This is the gtk3 version of !6722.

This fixes a problem that is apparent in
https://bugzilla.mozilla.org/show_bug.cgi?id=1869724, but that also
reproduces on any GTK application as described in
https://bugzilla.mozilla.org/show_bug.cgi?id=1869724#c16.

xdg_output sizes might be physical if the compositor doesn't scale them,
it seems. So to report the correct logical geometry in GDK pixels, we
need to detect this case. We do this by checking whether the wl_output
size matches the xdg_output size.

The first time this function is called, has_xdg_output() returns
true, but haven't yet received all the xdg-output events, so wait
for that to be done. Otherwise, the logical size is 0, and nothing
useful comes from that.

Use the CSS cursor names instead.

[gtk3/wayland] Fix GdkMonitor sizes on Mutter without fractional scaling.

See merge request GNOME/gtk!6723

Replace some instances of outdated cursor types

See merge request GNOME/gtk!6835

Fixes crash when handling tablet pad removal event after the
current_tablet has already been freed.

gdk/wayland: Clear current_tablet on pad leave

See merge request GNOME/gtk!6884

The version should not end with a trailing dot.

docs: Fix the Since tag for gtk_render_background()

See merge request GNOME/gtk!6906

The ModelButton update logic was unconditionally emitting them even if nothing
changed.

Fixes #6449.

a11y: Don't emit redundant checked notifications

See merge request GNOME/gtk!6923

This fixes https://bugzilla.mozilla.org/show_bug.cgi?id=1871627

All GTK3/4 apps on non-GNOME desktops aren't resizable if they are
tiled.

This patch makes the non-tiled edges resizable if we have per-edge
tiling, but not per-edge resizing information.

We might want to make all edges resizable, but that might not be fine
for tiling compositors.

[gtk3/wayland] Infer resizable edges for tiled windows if possible.

See merge request GNOME/gtk!6742

Corresponds to gdkselection-quartz changes in f30ed5dd.
Fixes https://gitlab.gnome.org/GNOME/gtk/-/issues/6480

It is at least theoretically possible that gtk_entry_get_pixel_ranges
will return no ranges, and we should handle that without an
out-of-bounds access or segfault.

dnd: Prevent a possible segfault

See merge request GNOME/gtk!6957

Finish fixing pasteboard constants macOS 10.14 and later.

See merge request GNOME/gtk!6954

The GtkFileChooserEntry widget creates a file filter instance, but never
sinks its floating reference. Newer versions of GLib correctly warn if
an instance with a floating reference gets finalized.

Fixes: #6527

Sink the floating ref on the file filter

See merge request GNOME/gtk!7011

Print backend can be disposed together with all its printers
as a reaction to user stopping enumeration of printers.
Adding a weak pointer help us to detect that the backend
was disposed and hence the backend and its printers should not
be used anymore.

Fixes #6265

Cursor themes have recently started to reduce their coverage of
'legacy' cursor names, and reduced to the standard names. Support
this for the few cursor types that are still used in GTK.

Nobody is going to track these allocation failures down in gtk3
at this point, so drop the alarming message. The failures are
pretty harmless.

wayland: Keep cursor types limping along

See merge request GNOME/gtk!7033

box gadget: Drop a critical

See merge request GNOME/gtk!7034

As fancy as property paths are, recursive resolution of files
to a location increases the big O complexity enough that it's
not a great option on large homedirs with many indexed files.

Ensure the files are from the right location through a URI
prefix match, which does hits an index. This may dramatically
improve performance on large indexed trees.

Testing this query in an isolated testcase with a total
1434099 indexed files shows that it can run more than 1500 times
per second in this computer (an average of 15200 queries in
several 10 second runs), which presumably is a tad faster than
anyone can type.

Closes: https://gitlab.gnome.org/GNOME/gtk/-/issues/4133

When starting a search over a very populated filesystem, it
is possible that typing the first chars will return a too
high number of results. Even though iterating through the
cursor is in itself very fast, extracting the GIO information
from those many files at once is not going to be as fast.

In order to increase interactivity (i.e. not make things
possibly sluggish) iterate the cursor in an idle function
and add search results to the filechooser model little by little.

If the user keeps typing (as it is likely will happen), there
will be better chances to cancel and proceed to the next
query timely. If not, the results will be there soon enough.

The search provider should make it sure there are some
specific GFileInfo fields set. Fix the mimetype extraction
from the query, and use that to fill in the missing gaps
the best we can.

Again on massive filesystems, the very first character
is likely to bring a likewise massive amount of search
results that we need to maybe query info for, then create
icons and widgets for. While it's impressive we can do
that, it's also expensive and likely pointless, for the
first character.

Typing a second character is however very likely to
considerably reduce the amount of items to categorize and
show. So start actually searching from there.

Testing on a filesystem with 1434099 files indexed, trying 5
semi-random 1 character searches (n, h, t, i, o) returns on
average 168K items (min. 78771, max. 331471), trying 5
semi-random 2 character searches (no, he, th, in, on)
returns on average 34K items (min. 11133, max. 94961),
which is a more approachable set.

Doing this is enough that typing on a filechooser search
entry feels completely fluid.

Provide the hits in an order that is more easily consumed by the
filechooser, and less jarring when populating.

Backport tracker3 performance improvements

See merge request GNOME/gtk!7051

This commit was identified as the cause for a GLArea transparency
regression. Bring the required code back.

Fixes: #6359

Revert 7237f5d0

See merge request GNOME/gtk!7079

We need to go back to using this, since move is widely abused
and thus can't change its image to one thats appropriate for dnd.

Use the same cursors during dnd as we do in GTK 4.

wayland: Guarantee dnd-move as a cursor name

See merge request GNOME/gtk!7102

It's mandatory to reply to a xdg_surface.configure with a
xdg_surface.ack_configure, as well as a wl_surface.commit to have the
ack_configure to have any effect. We failed to do that if the new
configuration didn't cause any actual anything to change, meaning we'd
just ack_configure, but potentially never commit.

Fix this by requesting a 'after-paint' frame clock phase, while setting
the pending commit state to true. If we were already to paint, the new
requested phase would happen anyway, but to avoid double
wl_surface.commit when also using eglSwapBuffers*(), also avoid
performing the commit if we know the swap buffers will happen.

wayland: Always commit soon after xdg_surface.ack_configure

See merge request GNOME/gtk!7139

We may get invalid UTF-16 sequences when third party
apps or hooks inject WM_CHAR messages.

Validate the unicode string before committing.

Fixes https://gitlab.com/inkscape/inkscape/-/issues/2088

GtkIMContextIME: Avoid committing NULL string

See merge request GNOME/gtk!7167

One of the confusing aspects of GOBject. When we say
'an uninitialized GValue', we mean one that is declared
via GValue v = G_VALUE_INIT;

Fixes: #6655

css: Avoid an uninitialized GValue

See merge request GNOME/gtk!7177

I'm unable to compile gtk+-3 on Fedora 40 with profiler support
"-Dprofiler=true" because it looks for sysprof-capture-3 while
Fedora 40 only provides sysprof-capture-4.

GTK4 for the same build option only looks for sysprof-capture-4.

Build option '-Dprofiler=true' fails with latest sysprof-capture

See merge request GNOME/gtk!7225

Port `gail_focus_notify` from using the deprecated
`atk_focus_tracker_notify` to notify about the new
focus object. Notify of a state of the focused state
instead, as suggested in the
`atk_focus_tracker_notify` doc [1].

Using the deprecated function e.g. resulted in
the Orca screen reader no longer announcing
initial focus on editable comboboxes in
LibreOffice Writer's toolbars after Orca dropped
support for the deprecated "focus" AT-SPI event
in this commit [2]:

    commit 9e2902dd46c7e583a097e235dfd7e3c50b016383
    Author: Joanmarie Diggs <jdiggs@igalia.com>
    Date:   Wed May 1 12:14:07 2024 +0200

        Remove the on_focus handler for the LibreOffice script

        The "focus:" event was deprecated many years ago. If the expected
        "object:state-changed:focused" event is absent, that bug should
        be fixed.

This commit makes the announcement work again.

It also fixes the scenario described in [3].

[1] https://docs.gtk.org/atk/func.focus_tracker_notify.html
[2] https://gitlab.gnome.org/GNOME/orca/-/commit/9e2902dd46c7e583a097e235dfd7e3c50b016383
[3] https://gitlab.gnome.org/GNOME/gtk/-/issues/454#note_2099344

Fixes: #454

a11y: Port from deprecated atk_focus_tracker_notify

See merge request GNOME/gtk!7239

Closes: https://gitlab.gnome.org/GNOME/gtk/-/issues/6704

quartz: Remove trackingArea double reference

See merge request GNOME/gtk!7250