When we are painting a stack of 5-10 maximized windows, the
standard bottom-to-top method of drawing every actor results
in a tremendous amount of overdraw and can easily max out
the available memory bandwidth on a low-end* graphics chipset.
It's even worse if window textures are being accessed over
the AGP bus.
When we have opaque windows, we can go ahead and compute visibility
ourselves (in classic X-server fashion) and use that information to
restrict drawing obscured actors.
* Add MutterWindowGroup - a ClutterGroup subclass with logic
for figuring out obscured regions.
* Add mutter_window_get_obscured_region() to get the region
obscured by that window.
* Add mutter_shaped_texture_set_clip_region() to hint
a clip region to the painting code; this is set based on
the computed visible region of MutterWindowGroup.
* Add tidy_texture_frame_set_needs_paint() to hint that the
paint can be skipped entirely; this is used when we detect
that the window shadow is entirely obscured.
http://bugzilla.gnome.org/show_bug.cgi?id=587344
If we are starting on something other than the first workspace,
meta_compositor_switch_workspace() will be called before
meta_compositor_manage_screen(); guard against that.
Previously, changes to the visibility of a window could be indicated
by meta_compositor_map_window(), meta_compositor_unminimize_window(),
meta_compositor_set_window_hidden(), etc, with the exact behavior
depending on the 'live_hidden_windows' preference.
Simplify this so that visibility is controlled by:
meta_compositor_show_window()
meta_compositor_hide_window()
With an 'effect' parameter provided to indicate the appropriate
effect (CREATE/UNMINIMIZE/MINIMIZE/DESTROY/NONE.)
The map state of the window is signalled separately by:
meta_compositor_map_window()
meta_compositor_unmap_window()
And is used only to control resource handling.
Other changes:
* The desired effect on show/hide is explicitly stored in
MetaWindow, avoiding the need for the was_minimized flag.
At idle, once we calculate the window state, we pass the
effect to the compositor if it matches the new window
state, and then clear the effect to start over for future
map state changes.
* meta_compositor_switch_workspace() is called before any windows
are hidden or shown, allowing the compositor to avoid hiding
or showing an effect for windows involved in the switch.
http://bugzilla.gnome.org/show_bug.cgi?id=582341
* Handling of post-effect cleanups for MutterWindow are
simplified - instead of trying to do different things based
on the individual needs of different effects, we just wait until
all effects complete and sync the window state to what it
should be.
* On unmap, once we destroy the pixmap, we tell ClutterX11Pixmap
that we've done so, so it can clean up and unbind. (The
unbinding doesn't seem to be working properly because of
ClutterGLXPixmap or video driver issues.)
http://bugzilla.gnome.org/show_bug.cgi?id=587251
Clean up mutter_window_effect_in_progress() by removing the
include_destroy parameter which was used only in one place that
could be easily done otherwise. (There was another use in
mutter_window_sync_actor_position() that had no point and looked
unintended.)
http://bugzilla.gnome.org/show_bug.cgi?id=587251
Add a paint function that checks all windows for repair and
shape updates; this:
- simplifies the logic for when a window needs to be repaired
- avoids duplicate work when we get multiple damage effects
- avoids the need to look ahead in the event queue
Instead of relying on repair to implicitly resize the
MutterWindow actor, set the size explicitly when the core
code updates the geometry. (This is needed because we haven't
repaired yet when we start an animation, and the animation
may depend on the size to, e.g., rescale from the center.)
Because the core geometry update happens before we start
maximize/unmaximize effects we need to work around this by
passing both the old and new geometry to the compositor.
http://bugzilla.gnome.org/show_bug.cgi?id=587251
Putting hidden windows in the desktop layer is pointless - in
the desktop layer isn't necessary below all visible windows,
and we are hiding the windows by other means. And the movement
isn't reliable because nothing sets stack->needs_relayer, so
windows can get stuck in the desktop layer after being
rehidden.
http://bugzilla.gnome.org/show_bug.cgi?id=587251
compositor.c: Move MutterWindow code to mutter-window.c;
rename map_win() to mutter_window_map(), etc.
mutter-window-private.h: New private header file for
MutterWindow functions used internally to the compositor.
compositor-mutter.h: Move MutterWindow declarations to
mutter-window.h; move a couple of private functions to
compositor-private.h
compositor-private.h: Move MetaCompScreen declaration to here:
Conceptually it's private to compositor.c, but MutterWindow
manipulates some of the lists directly for now.
mutter-plugin.c compositor.c: Don't call mutter_window_effect_completed()
for MUTTER_PLUGIN_SWITCH_WORKSPACE, but use a new
mutter_switch_workspace_completed(), since the window is
just used to identify a screen.
http://bugzilla.gnome.org/show_bug.cgi?id=587251
Separate code related to creating the gaussian-blurred shadow texture
into a separate file.
Move the definition of MetaCompositor into a compositor-private.h
so that the shadow code can cache the source in the compositor
structure.
http://bugzilla.gnome.org/show_bug.cgi?id=587251
The patch adds GLib marshalling code to Mutter, since it's required for the "workspace-switched" signal.
The definition of MetaMotionDirection enum is moved to common.h since it's now used in workspace.c.
A little cleaning is done in workspace.c:meta_workspace_activate_with_focus(), where compositor-specific code is merged with the rest of the function (required to emit signal), removing #ifdefs.
Remove a number of functions that were either entirely unimplemented
or had empty implementations for the Clutter-compositor.
meta_compositor_begin_move()
meta_compositor_update_move()
meta_compositor_end_move()
meta_compositor_set_active_window()
meta_compositor_free_window()
http://bugzilla.gnome.org/show_bug.cgi?id=581813
Now that we only have one compositor, there's no reason to access the
compositor functions through a vtable. Remove the MetaCompositor virtualization
and make the clutter code implement the meta_compositor_* functions
directly.
Move the checks for the compositor being NULL from the vtable wrappers
to the calling code (most of them were already there, so just a few
needed to be added)
Note: the compositor is actually hard-coded on at the moment and the plan
is to remove the non-composited code entirely, so the checks are
added only to keep things neat: they have no practical effect.
http://bugzilla.gnome.org/show_bug.cgi?id=581813
Mutter is a Clutter-based compositing manager. So, remove the code for
the XRender-based compositor, and make it mandatory to have XComposite,
XRender and Clutter.
Run-time support for non-composited operation is left for now.
* src/compositor/mutter/: Move files from this subdirectory into
the main compositor/ directory.
* compositor/compositor-xrender.ccompositor/compositor-xrender.h:
Remove
* include/compositor-clutter.h: Remove this stray file, it had been
replaced with compositor-mutter.h some time back.
http://bugzilla.gnome.org/show_bug.cgi?id=581813
Since the stack passed to the compositor now accurately reflects
the X stacking order, we need to treat hidden windows (which are
at the bottom of the X stacking order) specially - when the
compositor stacking order is synced, try to keep animating hidden
actors in their old positions in the stack.
http://bugzilla.gnome.org/show_bug.cgi?id=585984
With MetaStackTracker, it's no longer necessary to XQueryTree to
get a reasonably-up-to-date view of the server stacking order.
Add some comments explaining unclear aspects of
raise_window_relative_to_managed_windows() and with future possible
improvements.
http://bugzilla.gnome.org/show_bug.cgi?id=585984
Don't add override-redirect windows to MetaStack; we shouldn't
be restacking them.
Since we *aren't* stacking the override-redirect windows, we need to
be careful that to ignore them when looking for the top managed
window.
http://bugzilla.gnome.org/show_bug.cgi?id=585984
In order to properly track the stacking order for override-redirect
windows, move meta_compositor_sync_stack() call into MetaStackTracker.
In the new location, we sync the stack as a before-redraw idle function,
rather then using the freeze-thaw facilities of MetaStack. This is
simpler, and also properly compresses multiple stack changes on
notifications received from the X server.
http://bugzilla.gnome.org/show_bug.cgi?id=585984
Wedging override-redirect windows into the constraint code in stack.c
results in Mutter getting confused about the stacking order of
these windows with respect to other windows, and may also in some
cases cause Mutter to restack override-redirect windows.
core/stack-tracker.c core/stack-tracker.h: MetaStackTracker - combine
events received from the X server with local changes we have made
to come up with the best possible idea of what the stacking order
is at any one point in time.
core/screen.c core/screen-private.h: Create a MetaStackTracker for
the screen.
core/display.c: Feed relevant events to MetaStackTracker
core/frame.c core/screen.c core/stack.c: When we make changes to the
stacking order or add windows, record those changes immediatley
in MetaStackTracker so we have the information without waiting
for a round-trip.
include/ui.h ui/ui.c: meta_ui_create_frame_window add a return value
for the X request serial used to create the window.
http://bugzilla.gnome.org/show_bug.cgi?id=585984
Override-redirect windows should not be moved or resized by the
window manager.
- Mark override-redirect windows as already placed to avoid
placing them when first shown.
- Don't move-resize newly created override-redirect MetaWindow
- Don't queue a resize on override-redirect windows when reading
their WM_TRANSIENT_FOR hint.
- Add g_return_if_fail (!window->override_redirect) to catch
unexpected code paths that might result in override-redirect
windows being moved or resized.
http://bugzilla.gnome.org/show_bug.cgi?id=582639
Normally a window that is "on all workspaces", is also on a particular
workspace (to deal with being unstuck.) This is pointless for
override-redirect windows.
http://bugzilla.gnome.org/show_bug.cgi?id=582639
Don't include override-redirect windows when iterating the windows
in the screen. We don't need them for any of the current uses:
- Queueing redraws and resizes on managed windows
- Checking which windows should be added to a new workspace
http://bugzilla.gnome.org/show_bug.cgi?id=582639
Don't include override-redirect windows in the list return by
meta_display_list_windows(), since we almost never want to handle
them when considering "all window" for the display. Add a separate
meta_display_list_all_windows() that includes override-redirect
windows.
http://bugzilla.gnome.org/show_bug.cgi?id=582639
Skipping handling of properties for override redirect windows has
two advantages: first it reduces the amount of work needed to get
an override-redirect window (menu, tooltip, drag icon) onto the
screen. But more importantly, it reduces the number of code-paths
for an override-redirect to get into some code portion where it
isn't expected.
* Integrate the list of properties we load initially with the
list of property hooks; this avoids having two separate lists
that we have to keep in sync.
* Add a flag to MetaWindowPropHooks to indicate whether the
property should be handled for override-redirect windows;
currently we load a) properties that identify the window -
useful for debugging purposes b) WM_TRANSIENT_FOR (could be
used to associate menus with toplevels.)
* For properties that aren't always loaded through window-props.c,
add !window->override checks to places that trigger loading,
and add g_return_if_fail(!window->override) to the load
functions as a double-check.
http://bugzilla.gnome.org/show_bug.cgi?id=582639
Add g_return_if_fail() to check that window-management functions like
meta_window_maximize() aren't called on override-redirect windows.
This reveals that were were "unminimizing" override-redirect windows
when adding them; avoid doing that.
http://bugzilla.gnome.org/show_bug.cgi?id=582639
If a property has a reload function, but the standard property-fetching
mechanism isn't used (hooks->type == META_PROP_VALUE_INVALID), then the
a logic error (introduced in January) caused the hook to never be run.
This meant that changes to struts and icons weren't noticed.
Same as: http://bugzilla.gnome.org/show_bug.cgi?id=572573
The fix here is different in detail from that applied to Metacity, but
similar in spirit.
http://bugzilla.gnome.org/show_bug.cgi?id=585980
On subsequent changes, if there is a NET_WM_USER_TIME_WINDOW, then
read the property from that rather than from the main window.
(Fix an accidental regression: the right Window was being computed
but no longer passed in.)
http://bugzilla.gnome.org/show_bug.cgi?id=585979
Since --with-clutter and --with-introspection refer to use of external
libraries, they should in fact be --with and not --enable. Fix accidental
change that got committed as part of the big metacity => mutter change.
http://bugzilla.gnome.org/show_bug.cgi?id=586821
This reverts commit 16d49695ad.
If we want an actor to go away deterministically, we should call
clutter_actor_destroy(); using unparent instead a) may result
leaks through cyclic ref counts, and b) will delay the deallocation
of the window texture in a garbage collected environment until
the next garbage collection, causing much greater transient use
of memory.
http://bugzilla.gnome.org/show_bug.cgi?id=580520
info->pending_input_region needed to be destroyed in
mutter_set_stage_input_region() before it was set to something
else or before it was reset to be full screen (region == None).
Set the initial input area to an empty region by
setting info->pending_input_region to an empty region by default
and setting it to None or a new region if a new input region is
specified.
Remove show_overlay_window() function that was used to set an empty
input region on start-up, and uses a call to do_set_stage_input_region()
with info->pending_input_region as an argument instead.