React to the session-state in StateChanged signals from
the inhibit portal and call QueryEndResponse as appropriate.

list image buttons

See merge request GNOME/gtk!656

Quartz GLArea

See merge request GNOME/gtk!146

The previous version of this patch sent an update message to the
NSOpenGLContext in a GdkGLContext::update vfunc, but that vfunc does not
exist any more.

See: #517

Current problems:
* other widgets in a GL-painted window are low-resolution on Retina
  display
* something wrong with paint updates; gdkgears demo only updates every
  couple of seconds but reports ~30fps

See: #517

Use GLSL 3.30 shaders

See merge request GNOME/gtk!671

We currently ask for anything above 3.2 GL contexts, but we're still
using GLSL 1.50 shaders all over the place. If a GL driver supports GL
3.2+ and GLSL 1.50 only then we'd be in trouble, but the chances of that
happening are really small.

Fix the introspection build with "-Wl,--as-needed -flto -O2". Fixes #1692

See merge request GNOME/gtk!665

They're either wrong (when using FALSE because the widgets don't account
for changes to the CSS) or unnecessary (when using TRUE because it's the
default).

Fixes !1777

Thats a valid way to end color picking.

- based on Tobias' and Allan's feedback, the app icon dropshadow
  were too prominent.

The above flags in combination with "-fvisibility=hidden" break the
g-i build because it results in the g-i generated dumper executable not
linking against the libraries because they are detected as unused and
thrown out.

Fix by only using -fvisibility=hidden for the library and not g-i.

Inspector: Find themes the same way GtkCSSProvider does

See merge request GNOME/gtk!663

Look for subdirectories named "gtk-3.x", where 'x' starts as current
minor version and counts down to 14, then drops to 0.
Only look for gtk.css in these directories though. If a theme only
provides gtk-dark.css, it won't be found.

GDK W32: Test for IME correctly

See merge request GNOME/gtk!658

The strings that are set are static, so g_value_set_static_string() should
be used to set them instead of g_value_set_string().

ImmIsIME() doesn't work (always returns TRUE) since Vista.
Use ITfActiveLanguageProfileNotifySink to detect TSF changes,
which are equal to IME changes for us.

Also make sure that IMMultiContext re-loads the IM when keyboard layout
changes, otherwise there's a subtle bug that could happen:
* Run GTK application with non-IME layout (US, for example)
* Focus on an editable widget (GtkEntry, for example)
* IM Context is initialized to use the simple IM
* Switch to an IME layout (such as Korean)
* Start typing
* Since IME module is not loaded yet, keypresses are handled
  by a default MS IME handler
* Once IME commits a character, GDK will get a WM_KEYDOWN,
  which will trigger a GdkKeyEvent, which will be handled by
  an event filter in IM Context, which will finally re-evaluate
  its status and load IME, and only after that GTK will get
  to handle IME by itself - but by that point input would
  already be broken.
To avoid this we can emit a dummy event (with Void keyval),
which will cause IM Context to load the appropriate module
immediately.

1. NSRect and CGRect aren't toll-free bridged in 32 bit builds.
2. NSWindowCollectionBehavior(Dis)AllowTiling is defined beginning in 10.11.

Adwaita: checked buttons workaround

See merge request GNOME/gtk!657

Adwaita: Adjust switch margins on headerbars

See merge request GNOME/gtk!650

- backport from master

fixes https://gitlab.gnome.org/GNOME/gtk/issues/1756

- Rather than making labeled buttons inside lists toned down,
  only tone down image buttons, so that we don't have to include
  icon-only actions. Places like Software can continue using label
  buttons with no change without having those less visible.

Fixes https://gitlab.gnome.org/GNOME/gtk/issues/1748

Fixes https://gitlab.gnome.org/GNOME/gtk/issues/1642 once and for all

(until it resurfaces)

Closes https://gitlab.gnome.org/GNOME/gtk/issues/1759

The problem here was that NSPasteboard would release the clipboard
owner if all data items were transferred. When trying to re-use this
owner at a later point, GTK+ would attempt a retain call on a released
object and crash.

Fix this by not immediately releasing the owner after declaring types,
so by keeping our own reference around.

Closes https://gitlab.gnome.org/GNOME/gtk/issues/529

Gdk sometimes misses crossing events on popups, so the cached toplevel
may be NULL. If it is, find the toplevel under the pointer and set it.

Fixes https://gitlab.gnome.org/GNOME/gtk/issues/623

Startup notification fixes

See merge request GNOME/gtk!647

The use of the startup ID is now twofold, we reply back with it to end any
corresponding startup notification, but we also use it on
gtk_surface1.request_focus to acknowledge that the activation might raise
the corresponding window.

We should preserve the startup ID for the second to work properly, so avoid
clearing it here. It is inconsequential if the underlying
gtk_shell1.set_startup_id request happens multiple times on no longer existing
startup IDs, so don't bother preventing that from happening.

Closes: https://gitlab.gnome.org/GNOME/gtk/issues/1754

It might be too late to do it at GtkApplication::add_platform_data time,
since the envvar may be consumed earlier on if gdk_display_open() happened
to be called before (eg. through gtk_get_option_group(TRUE)).

Stash the envvar in a constructor function, so its ensured to happen before
it can get consumed.

Closes: https://gitlab.gnome.org/GNOME/gtk/issues/1754

gdk: Include copy of glib/gconstructor.h

See merge request GNOME/gtk!643

This is named gdkconstructor.h to avoid any possible conflicts. This fixes
the current usages of G_HAS_CONSTRUCTORS, as that header is not installed
by glib.

Closes: https://gitlab.gnome.org/GNOME/gtk/issues/1029

Should fix most if not all other cases where system-caused changes to
the NSWindow result in the Gdk coordinates not mapping correctly to the
AppKit coordinates.

Use the same icon we use in master now.