Age | Commit message (Collapse) | Author | Files | Lines |
|
|
|
If we haven't seen cursor_serial before, we need to xcb_xfixes_get_cursor_image()
and convert to native cursor.
A mechanism for caching of converted cursor images would be useful, perhaps
something like ???
- If cursor_serial not in cache
-- If cache is full, call a registered function to dispose of LRU entry in cache
-- Call a registered function do the conversion
-- Put result in cache
- Pass result from cache in callback here
|
|
|
|
XXX: incomplete
|
|
callback fn should not free() it now
|
|
xcwm_window_t
|
|
|
|
|
|
Retire xcwm_window_sizing_t in favour of xcb_size_hints_t, which contains
the full data it was being derived from
|
|
|
|
XXX: Is it ok to select for just one of SubstructureRedirectMask
ResizeRedirectMask or ButtonPressMask, the events that only one client may
select for? Or should we select for all of them, to ensure no other WM is
running?
|
|
After testing with metacity-window-demo
Use correct length of atom array from xcb_ewmh_get_wm_window_type_reply()
Recognize _NET_WM_WINDOW_TYPE_DOCK and _NET_WM_WINDOW_TYPE_DESKTOP
Recognize _NET_WM_WINDOW_TYPE_SPLASHSCREEN as a synonym for _NET_WM_WINDOW_TYPE_SPLASH
|
|
|
|
|
|
Fix typos, whitespace, coments
Fix some comments which say "context" when they mean "window", which is
confusing because we have a different data type which is called a "context"
Remove unused set_wm_name_in_context() prototype
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
|
|
misplaced or misized
Would this be fixed by doing xsync() after xcb_configure_window()?
Server values for values not in value mask in configure_request can be out
of date if xcb_configure_window() we have made to pass on a configure_request
hasn't completed yet...
Ignoring configure requests for windows we are managing is wrong, it means
they can't request to move themselves?
|
|
|
|
It's not nice to fail to call xcb_map_window() in XCB_MAP_REQUEST, it leaves
a window which tries to map in the unmapped state...
|
|
|
|
If the window is deleted very quickly after it is created, we might still be
in xcwm_window_create(), trying to obtain window information. Make sure that
we just return NULL if the window no longer exists...
|
|
Don't create a xcwm_window_t for an InputOnly window
Handle _xcwm_window_create returning NULL in MAP_NOTIFY
|
|
|
|
|
|
Add a .dir-locals.el which sets the something approximating the code standards
specified in doc/internal_docs/coding_standard.txt
|
|
|
|
of interest changes
|
|
|
|
|
|
use the bounding box damage reoprting level
It also seems to be buggy somehow...
|
|
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
|
|
|
|
Use xcb_composite_name_pixmap() to name the off-screen pixmap for the window,
replacing it with a new one when the window is mapped or resized. Also name the
composite pixmap for adopted windows.
(We need to be more careful than currently in MAP_NOTIFY. Always update the
name of the off-fscreen pixmap in MAP_NOTIFY, but only generate a
XCWM_EVENT_WINDOW_CREATE for override-redirect windows (since one was already
generated in MAP_REQUEST for non-override-redirect windows)
Ensure border width is zero, as border is included in composite pixmap image of
the window.
|
|
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
|
|
XtoQ appears to rely on the native windowing system to constrain the window size.
For Windows, the application has to do this itself. Rather than marhsalling all
WM_NORMAL_HINTS data through xcwm, just add a helper function to perform the size
constraints for WM_NORMAL_HINTS
This supports the aspect ratio constraint, which we seem to be ignoring before.
Future work: libxcwm appears to assume that WM_NORMAL_HINTS doesn't change after
the window is mapped, which doesn't seem to be guaranteed.
|
|
This helps when comparing with the output of xwininfo -tree -root
|
|
Handle XCB_KEY_RELEASE as well as XCB_KEY_PRESS
Reporting keycode as a character isn't that helpful, use a number instead
Use correct datatype for each event
Report XID in hex
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
|
|
They are sent to all clients. There is no mechanism to express disinterest in this event.
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
|
|
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
|
|
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
|
|
The helps when testing changes to a libxcwm-based WM, as you don't need to
restart the clients to see the effect.
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
|
|
xcwm_window_configure() has width and height parameters in the opposite order to
the order used everywhere else.
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
|
|
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
|
|
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
|
|
contacted
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
|
|
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
|
|
xcb-util-wm commit 23a4eb6059e0a694e342c3ff41ab076dad50b745, present in 0.3.9,
changes the xcb_ewmh_set_supporting_wm_check() API. Update for that and require
xcb-util-wm 0.3.9
Use the updated xcb_ewmh_set_supporting_wm_check() to more correctly implement
EWMH, setting the _NET_SUPPORTING_WM_CHECK property on the root *and* child
window.
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
|
|
Also, make debug output for injected input events a bit clearer
Also, add button number to xcwm_input_mouse_button_event debug output
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
|
|
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
|
|
Also keep a few Apple specific patterns
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
|
|
Install a pkg-config .pc file, along with the library and headers
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
|