We use to rely on grab broken events for most of the event sequence
lifetime, this breaks though on GDK_BUTTON_RELEASE/GDK_TOUCH_END, as there's
no longer a grab at that time.
For these cases (and all others where there's destroy/unrealize calls
involved during event dispatching), catch this on the late
WIDGET_REALIZED_FOR_EVENT calls on widget event handling functions.
https://bugzilla.gnome.org/show_bug.cgi?id=754098
This avoids a lot of overhead in the common case where a signal
is not connected and we're just using the class vfunc (which is true
for all in-libgtk widgets). Additionally it makes backtraces in
debuggers and profiles much much nicer to look at.
https://bugzilla.gnome.org/show_bug.cgi?id=754986
These days exposure happens only on the native windows (generally the
toplevel window) and is propagated down recursively. The expose event
is only useful for backwards compat, and in fact, for double buffered
widgets we totally ignore the event (and non-double buffering breaks
on wayland).
So, by not setting the mask we avoid emitting these events and then
later ignoring them.
We still keep it on eventbox, fixed and layout as these are used
in weird ways that want backwards compat.
There is no need to ref the windows we're ignoring, so collect and ref
only the affected child windows. Also, use a on-stack array rather
than allocating a linked list.
Also, we don't need to ref during the event emissions too, as we
already hold a ref.
https://bugzilla.gnome.org/show_bug.cgi?id=754687
It turns out that xgettext does not understand translatable="1",
so don't make gtk-builder-tool produce this, even though
GtkBuilder can parse it just fine.
https://bugzilla.gnome.org/show_bug.cgi?id=754928
This changes textview to share the style context with the pixelcache.
Doing so allows pixel cache to optimize the surface creation and use
a CAIRO_CONTENT_COLOR instead of CAIRO_CONTENT_COLOR_ALPHA when
appropriate.
https://bugzilla.gnome.org/show_bug.cgi?id=754658
We can take a fast path if the background for a widget is opaque by using
a CAIRO_CONTENT_COLOR instead of a CAIRO_CONTENT_COLOR_ALPHA surface. Most
blit'ing backends have a fast path for this, including Pixman and Quartz.
https://bugzilla.gnome.org/show_bug.cgi?id=754658
This new private API, _gtk_style_context_is_background_opaque(), is meant to
be used by internal Gtk+ wigets to optimize fast paths for cases where
applicable. One such use would be to use a CAIRO_CONTENT_COLOR surface
instead of CAIRO_CONTENT_COLOR_ALPHA.
https://bugzilla.gnome.org/show_bug.cgi?id=754658
This avoids a bunch of allocations, and additionally it has better
cache behaviour, as we don't follow pointers to the separate GList
node memory areas during traversal.
From Christian Hergert:
This machine is a Retina mac book pro so I've been working on getting
GtkTextView (GtkPixelCache) up to our performance level on
X11/Wayland. I'm seeing a jump from about 43 FPS to about 50 FPS.
https://bugzilla.gnome.org/show_bug.cgi?id=754687
Merge it into GtkWidgetPrivate. In my measurements, about half
of all widgets have a non-default auxinfo struct, and we use this
information in size allocation, so it is nice to avoid the gdata
overhead.
We only use widget paths for a few widgets nowadays (notebook,
treeview, pathbar, combobox), so we can save some space by
not having this field in GtkWidgetPrivate.
The hash table is only accessed at creation and destruction time,
and many widgets don't use templates at all, so no need to have
this permanently occupying space.
This reverts commit 3eacfa88f2.
Apart from the patch not being correct, we don't want to expose private
structures in header files if we can avoid it.
And this type-checking overhead is not an optimization that is even
measurable.
https://bugzilla.gnome.org/show_bug.cgi?id=754932
