC_DRM == ps->o.vsync) {
|
|
|
|
vsync_drm_wait(ps);
|
|
|
|
return;
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
|
|
|
#ifdef CONFIG_VSYNC_OPENGL
|
|
|
|
if (VSYNC_OPENGL == ps->o.vsync) {
|
|
|
|
vsync_opengl_wait(ps);
|
|
|
|
return;
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
|
|
|
// This place should not reached!
|
|
|
|
assert(0);
|
|
|
|
|
|
|
|
return;
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Pregenerate alpha pictures.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
init_alpha_picts(session_t *ps) {
|
|
|
|
int i;
|
|
|
|
int num = round(1.0 / ps->o.alpha_step) + 1;
|
|
|
|
|
|
|
|
ps->alpha_picts = malloc(sizeof(Picture) * num);
|
|
|
|
|
|
|
|
for (i = 0; i < num; ++i) {
|
|
|
|
double o = i * ps->o.alpha_step;
|
|
|
|
if ((1.0 - o) > ps->o.alpha_step)
|
|
|
|
ps->alpha_picts[i] = solid_picture(ps, false, o, 0, 0, 0);
|
|
|
|
else
|
|
|
|
ps->alpha_picts[i] = None;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Initialize double buffer.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
init_dbe(session_t *ps) {
|
|
|
|
if (!(ps->root_dbe = XdbeAllocateBackBufferName(ps->dpy,
|
|
|
|
(ps->o.paint_on_overlay ? ps->overlay: ps->root), XdbeCopied))) {
|
|
|
|
fprintf(stderr, "Failed to create double buffer. Double buffering "
|
|
|
|
"turned off.\n");
|
|
|
|
ps->o.dbe = false;
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Initialize X composite overlay window.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
init_overlay(session_t *ps) {
|
|
|
|
ps->overlay = XCompositeGetOverlayWindow(ps->dpy, ps->root);
|
|
|
|
if (ps->overlay) {
|
|
|
|
// Set window region of the overlay window, code stolen from
|
|
|
|
// compiz-0.8.8
|
|
|
|
XserverRegion region = XFixesCreateRegion(ps->dpy, NULL, 0);
|
|
|
|
XFixesSetWindowShapeRegion(ps->dpy, ps->overlay, ShapeBounding, 0, 0, 0);
|
|
|
|
XFixesSetWindowShapeRegion(ps->dpy, ps->overlay, ShapeInput, 0, 0, region);
|
|
|
|
XFixesDestroyRegion(ps->dpy, region);
|
|
|
|
|
|
|
|
// Listen to Expose events on the overlay
|
|
|
|
XSelectInput(ps->dpy, ps->overlay, ExposureMask);
|
|
|
|
|
|
|
|
// Retrieve DamageNotify on root window if we are painting on an
|
|
|
|
// overlay
|
|
|
|
// root_damage = XDamageCreate(ps->dpy, root, XDamageReportNonEmpty);
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
fprintf(stderr, "Cannot get X Composite overlay window. Falling "
|
|
|
|
"back to painting on root window.\n");
|
|
|
|
ps->o.paint_on_overlay = false;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Query needed X Render filters to check for their existence.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
init_filters(session_t *ps) {
|
|
|
|
if (ps->o.blur_background || ps->o.blur_background_frame) {
|
|
|
|
// Query filters
|
|
|
|
XFilters *pf = XRenderQueryFilters(ps->dpy, get_tgt_window(ps));
|
|
|
|
if (pf) {
|
|
|
|
for (int i = 0; i < pf->nfilter; ++i) {
|
|
|
|
// Convolution filter
|
|
|
|
if (!strcmp(pf->filter[i], XRFILTER_CONVOLUTION))
|
|
|
|
ps->xrfilter_convolution_exists = true;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
XFree(pf);
|
|
|
|
|
|
|
|
// Turn features off if any required filter is not present
|
|
|
|
if (!ps->xrfilter_convolution_exists) {
|
|
|
|
fprintf(stderr, "X Render convolution filter unsupported by your X server. Background blur disabled.\n");
|
|
|
|
ps->o.blur_background = false;
|
|
|
|
ps->o.blur_background_frame = false;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Redirect all windows.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
redir_start(session_t *ps) {
|
|
|
|
if (!ps->redirected) {
|
|
|
|
#ifdef DEBUG_REDIR
|
|
|
|
printf("redir_start(): Screen redirected.\n");
|
|
|
|
#endif
|
|
|
|
|
|
|
|
// Map overlay window. Done firstly according to this:
|
|
|
|
// https://bugzilla.gnome.org/show_bug.cgi?id=597014
|
|
|
|
if (ps->overlay)
|
|
|
|
XMapWindow(ps->dpy, ps->overlay);
|
|
|
|
|
|
|
|
XCompositeRedirectSubwindows(ps->dpy, ps->root, CompositeRedirectManual);
|
|
|
|
|
|
|
|
// Must call XSync() here
|
|
|
|
XSync(ps->dpy, False);
|
|
|
|
|
|
|
|
ps->redirected = true;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Get the poll time.
|
|
|
|
*/
|
|
|
|
static time_ms_t
|
|
|
|
timeout_get_poll_time(session_t *ps) {
|
|
|
|
const time_ms_t now = get_time_ms();
|
|
|
|
time_ms_t wait = TIME_MS_MAX;
|
|
|
|
|
|
|
|
// Traverse throught the timeout linked list
|
|
|
|
for (timeout_t *ptmout = ps->tmout_lst; ptmout; ptmout = ptmout->next) {
|
|
|
|
if (ptmout->enabled) {
|
|
|
|
// Truncate the last run time to the closest interval
|
|
|
|
time_ms_t newrun = ptmout->firstrun + ((ptmout->lastrun - ptmout->firstrun) / ptmout->interval + 1) * ptmout->interval;
|
|
|
|
if (newrun <= now) {
|
|
|
|
wait = 0;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
time_ms_t newwait = newrun - now;
|
|
|
|
if (newwait < wait)
|
|
|
|
wait = newwait;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return wait;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Insert a new timeout.
|
|
|
|
*/
|
|
|
|
static timeout_t *
|
|
|
|
timeout_insert(session_t *ps, time_ms_t interval,
|
|
|
|
bool (*callback)(session_t *ps, timeout_t *ptmout), void *data) {
|
|
|
|
const static timeout_t tmout_def = {
|
|
|
|
.enabled = true,
|
|
|
|
.data = NULL,
|
|
|
|
.callback = NULL,
|
|
|
|
.firstrun = 0L,
|
|
|
|
.lastrun = 0L,
|
|
|
|
.interval = 0L,
|
|
|
|
};
|
|
|
|
|
|
|
|
const time_ms_t now = get_time_ms();
|
|
|
|
timeout_t *ptmout = malloc(sizeof(timeout_t));
|
|
|
|
if (!ptmout)
|
|
|
|
printf_errfq(1, "(): Failed to allocate memory for timeout.");
|
|
|
|
memcpy(ptmout, &tmout_def, sizeof(timeout_t));
|
|
|
|
|
|
|
|
ptmout->interval = interval;
|
|
|
|
ptmout->firstrun = now;
|
|
|
|
ptmout->lastrun = now;
|
|
|
|
ptmout->data = data;
|
|
|
|
ptmout->callback = callback;
|
|
|
|
ptmout->next = ps->tmout_lst;
|
|
|
|
ps->tmout_lst = ptmout;
|
|
|
|
|
|
|
|
return ptmout;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Drop a timeout.
|
|
|
|
*
|
|
|
|
* @return true if we have found the timeout and removed it, false
|
|
|
|
* otherwise
|
|
|
|
*/
|
|
|
|
static bool
|
|
|
|
timeout_drop(session_t *ps, timeout_t *prm) {
|
|
|
|
timeout_t **pplast = &ps->tmout_lst;
|
|
|
|
|
|
|
|
for (timeout_t *ptmout = ps->tmout_lst; ptmout;
|
|
|
|
pplast = &ptmout->next, ptmout = ptmout->next) {
|
|
|
|
if (prm == ptmout) {
|
|
|
|
*pplast = ptmout->next;
|
|
|
|
free(ptmout);
|
|
|
|
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Clear all timeouts.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
timeout_clear(session_t *ps) {
|
|
|
|
timeout_t *ptmout = ps->tmout_lst;
|
|
|
|
timeout_t *next = NULL;
|
|
|
|
while (ptmout) {
|
|
|
|
next = ptmout->next;
|
|
|
|
free(ptmout);
|
|
|
|
ptmout = next;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Run timeouts.
|
|
|
|
*
|
|
|
|
* @return true if we have ran a timeout, false otherwise
|
|
|
|
*/
|
|
|
|
static bool
|
|
|
|
timeout_run(session_t *ps) {
|
|
|
|
const time_ms_t now = get_time_ms();
|
|
|
|
bool ret = false;
|
|
|
|
|
|
|
|
for (timeout_t *ptmout = ps->tmout_lst; ptmout; ptmout = ptmout->next) {
|
|
|
|
if (ptmout->enabled) {
|
|
|
|
const time_ms_t max = now +
|
|
|
|
(time_ms_t) (ptmout->interval * TIMEOUT_RUN_TOLERANCE);
|
|
|
|
time_ms_t newrun = ptmout->firstrun + ((ptmout->lastrun - ptmout->firstrun) / ptmout->interval + 1) * ptmout->interval;
|
|
|
|
if (newrun <= max) {
|
|
|
|
ret = true;
|
|
|
|
timeout_invoke(ps, ptmout);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Invoke a timeout.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
timeout_invoke(session_t *ps, timeout_t *ptmout) {
|
|
|
|
const time_ms_t now = get_time_ms();
|
|
|
|
ptmout->lastrun = now;
|
|
|
|
// Avoid modifying the timeout structure after running timeout, to
|
|
|
|
// make it possible to remove timeout in callback
|
|
|
|
if (ptmout->callback)
|
|
|
|
ptmout->callback(ps, ptmout);
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Unredirect all windows.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
redir_stop(session_t *ps) {
|
|
|
|
if (ps->redirected) {
|
|
|
|
#ifdef DEBUG_REDIR
|
|
|
|
printf("redir_stop(): Screen unredirected.\n");
|
|
|
|
#endif
|
|
|
|
// Destroy all Pictures as they expire once windows are unredirected
|
|
|
|
// If we don't destroy them here, looks like the resources are just
|
|
|
|
// kept inaccessible somehow
|
|
|
|
for (win *w = ps->list; w; w = w->next) {
|
|
|
|
free_pixmap(ps, &w->pixmap);
|
|
|
|
free_picture(ps, &w->picture);
|
|
|
|
}
|
|
|
|
|
|
|
|
XCompositeUnredirectSubwindows(ps->dpy, ps->root, CompositeRedirectManual);
|
|
|
|
// Unmap overlay window
|
|
|
|
if (ps->overlay)
|
|
|
|
XUnmapWindow(ps->dpy, ps->overlay);
|
|
|
|
|
|
|
|
// Must call XSync() here
|
|
|
|
XSync(ps->dpy, False);
|
|
|
|
|
|
|
|
ps->redirected = false;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
/**
|
|
|
|
* Main loop.
|
|
|
|
*/
|
|
|
|
static bool
|
|
|
|
mainloop(session_t *ps) {
|
|
|
|
// Process existing events
|
|
|
|
// Sometimes poll() returns 1 but no events are actually read,
|
|
|
|
// causing XNextEvent() to block, I have no idea what's wrong, so we
|
|
|
|
// check for the number of events here.
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
if (XEventsQueued(ps->dpy, QueuedAfterReading)) {
|
|
|
|
XEvent ev = { };
|
|
|
|
|
|
|
|
XNextEvent(ps->dpy, &ev);
|
|
|
|
ev_handle(ps, &ev);
|
|
|
|
ps->ev_received = true;
|
|
|
|
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (ps->reset)
|
|
|
|
return false;
|
|
|
|
|
|
|
|
// Calculate timeout
|
|
|
|
struct timeval *ptv = NULL;
|
|
|
|
{
|
|
|
|
// Consider ev_received firstly
|
|
|
|
if (ps->ev_received) {
|
|
|
|
ptv = malloc(sizeof(struct timeval));
|
|
|
|
ptv->tv_sec = 0L;
|
|
|
|
ptv->tv_usec = 0L;
|
|
|
|
}
|
|
|
|
// Then consider fading timeout
|
|
|
|
else if (!ps->idling) {
|
|
|
|
ptv = malloc(sizeof(struct timeval));
|
|
|
|
*ptv = ms_to_tv(fade_timeout(ps));
|
|
|
|
}
|
|
|
|
|
|
|
|
// Software optimization is to be applied on timeouts that require
|
|
|
|
// immediate painting only
|
|
|
|
if (ptv && ps->o.sw_opti)
|
|
|
|
swopti_handle_timeout(ps, ptv);
|
|
|
|
|
|
|
|
// Don't continue looping for 0 timeout
|
|
|
|
if (ptv && timeval_isempty(ptv)) {
|
|
|
|
free(ptv);
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
return false;
|
|
|
|
}
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
|
|
|
|
// Now consider the waiting time of other timeouts
|
|
|
|
time_ms_t tmout_ms = timeout_get_poll_time(ps);
|
|
|
|
if (tmout_ms < TIME_MS_MAX) {
|
|
|
|
if (!ptv) {
|
|
|
|
ptv = malloc(sizeof(struct timeval));
|
|
|
|
*ptv = ms_to_tv(tmout_ms);
|
|
|
|
}
|
|
|
|
else if (timeval_ms_cmp(ptv, tmout_ms) > 0) {
|
|
|
|
*ptv = ms_to_tv(tmout_ms);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
// Don't continue looping for 0 timeout
|
|
|
|
if (ptv && timeval_isempty(ptv)) {
|
|
|
|
free(ptv);
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
}
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
|
|
|
|
// Polling
|
|
|
|
fds_poll(ps, ptv);
|
|
|
|
free(ptv);
|
|
|
|
ptv = NULL;
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
|
|
|
|
timeout_run(ps);
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Initialize a session.
|
|
|
|
*
|
|
|
|
* @param ps_old old session, from which the function will take the X
|
|
|
|
* connection, then free it
|
|
|
|
* @param argc number of commandline arguments
|
|
|
|
* @param argv commandline arguments
|
|
|
|
*/
|
|
|
|
static session_t *
|
|
|
|
session_init(session_t *ps_old, int argc, char **argv) {
|
|
|
|
const static session_t s_def = {
|
|
|
|
.dpy = NULL,
|
|
|
|
.scr = 0,
|
|
|
|
.vis = NULL,
|
|
|
|
.depth = 0,
|
|
|
|
.root = None,
|
|
|
|
.root_height = 0,
|
|
|
|
.root_width = 0,
|
|
|
|
// .root_damage = None,
|
|
|
|
.overlay = None,
|
|
|
|
.root_tile = None,
|
|
|
|
.screen_reg = None,
|
|
|
|
.tgt_picture = None,
|
|
|
|
.tgt_buffer = None,
|
|
|
|
.root_dbe = None,
|
|
|
|
.reg_win = None,
|
|
|
|
.o = {
|
|
|
|
.display = NULL,
|
|
|
|
.mark_wmwin_focused = false,
|
|
|
|
.mark_ovredir_focused = false,
|
|
|
|
.fork_after_register = false,
|
|
|
|
.synchronize = false,
|
|
|
|
.detect_rounded_corners = false,
|
|
|
|
.paint_on_overlay = false,
|
|
|
|
.unredir_if_possible = false,
|
|
|
|
.dbus = false,
|
|
|
|
.logpath = NULL,
|
|
|
|
|
|
|
|
.refresh_rate = 0,
|
|
|
|
.sw_opti = false,
|
|
|
|
.vsync = VSYNC_NONE,
|
|
|
|
.dbe = false,
|
|
|
|
.vsync_aggressive = false,
|
|
|
|
|
|
|
|
.wintype_shadow = { false },
|
|
|
|
.shadow_red = 0.0,
|
|
|
|
.shadow_green = 0.0,
|
|
|
|
.shadow_blue = 0.0,
|
|
|
|
.shadow_radius = 12,
|
|
|
|
.shadow_offset_x = -15,
|
|
|
|
.shadow_offset_y = -15,
|
|
|
|
.shadow_opacity = .75,
|
|
|
|
.clear_shadow = false,
|
|
|
|
.shadow_blacklist = NULL,
|
|
|
|
.shadow_ignore_shaped = false,
|
|
|
|
.respect_prop_shadow = false,
|
|
|
|
|
|
|
|
.wintype_fade = { false },
|
|
|
|
.fade_in_step = 0.028 * OPAQUE,
|
|
|
|
.fade_out_step = 0.03 * OPAQUE,
|
|
|
|
.fade_delta = 10,
|
|
|
|
.no_fading_openclose = false,
|
|
|
|
.fade_blacklist = NULL,
|
|
|
|
|
|
|
|
.wintype_opacity = { 0.0 },
|
|
|
|
.inactive_opacity = 0,
|
|
|
|
.inactive_opacity_override = false,
|
|
|
|
.frame_opacity = 0.0,
|
|
|
|
.detect_client_opacity = false,
|
|
|
|
.inactive_dim = 0.0,
|
|
|
|
.inactive_dim_fixed = false,
|
|
|
|
.alpha_step = 0.03,
|
|
|
|
.blur_background = false,
|
|
|
|
.blur_background_frame = false,
|
|
|
|
.blur_background_fixed = false,
|
|
|
|
|
|
|
|
.wintype_focus = { false },
|
|
|
|
.use_ewmh_active_win = false,
|
|
|
|
.focus_blacklist = NULL,
|
|
|
|
.detect_transient = false,
|
|
|
|
.detect_client_leader = false,
|
|
|
|
|
|
|
|
.track_focus = false,
|
|
|
|
.track_wdata = false,
|
|
|
|
.track_leader = false,
|
|
|
|
},
|
|
|
|
|
|
|
|
.pfds_read = NULL,
|
|
|
|
.pfds_write = NULL,
|
|
|
|
.pfds_except = NULL,
|
|
|
|
.nfds_max = 0,
|
|
|
|
.tmout_lst = NULL,
|
|
|
|
|
|
|
|
.all_damage = None,
|
|
|
|
.time_start = { 0, 0 },
|
|
|
|
.redirected = false,
|
|
|
|
.unredir_possible = false,
|
|
|
|
.alpha_picts = NULL,
|
|
|
|
.reg_ignore_expire = false,
|
|
|
|
.idling = false,
|
|
|
|
.fade_time = 0L,
|
|
|
|
.ignore_head = NULL,
|
|
|
|
.ignore_tail = NULL,
|
|
|
|
.reset = false,
|
|
|
|
|
|
|
|
.expose_rects = NULL,
|
|
|
|
.size_expose = 0,
|
|
|
|
.n_expose = 0,
|
|
|
|
|
|
|
|
.list = NULL,
|
|
|
|
.active_win = NULL,
|
|
|
|
.active_leader = None,
|
|
|
|
|
|
|
|
.black_picture = None,
|
|
|
|
.cshadow_picture = None,
|
|
|
|
.gaussian_map = NULL,
|
|
|
|
.cgsize = 0,
|
|
|
|
.shadow_corner = NULL,
|
|
|
|
.shadow_top = NULL,
|
|
|
|
|
|
|
|
.refresh_rate = 0,
|
|
|
|
.refresh_intv = 0UL,
|
|
|
|
.paint_tm_offset = 0L,
|
|
|
|
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
#ifdef CONFIG_VSYNC_DRM
|
|
|
|
.drm_fd = 0,
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
#endif
|
|
|
|
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
#ifdef CONFIG_VSYNC_OPENGL
|
|
|
|
.glx_context = None,
|
|
|
|
.glx_get_video_sync = NULL,
|
|
|
|
.glx_wait_video_sync = NULL,
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
#endif
|
|
|
|
|
|
|
|
.xfixes_event = 0,
|
|
|
|
.xfixes_error = 0,
|
|
|
|
.damage_event = 0,
|
|
|
|
.damage_error = 0,
|
|
|
|
.render_event = 0,
|
|
|
|
.render_error = 0,
|
|
|
|
.composite_event = 0,
|
|
|
|
.composite_error = 0,
|
|
|
|
.composite_opcode = 0,
|
|
|
|
.has_name_pixmap = false,
|
|
|
|
.shape_exists = false,
|
|
|
|
.shape_event = 0,
|
|
|
|
.shape_error = 0,
|
|
|
|
.randr_exists = 0,
|
|
|
|
.randr_event = 0,
|
|
|
|
.randr_error = 0,
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
#ifdef CONFIG_VSYNC_OPENGL
|
|
|
|
.glx_exists = false,
|
|
|
|
.glx_event = 0,
|
|
|
|
.glx_error = 0,
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
#endif
|
|
|
|
.dbe_exists = false,
|
|
|
|
.xrfilter_convolution_exists = false,
|
|
|
|
|
|
|
|
.atom_opacity = None,
|
|
|
|
.atom_frame_extents = None,
|
|
|
|
.atom_client = None,
|
|
|
|
.atom_name = None,
|
|
|
|
.atom_name_ewmh = None,
|
|
|
|
.atom_class = None,
|
|
|
|
.atom_role = None,
|
|
|
|
.atom_transient = None,
|
|
|
|
.atom_ewmh_active_win = None,
|
|
|
|
.atom_compton_shadow = None,
|
|
|
|
.atom_win_type = None,
|
|
|
|
.atoms_wintypes = { 0 }
|
|
|
|
};
|
|
|
|
|
|
|
|
// Allocate a session and copy default values into it
|
|
|
|
session_t *ps = malloc(sizeof(session_t));
|
|
|
|
memcpy(ps, &s_def, sizeof(session_t));
|
|
|
|
ps_g = ps;
|
|
|
|
ps->ignore_tail = &ps->ignore_head;
|
|
|
|
gettimeofday(&ps->time_start, NULL);
|
|
|
|
|
|
|
|
wintype_arr_enable(ps->o.wintype_focus);
|
|
|
|
ps->o.wintype_focus[WINTYPE_UNKNOWN] = false;
|
|
|
|
ps->o.wintype_focus[WINTYPE_NORMAL] = false;
|
|
|
|
ps->o.wintype_focus[WINTYPE_UTILITY] = false;
|
|
|
|
|
|
|
|
get_cfg(ps, argc, argv);
|
|
|
|
|
|
|
|
// Inherit old Display if possible, primarily for resource leak checking
|
|
|
|
if (ps_old && ps_old->dpy)
|
|
|
|
ps->dpy = ps_old->dpy;
|
|
|
|
|
|
|
|
// Open Display
|
|
|
|
if (!ps->dpy) {
|
|
|
|
ps->dpy = XOpenDisplay(ps->o.display);
|
|
|
|
if (!ps->dpy) {
|
|
|
|
fprintf(stderr, "Can't open display\n");
|
|
|
|
exit(1);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
XSetErrorHandler(error);
|
|
|
|
if (ps->o.synchronize) {
|
|
|
|
XSynchronize(ps->dpy, 1);
|
|
|
|
}
|
|
|
|
|
|
|
|
ps->scr = DefaultScreen(ps->dpy);
|
|
|
|
ps->root = RootWindow(ps->dpy, ps->scr);
|
|
|
|
|
|
|
|
ps->vis = DefaultVisual(ps->dpy, ps->scr);
|
|
|
|
ps->depth = DefaultDepth(ps->dpy, ps->scr);
|
|
|
|
|
|
|
|
if (!XRenderQueryExtension(ps->dpy,
|
|
|
|
&ps->render_event, &ps->render_error)) {
|
|
|
|
fprintf(stderr, "No render extension\n");
|
|
|
|
exit(1);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (!XQueryExtension(ps->dpy, COMPOSITE_NAME, &ps->composite_opcode,
|
|
|
|
&ps->composite_event, &ps->composite_error)) {
|
|
|
|
fprintf(stderr, "No composite extension\n");
|
|
|
|
exit(1);
|
|
|
|
}
|
|
|
|
|
|
|
|
{
|
|
|
|
int composite_major = 0, composite_minor = 0;
|
|
|
|
|
|
|
|
XCompositeQueryVersion(ps->dpy, &composite_major, &composite_minor);
|
|
|
|
|
|
|
|
if (composite_major > 0 || composite_minor >= 2) {
|
|
|
|
ps->has_name_pixmap = true;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
if (!XDamageQueryExtension(ps->dpy, &ps->damage_event, &ps->damage_error)) {
|
|
|
|
fprintf(stderr, "No damage extension\n");
|
|
|
|
exit(1);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (!XFixesQueryExtension(ps->dpy, &ps->xfixes_event, &ps->xfixes_error)) {
|
|
|
|
fprintf(stderr, "No XFixes extension\n");
|
|
|
|
exit(1);
|
|
|
|
}
|
|
|
|
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
// Query X Shape
|
|
|
|
if (XShapeQueryExtension(ps->dpy, &ps->shape_event, &ps->shape_error)) {
|
|
|
|
ps->shape_exists = true;
|
|
|
|
}
|
|
|
|
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
// Query X RandR
|
|
|
|
if (ps->o.sw_opti && !ps->o.refresh_rate) {
|
|
|
|
if (XRRQueryExtension(ps->dpy, &ps->randr_event, &ps->randr_error))
|
|
|
|
ps->randr_exists = true;
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
else
|
|
|
|
fprintf(stderr, "No XRandR extension, automatic refresh rate "
|
|
|
|
"detection impossible.\n");
|
|
|
|
}
|
|
|
|
|
|
|
|
#ifdef CONFIG_VSYNC_OPENGL
|
|
|
|
// Query X GLX extension
|
|
|
|
if (VSYNC_OPENGL == ps->o.vsync) {
|
|
|
|
if (glXQueryExtension(ps->dpy, &ps->glx_event, &ps->glx_error))
|
|
|
|
ps->glx_exists = true;
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
else {
|
|
|
|
fprintf(stderr, "No GLX extension, OpenGL VSync impossible.\n");
|
|
|
|
ps->o.vsync = VSYNC_NONE;
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
}
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
|
|
|
// Query X DBE extension
|
|
|
|
if (ps->o.dbe) {
|
|
|
|
int dbe_ver_major = 0, dbe_ver_minor = 0;
|
|
|
|
if (XdbeQueryExtension(ps->dpy, &dbe_ver_major, &dbe_ver_minor))
|
|
|
|
if (dbe_ver_major >= 1)
|
|
|
|
ps->dbe_exists = true;
|
|
|
|
else
|
|
|
|
fprintf(stderr, "DBE extension version too low. Double buffering "
|
|
|
|
"impossible.\n");
|
|
|
|
else {
|
|
|
|
fprintf(stderr, "No DBE extension. Double buffering impossible.\n");
|
|
|
|
}
|
|
|
|
if (!ps->dbe_exists)
|
|
|
|
ps->o.dbe = false;
|
|
|
|
}
|
|
|
|
|
|
|
|
register_cm(ps, (VSYNC_OPENGL == ps->o.vsync));
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
|
|
|
|
// Initialize software optimization
|
|
|
|
if (ps->o.sw_opti)
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
ps->o.sw_opti = swopti_init(ps);
|
|
|
|
|
|
|
|
// Initialize DRM/OpenGL VSync
|
|
|
|
if ((VSYNC_DRM == ps->o.vsync && !vsync_drm_init(ps))
|
|
|
|
|| (VSYNC_OPENGL == ps->o.vsync && !vsync_opengl_init(ps)))
|
|
|
|
ps->o.vsync = VSYNC_NONE;
|
|
|
|
|
|
|
|
// Overlay must be initialized before double buffer
|
|
|
|
if (ps->o.paint_on_overlay)
|
|
|
|
init_overlay(ps);
|
|
|
|
|
|
|
|
if (ps->o.dbe)
|
|
|
|
init_dbe(ps);
|
|
|
|
|
|
|
|
init_atoms(ps);
|
|
|
|
init_alpha_picts(ps);
|
|
|
|
|
|
|
|
ps->gaussian_map = make_gaussian_map(ps->o.shadow_radius);
|
|
|
|
presum_gaussian(ps, ps->gaussian_map);
|
|
|
|
|
|
|
|
ps->root_width = DisplayWidth(ps->dpy, ps->scr);
|
|
|
|
ps->root_height = DisplayHeight(ps->dpy, ps->scr);
|
|
|
|
|
|
|
|
rebuild_screen_reg(ps);
|
|
|
|
|
|
|
|
{
|
|
|
|
XRenderPictureAttributes pa;
|
|
|
|
pa.subwindow_mode = IncludeInferiors;
|
|
|
|
|
|
|
|
ps->root_picture = XRenderCreatePicture(ps->dpy, ps->root,
|
|
|
|
XRenderFindVisualFormat(ps->dpy, ps->vis),
|
|
|
|
CPSubwindowMode, &pa);
|
|
|
|
if (ps->o.paint_on_overlay) {
|
|
|
|
ps->tgt_picture = XRenderCreatePicture(ps->dpy, ps->overlay,
|
|
|
|
XRenderFindVisualFormat(ps->dpy, ps->vis),
|
|
|
|
CPSubwindowMode, &pa);
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
ps->tgt_picture = ps->root_picture;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
init_filters(ps);
|
|
|
|
|
|
|
|
ps->black_picture = solid_picture(ps, true, 1, 0, 0, 0);
|
|
|
|
|
|
|
|
// Generates another Picture for shadows if the color is modified by
|
|
|
|
// user
|
|
|
|
if (!ps->o.shadow_red && !ps->o.shadow_green && !ps->o.shadow_blue) {
|
|
|
|
ps->cshadow_picture = ps->black_picture;
|
|
|
|
} else {
|
|
|
|
ps->cshadow_picture = solid_picture(ps, true, 1,
|
|
|
|
ps->o.shadow_red, ps->o.shadow_green, ps->o.shadow_blue);
|
|
|
|
}
|
|
|
|
|
|
|
|
fds_insert(ps, ConnectionNumber(ps->dpy), POLLIN);
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
|
|
|
|
XGrabServer(ps->dpy);
|
|
|
|
|
|
|
|
redir_start(ps);
|
|
|
|
|
|
|
|
XSelectInput(ps->dpy, ps->root,
|
|
|
|
SubstructureNotifyMask
|
|
|
|
| ExposureMask
|
|
|
|
| StructureNotifyMask
|
|
|
|
| PropertyChangeMask);
|
|
|
|
|
|
|
|
{
|
|
|
|
Window root_return, parent_return;
|
|
|
|
Window *children;
|
|
|
|
unsigned int nchildren;
|
|
|
|
|
|
|
|
XQueryTree(ps->dpy, ps->root, &root_return,
|
|
|
|
&parent_return, &children, &nchildren);
|
|
|
|
|
|
|
|
for (unsigned i = 0; i < nchildren; i++) {
|
|
|
|
add_win(ps, children[i], i ? children[i-1] : None);
|
|
|
|
}
|
|
|
|
|
|
|
|
XFree(children);
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
if (ps->o.track_focus) {
|
|
|
|
recheck_focus(ps);
|
|
|
|
}
|
|
|
|
|
|
|
|
XUngrabServer(ps->dpy);
|
|
|
|
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
// Fork to background, if asked
|
|
|
|
if (ps->o.fork_after_register) {
|
|
|
|
if (!fork_after(ps)) {
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
session_destroy(ps);
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
// Free the old session
|
|
|
|
if (ps_old)
|
|
|
|
free(ps_old);
|
|
|
|
|
|
|
|
return ps;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Destroy a session.
|
|
|
|
*
|
|
|
|
* Does not close the X connection or free the <code>session_t</code>
|
|
|
|
* structure, though.
|
|
|
|
*
|
|
|
|
* @param ps session to destroy
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
session_destroy(session_t *ps) {
|
|
|
|
redir_stop(ps);
|
|
|
|
|
|
|
|
// Stop listening to events on root window
|
|
|
|
XSelectInput(ps->dpy, ps->root, 0);
|
|
|
|
|
|
|
|
// Free window linked list
|
|
|
|
{
|
|
|
|
win *next = NULL;
|
|
|
|
for (win *w = ps->list; w; w = next) {
|
|
|
|
// Must be put here to avoid segfault
|
|
|
|
next = w->next;
|
|
|
|
|
|
|
|
if (IsViewable == w->a.map_state && !w->destroyed)
|
|
|
|
win_ev_stop(ps, w);
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
|
|
|
|
free_win_res(ps, w);
|
|
|
|
free(w);
|
|
|
|
}
|
|
|
|
|
|
|
|
ps->list = NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
// Free alpha_picts
|
|
|
|
{
|
|
|
|
const int max = round(1.0 / ps->o.alpha_step) + 1;
|
|
|
|
for (int i = 0; i < max; ++i)
|
|
|
|
free_picture(ps, &ps->alpha_picts[i]);
|
|
|
|
free(ps->alpha_picts);
|
|
|
|
ps->alpha_picts = NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
// Free blacklists
|
|
|
|
free_wincondlst(&ps->o.shadow_blacklist);
|
|
|
|
free_wincondlst(&ps->o.fade_blacklist);
|
|
|
|
|
|
|
|
// Free ignore linked list
|
|
|
|
{
|
|
|
|
ignore_t *next = NULL;
|
|
|
|
for (ignore_t *ign = ps->ignore_head; ign; ign = next) {
|
|
|
|
next = ign->next;
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
|
|
|
|
free(ign);
|
|
|
|
}
|
|
|
|
|
|
|
|
// Reset head and tail
|
|
|
|
ps->ignore_head = NULL;
|
|
|
|
ps->ignore_tail = &ps->ignore_head;
|
|
|
|
}
|
|
|
|
|
|
|
|
// Free cshadow_picture and black_picture
|
|
|
|
if (ps->cshadow_picture == ps->black_picture)
|
|
|
|
ps->cshadow_picture = None;
|
|
|
|
else
|
|
|
|
free_picture(ps, &ps->cshadow_picture);
|
|
|
|
|
|
|
|
free_picture(ps, &ps->black_picture);
|
|
|
|
|
|
|
|
// Free tgt_{buffer,picture} and root_picture
|
|
|
|
if (ps->tgt_buffer == ps->tgt_picture)
|
|
|
|
ps->tgt_buffer = None;
|
|
|
|
else
|
|
|
|
free_picture(ps, &ps->tgt_buffer);
|
|
|
|
|
|
|
|
if (ps->tgt_picture == ps->root_picture)
|
|
|
|
ps->tgt_picture = None;
|
|
|
|
else
|
|
|
|
free_picture(ps, &ps->tgt_picture);
|
|
|
|
|
|
|
|
free_picture(ps, &ps->root_picture);
|
|
|
|
|
|
|
|
// Free other X resources
|
|
|
|
free_picture(ps, &ps->root_tile);
|
|
|
|
free_region(ps, &ps->screen_reg);
|
|
|
|
free_region(ps, &ps->all_damage);
|
|
|
|
free(ps->expose_rects);
|
|
|
|
free(ps->shadow_corner);
|
|
|
|
free(ps->shadow_top);
|
|
|
|
free(ps->gaussian_map);
|
|
|
|
free(ps->o.display);
|
|
|
|
free(ps->o.logpath);
|
|
|
|
free(ps->pfds_read);
|
|
|
|
free(ps->pfds_write);
|
|
|
|
free(ps->pfds_except);
|
|
|
|
|
|
|
|
// Free reg_win and glx_context
|
|
|
|
if (ps->reg_win) {
|
|
|
|
XDestroyWindow(ps->dpy, ps->reg_win);
|
|
|
|
ps->reg_win = None;
|
|
|
|
}
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
#ifdef CONFIG_VSYNC_OPENGL
|
|
|
|
if (ps->glx_context) {
|
|
|
|
glXDestroyContext(ps->dpy, ps->glx_context);
|
|
|
|
ps->glx_context = None;
|
|
|
|
}
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
#endif
|
|
|
|
|
|
|
|
// Free double buffer
|
|
|
|
if (ps->root_dbe) {
|
|
|
|
XdbeDeallocateBackBufferName(ps->dpy, ps->root_dbe);
|
|
|
|
ps->root_dbe = None;
|
|
|
|
}
|
|
|
|
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
#ifdef CONFIG_VSYNC_DRM
|
|
|
|
// Close file opened for DRM VSync
|
|
|
|
if (ps->drm_fd) {
|
|
|
|
close(ps->drm_fd);
|
|
|
|
ps->drm_fd = 0;
|
|
|
|
}
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
#endif
|
|
|
|
|
|
|
|
// Release overlay window
|
|
|
|
if (ps->overlay) {
|
|
|
|
XCompositeReleaseOverlayWindow(ps->dpy, ps->overlay);
|
|
|
|
ps->overlay = None;
|
|
|
|
}
|
|
|
|
|
|
|
|
// Flush all events
|
|
|
|
XSync(ps->dpy, True);
|
|
|
|
|
|
|
|
// Free timeouts
|
|
|
|
timeout_clear(ps);
|
|
|
|
|
|
|
|
if (ps == ps_g)
|
|
|
|
ps_g = NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Do the actual work.
|
|
|
|
*
|
|
|
|
* @param ps current session
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
session_run(session_t *ps) {
|
|
|
|
win *t;
|
|
|
|
|
|
|
|
if (ps->o.sw_opti)
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
ps->paint_tm_offset = get_time_timeval().tv_usec;
|
|
|
|
|
|
|
|
ps->reg_ignore_expire = true;
|
|
|
|
|
|
|
|
t = paint_preprocess(ps, ps->list);
|
|
|
|
|
|
|
|
if (ps->redirected)
|
|
|
|
paint_all(ps, None, t);
|
|
|
|
|
|
|
|
// Initialize idling
|
|
|
|
ps->idling = false;
|
|
|
|
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
// Main loop
|
|
|
|
while (!ps->reset) {
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
ps->ev_received = false;
|
|
|
|
|
|
|
|
while (mainloop(ps))
|
|
|
|
continue;
|
|
|
|
|
|
|
|
// idling will be turned off during paint_preprocess() if needed
|
|
|
|
ps->idling = true;
|
|
|
|
|
|
|
|
t = paint_preprocess(ps, ps->list);
|
Feature: #7: VSync
- Add VSync feature. 3 possible VSync methods available: "sw" (software,
not too reliable, but at least you have something to fallback to),
"drm" (using DRM_IOCTL_WAIT_VBLANK, should work only on DRI drivers),
"opengl" (using SGI_swap_control extension OpenGL, might work on more
drivers than the DRM method). "sw" and "opengl" are briefly tested,
"drm" received utterly no test (because I use the nVidia binary blob).
They are enabled with "--vsync sw" / "--vsync drm" / "--vsync opengl".
- Add --refresh-rate to let user specify a refresh rate for software
VSync, in case the automatic refresh rate detection does not work
well.
- Seemingly the automatic refresh rate detection using X RandR in
software VSync detects refresh rate incorrectly. Need further investigation.
- Fix a few bugs in fading timing.
- Add a workaround for client window detection on Fluxbox, as Fluxbox
(incorrectly?) sets the override-redirect flag upon all frame
windows.
- Software VSync adds dependency on librt (a part of glibc) for
nanosecond-level timing functions, and libXrandr for automatic refresh
rate detection; DRM VSync adds dependency on libdrm to use its drm.h,
but does not link to libdrm; OpenGL VSync adds dependency on libGL.
- Print timing information on DEBUG_REPAINT.
12 years ago
|
|
|
|
|
|
|
// If the screen is unredirected, free all_damage to stop painting
|
|
|
|
if (!ps->redirected)
|
|
|
|
free_region(ps, &ps->all_damage);
|
|
|
|
|
|
|
|
if (ps->all_damage && !is_region_empty(ps, ps->all_damage)) {
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
static int paint = 0;
|
|
|
|
paint_all(ps, ps->all_damage, t);
|
|
|
|
ps->reg_ignore_expire = false;
|
|
|
|
paint++;
|
|
|
|
XSync(ps->dpy, False);
|
|
|
|
ps->all_damage = None;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (ps->idling)
|
|
|
|
ps->fade_time = 0L;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Turn on the program reset flag.
|
|
|
|
*
|
|
|
|
* This will result in compton resetting itself after next paint.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
reset_enable(int __attribute__((unused)) signum) {
|
|
|
|
session_t * const ps = ps_g;
|
|
|
|
|
|
|
|
ps->reset = true;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* The function that everybody knows.
|
|
|
|
*/
|
|
|
|
int
|
|
|
|
main(int argc, char **argv) {
|
|
|
|
// Set locale so window names with special characters are interpreted
|
|
|
|
// correctly
|
|
|
|
setlocale(LC_ALL, "");
|
|
|
|
|
|
|
|
// Set up SIGUSR1 signal handler to reset program
|
|
|
|
{
|
|
|
|
sigset_t block_mask;
|
|
|
|
sigemptyset(&block_mask);
|
|
|
|
const struct sigaction action= {
|
|
|
|
.sa_handler = reset_enable,
|
|
|
|
.sa_mask = block_mask,
|
|
|
|
.sa_flags = 0
|
|
|
|
};
|
|
|
|
sigaction(SIGUSR1, &action, NULL);
|
|
|
|
}
|
|
|
|
|
|
|
|
// Main loop
|
|
|
|
session_t *ps_old = ps_g;
|
|
|
|
while (1) {
|
|
|
|
ps_g = session_init(ps_old, argc, argv);
|
Improvement #74: Use libevent for main loop
- Use libevent for main loop. I will explain the reasons in #56 later.
The preferred libevent version is 2.x, yet 1.4.x should work as well.
- As a result, compton now should build fine on *BSD. Thanks to
DachiChang for the FreeBSD build issue report.
- Another consequence is we now use microsecond-level timing all the
way. Nanosecond-level code will be dropped soon. Start using long
instead of unsigned long to represent time in milliseconds, as both
can't hold the full epoch time in ms, anyway, and a signed type
requires less care in subtraction. Wrap the epoch time in ms to 15
days.
- Fix broken NO_VSYNC_DRM and NO_VSYNC_OPENGL compile-time options.
- Use git revision number for versioning in Makefile, and other small
improvements.
- Reorganize struct _win. Drop unused w->damaged_sequence. w->damaged is
turned to bool.
- Add type and format to winprop_t, as preparation for the new condition
format.
- Add w->shadow_force and w->focus_force, to prepare for D-Bus support.
- Rename wid_get_prop() to wid_get_prop_adv() with more options. Add
wrapper function wid_get_prop().
- Add some extra helper functions, for D-Bus support later.
- Make some functions return a bool value to indicate if it's
successful.
- Modify add_win(), use a static const structure to initialize the new
struct _win.
- Add some helper macros, like printf_err(f)(q). Make some errors fatal.
- Rename some types, constants, and functions. Code clean-up.
- Check for time disorder in paint_preprocess() when calculating fading
steps.
- Rename evpoll() to swopti_handle_timeout(), and partially rewrite it.
- Make -h / --help legal.
- Known issue: compton segfaults on FreeBSD with nvidia-drivers, unless
NO_VSYNC_OPENGL is used. Will look into it later. Thamls to DachiChang
for reporting.
12 years ago
|
|
|
if (!ps_g) {
|
|
|
|
printf_errf("Failed to create new session.");
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
session_run(ps_g);
|
|
|
|
ps_old = ps_g;
|
|
|
|
session_destroy(ps_g);
|
|
|
|
}
|
|
|
|
|
|
|
|
free(ps_g);
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|