And make the GdkCairoContext as abstract.
The idea of this and thje following commits is to get rid of all
Cairo code in gdksurface.c (and $backend/gdksurface-$backend.c)
by moving that code into the Cairo context files.
In particular, the GdkSurfaceClass.begin_frame/end_frame()
functions (which are currently exclusively used by the Cairo code
should end up being moved to GdkDrawContextClass.begin/end_frame().
This has multiple benefits:
1. It unifies code between the different drawing contexts.
GL lives in GLContext, Vulkan in VulkanContext and Cairo in
CairoContext. In turn, this makes it way easier to reason about
what's going on in surface-specific code. Currently pretty much
all backends do things wrong when they want to sync to drawing
or to the frame clock.
2. It makes the API of GdkSurface smaller. No drawing code (apart
from creating the contexts) needs to remain.
3. It confines Cairo to the Drawcontext, thereby making it way
more obvious when backends are still using it in situations
where it may now conflict with OpenGL (like when doing the dnd
failed animation or in the APIs that I'm removing in this
branch).
4. We have 2 very different types of Cairo contexts: The X/win32
model, where we have a natively supported Cairo backend but do
double buffering ourselves and use similar surfaces and the
Wayland/Broadway model where we use image surfaces without any
Cairo backend support and have to submit the buffers manually.
By not sharing code between those 2 versions, we can make the
actual code way smaller. We also get around the need to create
1x1 image surfaces in the Wayland backend where we pretend
there's a native Cairo surface.
This does nothing but disallow passing NULL to gdk_surface_begin_paint()
and instead require this context.
The ultimate goal is to split out Cairo drawing into its own source file
so it doesn't clutter up the generic rendering path.
The shortcuts inhibitors hash table is created when we create a
GdkWaylandWindow implementation for a GdkWindow, and it's destroyed once
we finalize the instance. The fake "root" window we create for the
Wayland display does not have a backing native window, so the shortcuts
inhibitors hash table is set to NULL; this causes a critical error
message when calling g_hash_table_destroy() on it. The finalization of
the root window happens when we close a display connection.
We should use g_clear_pointer(), instead, as it's NULL safe.
Without this change, the displayclose test fails, as all warnings are
considered fatal.
We can't freeze the frame clock on commit, but only after-paint,
otherwise the frameclock will resume in the paint stage.
So freeze the frame clock at the end of the frame if we are waiting for
a frame callback.
Note; The diff is only lage because of indentation changes due to
avoiding early returns in favor of a branch.
Instead of going through an ancillary script to strip away the
`WL_EXPORT` annotation from the generated code, we should bump up the
required version of Wayland, and use the `private-code` argument for
wayland-scanner, which does the right thing for us.
* Previous commit had misleading info. The code was
added to begin_paint() instead of end_paint(). Though
that did not affect its performance in any visible way.
* Company advised to move the code to an "after_paint" signal
handler, so that it works on all renderers, not just Cairo.
This change caused high fluctuation in FPS values in fishbowl
when it is put in a situation where it cannot achieve 60fps
(such as using Cairo renderer at ultra-high resolution).
This seems to be deliberate and not a bug.
There is no easily apparent way of being notified when frame updates
happene exactly, so we just query frame info at the end of each paint.
If we query too often (faster than DWM refresh rate), we just get
the same values twice in a row, but that is, hopefully, highly unlikely.
When asked for a nonexistent (positive) monitor number,
gdk_x11_display_get_monitor would (at best) return an uninitialized pointer,
instead of returning NULL.
That way, we can store the right region there: The actual painted area
instead of the exposed area (which is way too small).
Also, the GL context is the only user of this data, so storing it there
seems way smarter.
This error message is printed into the journal if a GTK app can't
connect to eithre Wayland or X11. Make it at least mention who is not
capable of connecting to a server.
Unrelated, we might want to improve our error reporting when a GTK app
can't start, so that debugging issues with system startup / login get
easier to resolve.
Instead of calling gdk_surface_invalidate_region(), just
gdk_surface_queue_expose() and rely on the renderer computing the diff
from the previous rendering.
This ensures that the frame clock gets updated with correct presentation
times even if nothing was drawn.
This is necessary for benchmarking but would also be relevant for videos
that want to sync to the frame clock but draw frames a lot less.
This commit ensures that each GdkSurface impl remembers the
cursor that GDK sets for it, and that this cursor is set
each time WM_SETCURSOR is called for that sufrace's HWND.
This is needed because W32, unlike X, has no per-window cursors -
the cursor on W32 is a global resource, and we need to keep track
of which cursor should be set when pointer is over which surface
ourselves (WM_SETCURSOR exists exactly for this reason).
This commit also makes GDK remember the surface that has an implicit
grab (since implicit grabs are gone from the upper levels of the toolkit),
and ensures that crossing events are correctly synthesized and the grab
is broken when surface focus changes. This fixes a bug where opening
a new window (by clicking something in some other, pre-existing window)
will make that new window not get any mouse input due to the fact
that the mouse-button-down event from that click caused an implicit
grab on the pre-existing window, and that grab was not released afterward.
Instead of now-unused GdkWin32Cursor class (a subclass of GdkCursor),
add a stand-alone GdkWin32HCursor class that is a wrapper around
HCURSOR handle.
On creation it's given a display instance, a HCURSOR handle and a boolean
that indicates whether the HCURSOR handle can or cannot be destroyed
(this depends on how the handle was obtained).
That information is stored in a hash table inside the GdkWin32Display
singleton, each entry of that table has reference count.
When the GdkWin32HCursor object is finalized, it reduces the reference
count on the table entry in the GdkWin32Display. When it's created,
it either adds such an entry or refs an existing one.
This way two pieces of code (or the same piece of code called
multiple times) that independently obtain the same HCURSOR from the OS
will get to different GdkWin32HCursor instances, but GdkWin32Display
will know that both use the same handle.
Once the reference count reaches 0 on the table entry, it is freed
and the handle (if destroyable) is put on the destruction list,
and an idle destruction function is queued.
If the same handle is once again registered for use before the
idle destructior is invoked (this happens, for example, when
an old cursor is destroyed and then replaced with a new one),
the handle gets removed from the destruction list.
The destructor just calls DestroyCursor() on each handle, calling
SetCursor(NULL) before doing that when the handle is in use.
This ensures that SetCursor(NULL) (which will cause cursor to disappear,
which is bad by itself, and which will also cause flickering if the
cursor is set to a non-NULL again shortly afterward)
is almost never called, unless GTK messes up and keeps using a cursor
beyond its lifetime.
This scheme also ensures that non-destructable cursors are not destroyed.
It's also possible to call _gdk_win32_display_hcursor_ref()
and _gdk_win32_display_hcursor_unref() manually instead of creating
GdkWin32HCursor objects, but that is not recommended.
Add a new W32 backend-specific message filtering mechanism.
Works roughly the same way old event filtering did, but without
events (events are GDK/X11 concept that never really made sense
on W32), so there's no functionality for 'altering' events being
emitted. If an event needs to be emitted in response to a message
do it yourself.
Implemented like this, it should give better performance than
if we were to use GLib signals for this, since W32 sends a LOT
of messages (unlike X11, which doesn't send events as often)
all the time, and invoking the signal machinery on *each* message
would probably be bad.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
Rename GdkWin32Selection to GdkWin32Clipdrop, since GdkSelection
is mostly gone, and the word "selection" does not reflect the
functionality of this object too well.
Clipboard is now handled by a separate thread, most of the code for
it now lives in gdkclipdrop-win32.c, gdkclipboard-win32.c just uses
clipdrop as a backend.
The DnD source part is also put into a thread.
The DnD target part does not spin the main loop, it just
emits a GDK event and returns a default value if it doesn't get a reply
by the time the event is processed.
Both clipboard and DnD use a new GOutputStream subclass to get data
from GTK and put it into a HGLOBAL.
GdkWin32DragContext is split into GdkWin32DragContext and GdkWin32DropContext,
anticipating a similar change that slated to happen to GdkDragContext.
OLE2 DnD protocol is now used by default, set GDK_WIN32_OLE2_DND envvar to 0
to make GDK use the old LOCAL and DROPFILES protocols.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
With the previous approach we would spend most of the time waiting for
the swapchain to be filled again because it seems the compositor takes
care of 2 images at once from time to time.
This is not visible in profiles because waiting for a frame is a
read/poll/whatever operation that does not take CPU. It's only
noticeable because the app becomes less responsive.
If G_ENABLE_CONSISTENCY_CHECKS is defined (i.e. if our buildtype is
'debug'), add a opengl debug callback that prints all debug messages
with a severity higher than SEVERITY_NOTIFICATION as a warning to the
console.
Set the display for each event that we put.
Also reorganize the dnd_event_put() function a bit, giving it a surface
directly instead of setting it by implication.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
dest_surface is going to always be NULL for source contexts.
Previously we used to put the root window there to pass this check,
but root windows are gone (and root surfaces never existed to begin
with), so we have to adapt.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
This affects gdk_device_query_state() for the virtual device. It has
no window, and is forced to query the display itself, and display
defaults its scale to 1 even for HiDPI desktops. Use the same
"query scale of a NULL monitor" trick that we use in other places
to get the global desktop scale.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
A side effect of vkQueuePresentKHR is the Vulkan implementation calling
wl_surface_commit() on the corresponding Wayland surface. Thus, before
this, we must synchronize the surface state (e.g. opaque region, window
geometry, etc) that changed since last time. Prior to this commit this
was done after calling vkQueuePresentKHR(), causing the surface state to
always correspond to the previous buffer state. As of this commit this
is now done before calling vkQueuePresentKHR(), thus before
wl_surface_commit().
A side effect of eglSwapBuffers* is the EGL implementation calling
wl_surface_commit() on the corresponding Wayland surface. Thus, before
swapping buffers, we must synchronize the surface state (e.g. opaque
region, window geometry, etc) that changed since the last buffer swap.
Prior to this commit, this was done after eglSwapBuffers*, causing the
surface state to always correspond to the previous buffer state. As of
this commit this is now done before swapping the buffers, thus before
wl_surface_commit().
If you want transparent region, you can just render them transparent.
If you want input shaping, use gdk_surface_input_shape_combine_region().
Also remove gtk_widget_shape_combine_region().
... and its implementation in the X11 backend.
GDK does lots of work trying to reduce the region in expose events
so that when the server sends multiple expose events, touching the
same area we can make sure to only redraw stuff once. However:
(1) this is only relevant of there's tons of delay and multiple
expose events get sent
(2) we coalesce multiple events into a single expose event anyway
(3) we do this on the frame clock
But most importantly:
(4) Since the invention of compositing, servers caches all contents
anyway
If a surface is unmapped by the client while gdk is processing updates,
(for example Firefox un-mapping its window on Expose events), the
windowing backend resources might be lost (for example with Wayland)
which can cause a crash in end_paint().
Make sure we drop the cairo surfaces as well when hiding the surface,
that will avoid the crash in gdk_surface_impl_wayland_end_paint() when
trying to attach the staging cairo surface to a released wl_surface,
these will be recreated when needed when the surface becomes visible
again and there is no need to keep such buffers around for a surface
which is not visible anyway.
https://bugzilla.gnome.org/show_bug.cgi?id=793062
This is an automated change doing these command:
git sed -f g gtk_widget_set_has_window gtk_widget_set_has_surface
git sed -f g gtk_widget_get_has_window gtk_widget_get_has_surface
git sed -f g gtk_widget_set_parent_window gtk_widget_set_parent_surface
git sed -f g gtk_widget_get_parent_window gtk_widget_get_parent_surface
git sed -f g gtk_widget_set_window gtk_widget_set_surface
git sed -f g gtk_widget_get_window gtk_widget_get_surface
git sed -f g gtk_widget_register_window gtk_widget_register_surface
git sed -f g gtk_widget_unregister_window gtk_widget_unregister_surface
git checkout NEWS*
This is an automatic rename of various things related
to the window->surface rename.
Public symbols changed by this is:
GDK_MODE_WINDOW
gdk_device_get_window_at_position
gdk_device_get_window_at_position_double
gdk_device_get_last_event_window
gdk_display_get_monitor_at_window
gdk_drag_context_get_source_window
gdk_drag_context_get_dest_window
gdk_drag_context_get_drag_window
gdk_draw_context_get_window
gdk_drawing_context_get_window
gdk_gl_context_get_window
gdk_synthesize_window_state
gdk_surface_get_window_type
gdk_x11_display_set_window_scale
gsk_renderer_new_for_window
gsk_renderer_get_window
gtk_text_view_buffer_to_window_coords
gtk_tree_view_convert_widget_to_bin_window_coords
gtk_tree_view_convert_tree_to_bin_window_coords
The commands that generated this are:
git sed -f g "GDK window" "GDK surface"
git sed -f g window_impl surface_impl
(cd gdk; git sed -f g impl_window impl_surface)
git sed -f g WINDOW_IMPL SURFACE_IMPL
git sed -f g GDK_MODE_WINDOW GDK_MODE_SURFACE
git sed -f g gdk_draw_context_get_window gdk_draw_context_get_surface
git sed -f g gdk_drawing_context_get_window gdk_drawing_context_get_surface
git sed -f g gdk_gl_context_get_window gdk_gl_context_get_surface
git sed -f g gsk_renderer_get_window gsk_renderer_get_surface
git sed -f g gsk_renderer_new_for_window gsk_renderer_new_for_surface
(cd gdk; git sed -f g window_type surface_type)
git sed -f g gdk_surface_get_window_type gdk_surface_get_surface_type
git sed -f g window_at_position surface_at_position
git sed -f g event_window event_surface
git sed -f g window_coord surface_coord
git sed -f g window_state surface_state
git sed -f g window_cursor surface_cursor
git sed -f g window_scale surface_scale
git sed -f g window_events surface_events
git sed -f g monitor_at_window monitor_at_surface
git sed -f g window_under_pointer surface_under_pointer
(cd gdk; git sed -f g for_window for_surface)
git sed -f g window_anchor surface_anchor
git sed -f g WINDOW_IS_TOPLEVEL SURFACE_IS_TOPLEVEL
git sed -f g native_window native_surface
git sed -f g source_window source_surface
git sed -f g dest_window dest_surface
git sed -f g drag_window drag_surface
git sed -f g input_window input_surface
git checkout NEWS* po-properties po docs/reference/gtk/migrating-3to4.xml
Rename all *window.[ch] source files.
This is an automatic operation, done by the following commands:
for i in $(git ls-files gdk | grep window); do
git mv $i $(echo $i | sed s/window/surface/);
git sed -f g $(basename $i) $(basename $i | sed s/window/surface/) ;
done
git checkout NEWS* po-properties po
This renames the GdkWindow class and related classes (impl, backend
subclasses) to surface. Additionally it renames related types:
GdkWindowAttr, GdkWindowPaint, GdkWindowWindowClass, GdkWindowType,
GdkWindowTypeHint, GdkWindowHints, GdkWindowState, GdkWindowEdge
This is an automatic conversion using the below commands:
git sed -f g GdkWindowWindowClass GdkSurfaceSurfaceClass
git sed -f g GdkWindow GdkSurface
git sed -f g "gdk_window\([ _\(\),;]\|$\)" "gdk_surface\1" # Avoid hitting gdk_windowing
git sed -f g "GDK_WINDOW\([ _\(]\|$\)" "GDK_SURFACE\1" # Avoid hitting GDK_WINDOWING
git sed "GDK_\([A-Z]*\)IS_WINDOW\([_ (]\|$\)" "GDK_\1IS_SURFACE\2"
git sed GDK_TYPE_WINDOW GDK_TYPE_SURFACE
git sed -f g GdkPointerWindowInfo GdkPointerSurfaceInfo
git sed -f g "BROADWAY_WINDOW" "BROADWAY_SURFACE"
git sed -f g "broadway_window" "broadway_surface"
git sed -f g "BroadwayWindow" "BroadwaySurface"
git sed -f g "WAYLAND_WINDOW" "WAYLAND_SURFACE"
git sed -f g "wayland_window" "wayland_surface"
git sed -f g "WaylandWindow" "WaylandSurface"
git sed -f g "X11_WINDOW" "X11_SURFACE"
git sed -f g "x11_window" "x11_surface"
git sed -f g "X11Window" "X11Surface"
git sed -f g "WIN32_WINDOW" "WIN32_SURFACE"
git sed -f g "win32_window" "win32_surface"
git sed -f g "Win32Window" "Win32Surface"
git sed -f g "QUARTZ_WINDOW" "QUARTZ_SURFACE"
git sed -f g "quartz_window" "quartz_surface"
git sed -f g "QuartzWindow" "QuartzSurface"
git checkout NEWS* po-properties
is_touchpad_device() for XI2 was hardcoded to look for libinput only.
Extend it slightly to correctly identify other Xorg touchpad drivers.
https://gitlab.gnome.org/GNOME/gtk/issues/97
This is necessary so that bidnings work properly and don't make
gdk_gl_texture_release() a function on GdkTexture.
It also allows code to identify what type of texture they are dealing
with.
Finally, we can now decide to add getters later without screwing
anything up, if we want to allow people to access GL textures directly.
This is a constructor, and it needs a transfer annotation.
Sadly, the resulting introspection representation is going to be a less
than satisfactory `Gdk.gl_texture_new()`, because there is no such thing
as a GdkGLTexture in the public API.
A problem with textures is that they can become too big for GPU memory,
which would require tiling. But for tiling we only need to download
the pixels needed by the tile.
Similarly, there might be interest to not upload full textures if a
renderer knows it only needs a small part.
Both of these methods require the ability to specify an area of the
texture to be downloaded. So change the download vfunc to include
this parameter now before we add even more textures later.
A private gdk_texture_download_area() function has also been added, but
nobody is using it yet.
GdkMemoryTexture is a texture implementation for holding data in memory
(read: GBytes). You specify the GdkMemoryFormat that data is in and off
you go.
Renderers can use this to add uploads in various different formats and
don't need to fallback to GDK doing the conersion on the CPU.
Supported formats can be extended if we need new ones, for now I just
added the relevant ones for Cairo and GdkPixbuf.
The constructor is also private still, because I'm not sure we want to
export GdkMemoryFormat.
Wrappers that do from_cairo_surface() and for_pixbuf() do exist though.
Put GdkGLTexture into its own file and rename the API to
gdk_gl_texture_foo() instead of gdk_texture_foo_for_gl().
Apart from naming, no actual code changes.
The header got included without config.h being included first which resulted in the
wrong _GDK_EXTERN macro being used. As a result some symbols weren't exported
and starting a DnD action would crash in the linker.
This patch adds config.h includes in all places where clang complained about
_GDK_EXTERN redefinitions.
See #32 for more info.
The Wayland backend was already not supporting this setting
since it is an XSetting that is not backed by a GSetting.
Drop this setting altogether, since we will stop supporting
general-purpose modules.
The internal known_globals hashtable is used to carry accounting for
interfaces that depend on others (as ordering is not guaranteed), extend
its usage so it also keeps track of unimplemented interfaces (here at
least).
The API call will then use this to allow querying the globals offered by
the compositor, it will be useful to determine whether we can use
text-input protocols or should fallback to other IMs.
Drop the public filtering API. The x11 backend already has
the ::xevent signal as replacement. The win32 backend needs
a similar signal to replace filtering.
Reshuffle header inclusions in the x11 backend a little bit
to avoid a cyclic inclusion between gdkprivate-x11.h and
gdkdisplay-x11.h that is otherwise causing problems.
Remove all the old 2.x and 3.x version annotations.
GTK+ 4 is a new start, and from the perspective of a
GTK+ 4 developer all these APIs have been around since
the beginning.
The standard Vulkan SDK ships with a pkg-config file, like a modern
library should.
We should fall back to finding the library and header only for platforms
where pkg-config is not really a thing.
Based on a patch by: Daniel Stone <daniels@collabora.com>
https://bugzilla.gnome.org/show_bug.cgi?id=793181
The GDK_POINTER_MOTION_HINT_MASK enumeration value is gone, but we're
still keeping around the "is_hint" field in GdkEventMotion, even though
every backend sets it to `false` — except for the core X11 device
manager.
GdkContentFormatsBuilder is currently not introspectable, as it does not
have a GType. We can turn it into a boxed type, but we need to implement
memory management for it.
The current gdk_content_formats_builder_free() function returns a newly
constructed value, so we cannot use it as a GBoxedFreeFunc; additionally
copying a GdkContentFormatsBuilder contents would make it a bit odd, as
you could get multiple identical GdkContentFormats out of the copies.
A simple approach is to model the GdkContentFormatsBuilder API to follow
the GBytes one: use reference counting for memory management, and have
a function to release a reference, return a GdkContentFormats, and reset
the GdkContentFormatsBuilder state.
For language bindings, we can provide a get_formats() function that
returns the GdkContentFormats instance and resets the builder instance,
leaving the reference count untouched.
For C convenience we can keep gdk_content_formats_builder_free(), and
make it a wrapper around gdk_content_formats_builder_get_formats(), with
the guarantee that it'll free the builder instance regardless of its
current reference count.
https://bugzilla.gnome.org/show_bug.cgi?id=793097https://blogs.gnome.org/otte/2018/02/03/builders/
GDK has a lock to mark critical sections inside the backends.
Additionally, code that would re-enter into the GTK main loop was
supposed to hold the lock.
Back in the Good Old Days™ this was guaranteed to kind of work only on
the X11 backend, and would cause a neat explosion on any other GDK
backend.
During GTK+ 3.x we deprecated the API to enter and leave the critical
sections, and now we can remove all the internal uses of the lock, since
external API that uses GTK+ 4.x won't be able to hold the GDK lock.
https://bugzilla.gnome.org/show_bug.cgi?id=793124
The main GDK thread lock is not portable and deprecated.
The only reason why gdk_threads_add_timeout() and
gdk_threads_add_timeout_full() exist is to allow invoking a callback
with the GDK lock held, in case 3rd party libraries still use the
deprecated gdk_threads_enter()/gdk_threads_leave() API.
Since we're removing the GDK lock, and we're releasing a new major API,
such code cannot exist any more; this means we can use the GLib API for
installing timeout callbacks.
https://bugzilla.gnome.org/show_bug.cgi?id=793124
The main GDK thread lock is not portable and deprecated.
The only reason why gdk_threads_add_idle() and
gdk_threads_add_idle_full() exist is to allow invoking a callback with
the GDK lock held, in case 3rd party libraries still use the deprecated
gdk_threads_enter()/gdk_threads_leave() API.
Since we're removing the GDK lock, and we're releasing a new major API,
such code cannot exist any more; this means we can use the GLib API for
installing idle callbacks.
https://bugzilla.gnome.org/show_bug.cgi?id=793124
BTN_STYLUS3 is defined by the Linux 4.15 kernel and is sent when the
third button on a stylus is pressed. At the moment, only Wacom's "Pro
Pen 3D" has three stylus buttons. Pressing this button triggers a button
8 event to be sent under X11, so we use the same mapping here.
https://bugzilla.gnome.org/show_bug.cgi?id=790033
We have a couple of Python 3.x scripts that parse C files, and since C
does not have any encoding, we need to force one ourselves, to avoid the
case when we're running the build in a non-UTF-8 locale.
https://bugzilla.gnome.org/show_bug.cgi?id=792497
This state flag is used in several places in GTK+, for example to
ignore RESIZE_INC hints if tiled. Setting it is also necessary for
backwards compatibility with applications that changed their behaviour
when tiled, such as GNOME Terminal and its MATE fork.
Signed-off-by: Simon McVittie <smcv@debian.org>
https://bugzilla.gnome.org/show_bug.cgi?id=789357
Commit c415bef5de introduced support for the new _GTK_EDGE_CONSTRAINTS
atom. If the compositor supports that atom, however, we were always
setting the tiled state, even if no actual tiling information is
available, where the correct action is to completely remove any traces
of the tiled state.
Fix that by correctly removing the tiled state when compositor supports
_GTK_EDGE_CONSTRAINTS Xatom.
https://bugzilla.gnome.org/show_bug.cgi?id=788516
The inspector may hold on to render nodes and textures
beyond the lifetime of the widget (and thus the GL
resources). To handle this situation, allow the widget
to explicitly release the GL resources, and make
the texture available on the clent-side as a cairo
surface. This lets the recorder still show the content
after the widget is gone.
The query function for cursor sizes and capabilities
are not very interesting. At least, they are not used
in GTK+, and all backends but X11 just hardcode
made-up values anyway. So, lets drop them.
As far as possible, use per-display debug flags.
This will minimize the debug spew that we get from
the inspector if it is running on a separate display.
We should be smarter in picking a good device eventually,
but for now, we just allow to explicitly choose one. To
see a list of all devices, use GDK_VULKAN_DEVICE=list
To specify which device to use, use GDK_VULKAN_DEVICE=<number>
Instead, pass the source window to gdk_drag_begin().
Also make Wayland use this window instead of the one under the pointer
(though those 2 Windows are most likely the same anyway).
