The intention of the ui file was to not let
the paned shrink both children down to nothing,
but using <child> for the children effectively
overrides the setting of the shrink properties.
Fix that by using child properties instead of
<child>.
The previous code would include CSS padding/margin/border in the
measurement and that is wrong.
Until commit a96c75ff02 this was not actually visible, but afterwards
listitems were allocated 16px too wide.
Test included
Put all the function checks in one place.
Remove functions we don't actually use,
and add ones that we have #ifdefs for in
in the code. Also add enough includes to
make these checks actually work.
Fixes: #5070
The availability of wl_surface.offset depends on the compositor, so we
can't call it unconditionally. Add a version check to so we only call
offset if we know we won't raise a protocol error.
Fixes: 0eb791eaaa ("Make mask nodes more versatile")
When looking at `gtk_widget_class_install_action()`, it isn't
immediately obvious how actions can be enabled or disabled. Add a
reference to the `gtk_widget_action_set_enabled()` method so people have
a quick link to that.
We don't use it most of the time and it's also confusing to new contributors who think it's important to fix it.
Keep it for manual runs, so that interested people can just click a button to see the results. It also makes it easy to turn back on later.
The API docs outline why quite well.
This should make it possible to do saving of textures to image files
without any private API with the same featureset that GTK uses.
Also remove the gsktextureprivate.h include where
gdk_texture_get_format() was the only reason for it.
GtkMenuButton currently does not provide a way to tell
if it's open programmatically. The existing methods,
`popup()` and `popdown()`, do not expose any state
to callers. If someone wanted to know whether or not
a menubutton was open, they needed the popover. Given
that GtkMenuButton can manage the popovers itself,
that's not always an option for app developers.
This commit adds the `active` property and associated
methods, where `gtk_menu_button_set_active ()` replaces
both `gtk_menu_popup ()` and `gtk_menu_popdown ()`.
This addition also mirrors changes in other places,
Such as `GtkWidget:visible` vs `show()`/`hide()`.
When we truncate the command queue because it
is too big, we were messing up our state accounting
and running into criticals as a consequence.
This can be reproduced by opening a well-populated
fishbowl demo in the inspectors recorder.
Fixes: #5188
Add GskMaskNode, and support it in the render node
parser, in the inspector and in GtkSnapshot.
The rendering is just fallback for now.
Based on old work by Timm Bäder.
The Common Print Dialog Backends (CPDB) concept has GUI-toolkit-independent
backends for each print technology (CUPS, Print to File, cloud printing
services, ...) and each print dialog (GTK, Qt, Chromium, ...) is supposed
to use this backend, so that changes in print technologies can be centrally
and quickly covered by changing the backends and everything new gets available
in all print dialogs.
This commit provides a GTK print dialog backend to add support for the CPDB
concept. It communicates with all installed CPDB backends and so gives support
for all these print technologies to the GTK print dialog.
To make use of CPDB the GTK print dialog is supposed to be installed with this
backend and the 'Print To File' backend, and not any others to prevent printer
duplication.
The 'optional' annotation should be used in these cases rather than 'nullable'.
NULL can provided to ignore these output parameters, but the function is
not setting the output parameter to NULL.
We no longer need to make much distinction between multiple logical
devices, plus it breaks esp. with the Xwayland input device distribution.
Just iterate across all devices and reset their scroll valuators.
Closes: https://gitlab.gnome.org/GNOME/gtk/-/issues/4160
This is a bit spaghetti right now, since seats and devices were
heavily entangled there are a number of crossed private API calls that
should ideally not be there.
Let this be a first step, so more bits may move from the seat
implementation to devices.
This commit implements the idea brought up in #5032, and provides a
simple function for GtkEditable implementations using a delegate object.
The accessible state is proxied from the outher GtkEditable to the
delegate.
This file, event though a clump of input-y objects, has more of
seats than anything else. Rename it so that we can start splitting
these objects out of it.
The `get_child_at_index()` API model comes from AT-SPI, and it's not an
efficient design, especially when coupled with large widgets.
Replace `get_child_at_index()` with `get_first_accessible_child()` and
`get_next_accessible_sibling()`.
That allows efficiently retrieving all the children, simplifies the
implementation of GtkAccessible in GtkWidget and closely resembeles the
GtkWidget API.
Getting the last child and previous sibling for iterating backwards is
not a part of the interface at the moment, but they can be added at a
later date.
Note that this change required tracking the next stack page in
GtkStackPage.
Reduce our dependency on linked lists; pointer arrays are better at
cache locality.
Additionally, we can avoid quadratic behaviors when finding a child at a
given index.
Otherwise, we end up with a single long row
pushing the content of all the other rows
off to the left, which is much worse than
ellipsizing.
Fixes: #4710
This is currently just used as a convenience storage of the startup ID
between the GtkApplication and the GtkWindow (after it's ready to notify
on it).
This could be untangled in the GTK layers so there is no involvement
from GDK in keeping the startup ID around, in the mean time just deprecate
these gdk_wayland* API calls.
These pieces were added in commit 8d2f81cca4, but are no longer
necessary since there's no capture_button_press special behavior
toggle.
This can be simplified again.
Do not get a GdkEvent first and foremost, and only do that in the
parts where the GdkEvent is absolutely necessary (i.e. popping up
the WM menu for the window).
We can abstract widgets about the specific ongoing sequences that
are triggering a GtkGesture. This used to be more necessary in
GTK3 world where complex widgets might have required handling
different events in different areas, but in GTK4 world that would
be done with multiple widgets.
This is no longer necessary to carry forward.
It does not require special knowledge about the ongoing event
sequences, and it can instead trust the event controller. Make
it use gtk_gesture_set_state() generically.
The design patterns using statusbar are no longer popular,
and it is pretty easy to make a statusbar yourself with boxes
and labels, if you need one. The only thing special about
GtkStatusbar was its window resize handle, but that has
been gone for a long time.
... and use this check in gdk_gl_context_make_current() and
gdk_gl_context_get_current() to make sure the context really is still
current.
The context no longer being current can happen when external GL
implementations make their own contexts current in the same threads GDK
contexts are used in.
And that can happen for example by WebKit.
Theoretically, this should also allow external EGL code to run in X11
applications when GDK chooses to use GLX, but I didn't try it.
Fixes#5392
When checking characteristics of the context
for downloading, we were using self->context,
even though we are using a possibly different
context for downloading.
Pass the right context along and use it.
File may not have paths, and we should handle
that without incident. While we are at it, add
some logging so GDK_DEBUG=dnd gives us enough
output to see what is going on.
It does not make sense to sync and wait in the
same context, that is just a no-op. The intention
of this code clearly was to sync in the gst
context, and wait in the gdk one.
That also matches what the gtk sink implementation
in gstreamer does.
Set the label to expand, so it actually fills
the width that we allocated for it, instead
of shrinking back to the minimum width for
its height.
Fixes: #5521
Setting max-width-chars to the number of characters
in the string works ok only as long as the average
char width we get from Pango matches reality. Sadly
that seems not always the case, and this code was
causing short Chinese tooltips to always be broken
into two lines.
Fixes: #4470
At the moment, GTK applications search for "desktop-startup-id" in the
platform data on Wayland , but desktop environments such as plasma set
"activation-token" property instead as indicated in the spec:
activation-token: This should be a string of the same value as would
be stored in the XDG_ACTIVATION_TOKEN environment variable, as specified
by the XDG Activation protocol for Wayland.
https://specifications.freedesktop.org/desktop-entry-spec/desktop-entry-spec-latest.html#dbus
Instead of adding events to the application event queue, dispatch
them directly to the right display. We know this when the event is
to be dispatched.
This is the same as used for the `sendEvent` method in `GdkMacosWindow`.
To achieve this I factored out the generic NSEvent to GdkEvent translation.
We can send an event directly, when we receive it in the GdkMacosWindow
directly from the OS.
The idea behind this code was to let scalable
images (i.e. mainly SVGs) provide twice as much
detail when the scale is 2. But we were also
using a scaler for pngs, causing them to be too
small on a hidpi screen. Fix that.
Note that there are cases where you want scaling
for pngs (when you display them scaled down, so
the image has 'hidden' detail). But we are not
attempting to handle that situation automatically.
By passing the events during a (midal-ish) drag operation to the main loop,
we're able to keep up with what's happening. This allows the internal
drag state (GtkDragSource) to be updated and be done when the drag is
done.
The Drag data should pass through the macos pasteboard system.
We need to provide some pasteboard type. Let's make it a "URL",
which is a pretty generic type anyway.
The handling is done similar to drag targets.
Note that dragging is a modal action on macos: no events
are sent to the main window. This could cause trouble when
we finish the drag, and not finish the gesture in GTK.
Escape XML tags in gi-docgen oriented comment e.g. from <child> to
`<child>`, so that they don't become HTML tag on the final webpage.
This fix includes everything from commit ff46ea64 and #5312.
Fixes#5312
Instead of keeping a GtkStringSet per IconTheme,
just make one for the whole GtkIconTheme.
This avoids loops of the themes in some places, and
due to the overlap in icon names between the themes,
it reduces the amount of memory we use for the icon
names with Adwaita+hicolor from 5+4 chunks to 6 chunks.
When gi-docgen added the "related" key, it also began checking
that the dependencies were actually dependencies otherwise they aren't
listed as dependencies. The capitalization needs to match the
namespace.
This will make it easier to reuse from drag integration so that we don't
require clipboards for everything.
We will need to subclass the pasteboard provider twice, however, both
for clipboard and dragging.
This was causing animation and transition to stop randomly and reset
their state to initial state.
This issue has existed since commit
7b68bdb831.
Closes#4426
Just to help static analysis out.
self->n_columns can't ever be 0, since
we clamp it between min_columns and
max_columns, with min_columns always
being at least one.
Clang was complaining that we never use the
value stored in mime_type. Just don't store it,
we are only interested in the side-effect
(interning the string).
The build is error-free and we want to keep it that way.
We'd also like to make Timm do real work instead of having to fix clang
warnings all the time and this build is a clang build.
When transitioning from internal to malloc, the strings were placed in
the wrong order to g_strconcat(). This fixes an issue with undo where
if you hit the boundary in just the right way, your undo stack will do
unexpected things.
Fixes#5506
When the model button just has a text label and accel text,
the button fills and the accel label is implicitly aigned to end.
When there's also a icon, even though it's not shown (because
icons are only shown if there's no text), the button doesn't fill
and the accel ends up not aligned (assuming one of the other buttons
is longer). Ensure that the accel label is aligned to the end.
Fixes: https://gitlab.gnome.org/GNOME/gtk/-/issues/5504
Use the fallback logic to generate the base path for the GtkAtSpiRoot
if the GApplication has no DBus object path to guarantee that
the base path will not stay NULL.
`gtk_at_spi_cache_add_context()` checks if the GtkAtSpiContext's path
is NULL before inserting the context object into the hash table.
Do the same in `gtk_at_spi_cache_remove_context()` to avoid a NULL
pointer dereference in `g_str_hash()` during the hash table lookup
if a context with NULL path is removed. That can happen when the
GtkAtSpiRoot::base_path is NULL, which, in turn, can happen if
`g_application_get_dbus_object_path()` returns NULL.
==394047==ERROR: AddressSanitizer: SEGV on unknown address 0x000000000000 (pc 0x7fd1966f8b84 bp 0x7fff11e3ded0 sp 0x7fff11e3de58 T0)
==394047==The signal is caused by a READ memory access.
==394047==Hint: address points to the zero page.
#0 0x7fd1966f8b84 in g_str_hash (/usr/lib/libglib-2.0.so.0+0x37b84)
#1 0x7fd1966f9c09 in g_hash_table_contains (/usr/lib/libglib-2.0.so.0+0x38c09)
#2 0x7fd196062c10 in gtk_at_spi_cache_remove_context ../gtk/a11y/gtkatspicache.c:447
#3 0x7fd19606e0a9 in gtk_at_spi_root_unregister ../gtk/a11y/gtkatspiroot.c:653
#4 0x7fd196067f58 in gtk_at_spi_context_unrealize ../gtk/a11y/gtkatspicontext.c:1559
#5 0x7fd195ced97f in gtk_at_context_unrealize ../gtk/gtkatcontext.c:668
#6 0x7fd195f5576e in gtk_widget_unroot_at_context ../gtk/gtkwidget.c:2399
#7 0x7fd195f55bd2 in gtk_widget_unroot ../gtk/gtkwidget.c:2499
...
The _gtk_file_system_model_update_file() function is not used outside
GtkFileSystemModel, so no need to expose it in the header.
Shuffle it around in code, and remove it from the header.
The marked days are set only as part of gtk_calendar_select_day().
This is insufficient, especially because the day-selected signal
is emitted after the marked days are set in gtk_calendar_select_day().
The marked day gets applied to the current month, and either the previous
or subsequent month if they are visible within the current month.
This doesn't make any sense and likely was an accidental regression
in a6f9052cf1. Clarify the docs.
As part of a6f9052cf1, marked days lost
their style, essentially making that function worthless. Previously,
they were simply bolded, but that doesn't give them proper justice.
This issues a warning when an enum value is compared to a value that is
out of range for the enum.
We do this a lot, either when using -1 for undefined values or when
comparing array sizes to enum values like so:
enum {
ONE,
TWO,
THREE
} some_enum_value;
const char *names= { "one", "two", "three" };
g_assert (some_enum_value < G_N_ELEMENTS (names));
switch_to_selected_folder can be called when the
selection contains more than one item. Handle it
like it used to be handled: switch to the first
folder we find.
Fixes: #5494
We have various layers where we store the startup ID for a request,
since this API does not have a GdkToplevel that we can refer about
for the Wayland platform, this is the most obvious candidate to
start untangling these various layers.
Deprecate this call, it is already unused in the gtk/ side.
This should do nothing worthwhile anymore, the X11/Wayland GtkApplication
implementations do already pass the startup ID from the platform_data
via windowing specific APIs, and the application handling the request
via show()/present() should trigger the activation request.
While this used to be tangential to windows showing or requesting
focus, the xdg-activation Wayland protocol does merge both concepts
together.
But also, for a correct interaction with the compositor, the
toolkit should ideally merge the activation request resulting from
both into the same one, so that the gdk_toplevel_focus() request
replies to the startup token that started the application and
correct focus-stealing prevention/etc happens, instead making up
one just in time for the focus request.
This kind of requires doing things in the right order, a show()
request on the GtkWindow should activate any pending activation
token on the toplevel, a present() request should additionally
create a new token if there was none pending. And
xdg_activation_v1_activate() should happen once on both.
Shuffle the gdk_toplevel_set_startup_id() calls so that this
happens in the right order for Wayland, while making X11 happy
too.
In the way towards deprecating gdk_display_notify_startup_complete(),
make gdk_toplevel_set_startup_id() on X11 perform this piece of messaging
itself. It should be harmless that the message is emitted twice, if
callers do still use that API.
This call has everything to perform activation as specified by the
xdg_activation protocol, notably a surface to activate as opposed to
gdk_display_notify_startup_complete().
Make activation happen here, so that the surface gets activated when
its gets a startup ID assigned.
The way the check is written, if the build is native, then the
introspection option has no effect.
Particularly yocto project does want to disable introspection in
native builds and enable it in cross builds (both via the option),
and without this patch the former is not possible.
Signed-off-by: Alexander Kanavin <alex@linutronix.de>
It's a big performance drag to create many cells (and their associated
widgets) for columns that may never be shown. Only create cells
is the column is visible, and subsequently create/remove cells
when in column_set_visible.
The value property is only getting notified when it's unset
or when it's set from gdk_drop_real_value_async(). Make sure
to always notify :value when its changed
The term "cancelling" is used with GCancellable, a user clicking a close
button is not that.
User input is usually described as "dismissing", so we use that term.
The owner_events=TRUE grab makes GDK on X11 see events happening
outside every client window as received on the grab window.
Additionally check that the pointer is inside the grab window
(i.e. it received GDK_CROSSING_NORMAL crossing events for the
core pointer) in order to handle clicks happening outside client
windows.
These new paths are expected to be a no-op on Wayland, and to
also work for touchscreen input on X11, due to emulated pointer
events.
A screen reader user is not interested in GTK internals, for example,
he does not care whether a button is an image button or not,
and a screen reader will report the fact that it is a button anyway.
Same applies for GtkEntry widgets, for example.
This actually is sufficient to fix gnome-control-center#2244.
And, according to the discussion in #5145, it should be fine.
Fixes build with only `-Dgtk_doc=true` without
`-Dintrospection=enabled`:
Program gi-docgen found: NO
Configuring gdk4.toml using configuration
docs/reference/gdk/meson.build:13:2: ERROR: Tried to use not-found external program in "command"
`introspection` is `auto` by default.
The rest of the docs build is only contingent on
`if get_option('gtk_doc')` so we should use the same restriction here.
We check for `build_gir` below already, and the gi-docgen subproject
itself does not need gobject-introspection so it's fine to do it like
this.
Installed tests require access to the system prefix, and thus a
system-wide installation of Meson, which we don't have.
We're going to restore this job at a later date.
We don't want to bring undefined dependencies into the image.
Additionally, Wayland depends on Meson, and we don't want to use
Fedora's version of Meson.
ClutterInputFocus/GtkIMContext uses char based offset for
delete_surrounding, however, text_input_v3 uses byte based offset for
it. Currently only GTK with mutter can work correctly via text_input_v3
because they both forget to convert between char based offset and byte
based offset.
This commit fixes it in GTK by converting byte based offset to char
based offset with the UTF-8 encoded surrounding text.
Fixes <https://gitlab.gnome.org/GNOME/gtk/-/issues/4566>.
If the drag events are claimed by another gesture (e.g. a GtkDragSource
in an item widget), list base still commits a rubberband selection, for
a rubberband which wasn't even visible yet. This is a problem for the
GNOME Files application which needs both rubberbanding and drag-n-drop.
My previous fix[0] was enough for the case where the event sequence is
claimed right before the first GtkDragGesture::drag-update emission,
but it's useless if the event is claimed later (e.g. after the drag
treashold), because a rubberband already exists by that time.
Therefore, the complete solution requres checking whether the event
sequence is no longer being handled by our gesture, and commit the
selection changes only if it is, but otherwise cleanup the rubberband.
This is what GtkFlowBox does already, so let's do the same here.
[0] commit dc4540fae9
Dummy dependencies are not required to execute a subproject
automatically for providing a program, nor do you need to explicitly
call subproject() to do that.
A `[provide]` section in the wrap file is enough.
Subprojects that use meson.override_dependency() do not require the
caller to provide the dependency variable name inside the subproject.
We also don't want to provide the *subproject* name, because the
subproject name can be `pango-1.50.12` instead of `pango` when using
wrap-file to download the tarball instead of using wrap-git. This
causes the pango subproject to be executed twice when using gtk as
a subproject inside gstreamer (which uses pango-1.50.12 as
a wrap-file).
All the dependencies we use can be switched in this way, but the
remaining ones need to be changed to use meson.override_dependency()
first.
The is_msvc_like change is wrong; it used a false correlation between
"compiler being used" and "dependency method" by saying that on
Windows, when building with MSVC, you will only use CMake to find png,
jpeg, tiff.
You can use pkgconfig to find these deps on Windows with MSVC -- when
the deps have been built with Autotools or Meson (with MSVC). You can
also find these deps using CMake on other platforms like macOS or
Linux.
The solution is simple: just search for both names on all platforms,
and just search for the pkgconfig name first.
The tooltips from the Grid View & List View buttons are unnecessarily long and look different from the tooltips used in Nautilus.
This commit makes the tooltips to be consistent with Nautilus and, consequently, makes them shorter.
Accept labels can be used for additional context regarding
the purpose of a file. The old GtkFileChooser APIs allowed
developers to set it, but the initial FileDialog API was missing
this functionality.
This commit adds `gtk_file_dialog_set_accept_label ()` to
restore the missing functionality.
Closes https://gitlab.gnome.org/GNOME/gtk/-/issues/5421
The EventControllerFocus on the list item, updates the list base focus
tracker and scrolled to position any time the list item enters focus.
This works when interacting within a single window, but has unexpected
results when changing focus between multiple windows.
Instead of using the focus controller workaround, just make
gtk_list_base_update_focus_tracker the set_focus_child vfunc
Fixes: https://gitlab.gnome.org/GNOME/gtk/-/issues/5433
Fixes: https://gitlab.gnome.org/GNOME/gtk/-/issues/5432
meson setup:
configuration
meson compile:
compilation
meson install
installation
Do not use ninja directly, and do not use `meson` as a synonym for
`meson setup`.
The unaligned-offscreen and upside-down-label-3d tests are failing after
upgrading our CI images, seemingly because of some font rendering issue
that is hard to track. Let's use the "failing" testsuite mechanism that
we also use for the reftests.
Try to get a native file:// URI instead of any other GVFS
scheme, for interoperability with apps only handling file:// URIs.
This is what GTK3 Nautilus and Thunar do, so apps should be tuned
for this behavior.
See also https://bugzilla.xfce.org/show_bug.cgi?id=13845Fixes: #5422
The path bar does a lot of manual management of buttons, mostly to
be able to show navigation arrows when there's not enough space to
show the full path.
Since the GTK4 migration, this is slightly broken in some cases, due
to the 'need_sliders' variable being always set to TRUE. Furthermore,
after the introduction of the Recent button as a special cased fake
root, the allocation of the buttons is generating warnings.
Reimplement the path bar as a GtkBox, inside a GtkScrolledWindow.
This mimics what Nautilus does, and allows us to make navigation more
predictable, and remove most of the complexity from GtkPathBar. It
also prevents it from generating allocation warnings.
The path bar itself now doesn't override GtkWidget.measure nor
GtkWidget.allocate; instead, it delegates layout to the GtkBinLayout
layout manager.
CSS is adjusted to account for the changed hierarchy of buttons.
It's positioned and looks exactly like the browse_toggle_view_button,
but due to the way things are organized, we cannot simply reuse that
button.
Add a clone of browse_toggle_view_button in the search entry page of
the toolbar stack. Make it toggle the same action as of the original
button, and bind the icon name and tooltip texts to it too.
Most of the pointer comparisons against 'browse_files_column_view'
should actually be performed against the current view widget. As
it turns out, it weren't that many places after all.
Add a grid view outside of the widgetry tree. The grid view mimics
the column view using bindings, so we only need to manage the column
view.
Also add a button in the path bar section to toggle the view. This
is handled as a new 'toggle-view' action in the file chooser.
The way switching between views currently work is by setting either
the column or grid view as the child of the GtkScrolledWindow. This
has the benefit of unmapping the unused view, which is nice and can
avoid some tricky situations with thumbnails.
Fixes https://gitlab.gnome.org/GNOME/gtk/-/issues/233
A Drag surface does not have a parent surface. Therefore, if we initialize
it with one, it's inheriting the frame clock from the parent, but the
drag surface is not linked to the parent. Once the drag surface is destroyed,
it's disposing the frame clock, which results in a "frozen" application.
This is an extra safeguard that avoids grabbing resources longer than
nessecary. It also ensures the resource is removed from the drag,
so it is not freed again.
Notifications are not going to be emitted during the finalization, and
GObject will warn if you try to acquire and release a reference to the
notification queue when the reference count of an object has reached
zero.
Fixes: #5420
Recommend that scope implementations should fall
back to or derive from GtkBuilderCScope in order
to not lose GTK's type-guessing machinery.
Related: #5398
Nested async calls are always a challenge.
Hopefully, things are straightened out now,
and we report GTK_DIALOG_ERROR errors for
the cases we care about.
If the parent window of the button gets destroyed
while the dialog is open, we cancel the async op,
but we need to be a little more careful about not
stepping on glass.
If the parent window of the button gets destroyed
while the dialog is open, we cancel the async op,
but we need to be a little more careful about not
stepping on glass.
Determine the location to save testcases in dynamically,
trying first a GTK_SOURCE_DIR environment variable
and then the current directory as the GTK source dir,
ultimatively falling back to just saving in the current
directory.
This avoids leaking details of the build environment
into the produced artifacts and should make GTK builds
more reproducible.
Fixes: #5403
The header in GtkColumnView has multiple event handlers
there is a ::pressed handler in GtkColumnView for
resizing the columns in CAPTURE as well as a motion
and drag controller. The ::release handler is in
GtkColumnViewTitle. We can't claim the event in the
existing handlers because then the ::release handler will
never get called. Currently, however, all clicks get propagated
to the ColumnView from the header which can be problematic.
Since we don't usually want the clicks from the header
handled on the view, claim it in the BUBBLE phase.
Fixes: https://gitlab.gnome.org/GNOME/gtk/-/issues/5425
Currently, the GdkSurfaceX11 implementation relies that the upper
layers hid the surface before destruction, and that no
GdkSurfaceClass.compute_resize happened between them. If these
circumstances happened, there would be a compute_size timeout left
dangling after the surface got destroyed, poking at incorrect data
later on. Something that looks like this was reported in the
recent mutter-x11-frames "SSD frames server":
mutter-x11-frames:423016): GLib-GObject-WARNING **: 19:41:16.869: invalid unclassed pointer in cast to 'GtkWindow'
Thread 1 "mutter-x11-fram" received signal SIGTRAP, Trace/breakpoint trap.
g_logv (log_domain=0x7ffff7f7c4f8 "GLib-GObject", log_level=G_LOG_LEVEL_WARNING, format=<optimized out>, args=<optimized out>) at ../../../glib/gmessages.c:1433
1433 ../../../glib/gmessages.c: No such file or directory.
(gdb) bt
#0 g_logv (log_domain=0x7ffff7f7c4f8 "GLib-GObject", log_level=G_LOG_LEVEL_WARNING, format=<optimized out>, args=<optimized out>) at ../../../glib/gmessages.c:1433
#1 0x00007ffff73470ff in g_log (log_domain=log_domain@entry=0x7ffff7f7c4f8 "GLib-GObject", log_level=log_level@entry=G_LOG_LEVEL_WARNING, format=format@entry=0x7ffff7f84da8 "invalid unclassed pointer in cast to '%s'")
at ../../../glib/gmessages.c:1471
#2 0x00007ffff7f72892 in g_type_check_instance_cast (type_instance=type_instance@entry=0x5555558e04b0, iface_type=<optimized out>) at ../../../gobject/gtype.c:4144
#3 0x00007ffff791e77d in toplevel_compute_size (toplevel=<optimized out>, size=0x7fffffffe170, widget=0x5555558e04b0) at ../../../gtk/gtkwindow.c:4227
#4 0x00007ffff7f4f3b0 in g_closure_invoke (closure=0x555555898cc0, return_value=return_value@entry=0x0, n_param_values=2, param_values=param_values@entry=0x7fffffffdeb0, invocation_hint=invocation_hint@entry=0x7fffffffde30)
at ../../../gobject/gclosure.c:832
#5 0x00007ffff7f62076 in signal_emit_unlocked_R
(node=node@entry=0x55555588feb0, detail=detail@entry=0, instance=instance@entry=0x55555560e990, emission_return=emission_return@entry=0x0, instance_and_params=instance_and_params@entry=0x7fffffffdeb0)
at ../../../gobject/gsignal.c:3796
#6 0x00007ffff7f68bf5 in g_signal_emit_valist (instance=<optimized out>, signal_id=<optimized out>, detail=<optimized out>, var_args=var_args@entry=0x7fffffffe050) at ../../../gobject/gsignal.c:3549
#7 0x00007ffff7f68dbf in <emit signal ??? on instance 0x55555560e990 [GdkX11Toplevel]> (instance=<optimized out>, signal_id=<optimized out>, detail=detail@entry=0) at ../../../gobject/gsignal.c:3606
#8 0x00007ffff7a8de96 in gdk_toplevel_notify_compute_size (toplevel=<optimized out>, size=size@entry=0x7fffffffe170) at ../../../gdk/gdktoplevel.c:112
#9 0x00007ffff7a4b15a in compute_toplevel_size (surface=surface@entry=0x55555560e990 [GdkX11Toplevel], update_geometry=update_geometry@entry=1, width=width@entry=0x7fffffffe220, height=height@entry=0x7fffffffe224)
at ../../../gdk/x11/gdksurface-x11.c:281
#10 0x00007ffff7a4c3b2 in compute_size_idle (user_data=0x55555560e990) at ../../../gdk/x11/gdksurface-x11.c:356
#11 0x00007ffff733f67f in g_main_dispatch (context=0x55555563f6e0) at ../../../glib/gmain.c:3444
#12 g_main_context_dispatch (context=context@entry=0x55555563f6e0) at ../../../glib/gmain.c:4162
#13 0x00007ffff733fa38 in g_main_context_iterate (context=0x55555563f6e0, block=block@entry=1, dispatch=dispatch@entry=1, self=<optimized out>) at ../../../glib/gmain.c:4238
#14 0x00007ffff733fcef in g_main_loop_run (loop=loop@entry=0x5555560874a0) at ../../../glib/gmain.c:4438
#15 0x0000555555557de0 in main (argc=<optimized out>, argv=<optimized out>) at ../src/frames/main.c:68
It perhaps makes sense to warn in these situations, but either way
it sounds like gdk_surface_x11_finalize() could enforce the correct
behavior by ensuring there is no dangling timeouts/data. This commit
does that.
The argument to xdg_activation_token_v1_set_surface is documented to be the
surface requesting the activation, not the surface to be activated, which is
given later when calling xdg_activation_v1_activate.
(c.f. 36cee4bdbc)
Use the same logic as in gdk_wayland_app_launch_context_get_startup_notify_id,
i.e. if we have a surface with focus, set that, otherwise set NULL.
This fixes requesting urgent/focus on wlroots (compositors like Sway, etc.),
which was blocked as the surface requesting the activation didn't have focus.
Signed-off-by: Joan Bruguera <joanbrugueram@gmail.com>
Active state is handled by main now. It appears that listitem and
treeexpander handled it manually (probably before main did). This
is unnecessary now, so let's remove it.
If the anchor is below the expanded item, the expanded item will
go out of view if there are sufficient children items. This is not
ideal, so make sure to scroll to the item to ensure it remains in
view.
With the introduction of the hide-expander property, I noticed that
the active state would persist in many cases because the release
signal was never emitted. In gtk3 tree expanders, expanded
on release. gtk4 expanded on press to match window explorer.
Per irc chat, the designers didn't have a strong preference
for press or release. In order to keep consistency and
fix this bug, let's move back to release.
We are caching the bus address as data on the display object when it
exists, but fail to set the data when the bus address doesn't exist.
That causing excessive calls to GetAddress when the accesssbility
bus doesn't exist. Make sure to cache a non-existent accessibility
bus by setting the "" string.
gtk_widget_set_visible and gtk_window_present
are better alternatives, and calling gtk_widget_show
on newly created widgets is no longer necessary
anyway.
With GtkText and GtkTextView (and in extension, all their subclasses)
handling OSK activation activation, this gesture is only useful for
all text input widgets that are not subclasses of these 2 widgets,
e.g. the VTEs and crosswords of the world.
These still do need a hand in handling OSK activation, so only
set up the gesture for such cases.
If the ::release handler is invoked, the press/release happened without
drags in between. Additionally check that there is no selection at all.
This makes OSK invoked on taps that move the caret around, while tapping
in the selection invokes edition popup and text handles without bringing
in the OSK.
This way, the drag gesture lets the click gesture ::release handler
happen if there was no actual changes to the selected text (i.e.
too short drags). This matches the ::release handler behavior match
the situations in which the OSK was being invoked by the wayland
GtkIMContext.
If the ::release handler is invoked, the press/release happened without
drags in between. Additionally check that the press did not happen within
the selection, and that there is no selection at all.
This makes OSK invoked on taps that move the caret around, while tapping
in the selection invokes edition popup and text handles without bringing
in the OSK.
This way, the drag gesture lets the click gesture ::release handler
happen if there was no actual changes to the selected text (i.e.
too short drags). This matches the ::release handler behavior match
the situations in which the OSK was being invoked by the wayland
GtkIMContext.
This method is so far private for both external GtkIMContext
implementations and external GtkIMContext users, and is meant
to activate the OSK in the environments where this may happen.
GTK depends on the a11y infrastructure to be in place unless GTK_A11Y is
set to none. It appears that despite that, users attempt to
get around the a11y requirement without setting GTK_A11Y.
This can cause, amongst other issues, performance problems
with gtk applications. Log failure to connect to the a11y
bus.
The python3-toml package is deprecated, and replaced by python3-tomli.
At least, until we bump up the dependency to Fedora 37: then we can
depend on Python 3.11, and its TOML parser in the standard library.
See also: https://gitlab.gnome.org/GNOME/gi-docgen/-/merge_requests/168
These tests can be run manually, but are not suitable for use as an
acceptance test, so let's not make frameworks like Debian's autopkgtest
run these when they run ginsttest-runner in the most obvious way.
Signed-off-by: Simon McVittie <smcv@debian.org>
There are two possible interpretations of "expected failure": either
the test *must* fail (exactly the inverse of an ordinary test, with
success becoming failure and failure becoming success), or the test
*may* fail (with success intended, but failure possible in some
environments). Autotools had the second interpretation, which seems
more useful in practice, but Meson has the first.
Instead of using should_fail, we can put the tests in one of two new
suites: "flaky" is intended for tests that succeed or fail unpredictably
according to the test environment or chance, while "failing" is for
tests that ought to succeed but currently never do as a result of a
bug or missing functionality. With a sufficiently new version of Meson,
the flaky and failing tests are not run by default, but can be requested
by running a setup that does not exclude them, with a command like:
meson test --setup=x11_unstable --suite=flaky --suite=failing
As a bonus, now that we're setting up setups and their excluded suites
programmatically, the gsk-compare-broadway tests are also excluded by
default when running the test setup for a non-broadway backend.
When running the tests in CI, --suite=gtk overrides the default
exclude_suites, so we have to specify --no-suite=flaky and
--no-suite=failing explicitly.
This arrangement is inspired by GNOME/glib!2987, which was contributed
by Marco Trevisan.
Signed-off-by: Simon McVittie <smcv@debian.org>
By dividing the blur radius to obtain the clip radius, we may end up
with halved values that result in an overshunk clip mask. Extend this
so that we ensure to cover the last pixel.
Fixes artifacts seen with the cairo renderer in X11 when resizing
windows horizontally, a black 1px high line would be seen in the
top of the window due to these outset bounds being used in clipping.
More mysteriously, also seems to fix resize lag in the GL renderer
(also X11), if e.g. the bottom-right corner of a window is resized
diagonally in bottom-left -> top-right direction, or
bottom-right -> top-left.
Related: https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2175#note_1599335
By using wl_output_release(), GDK lets the compositor to clean up the
output global more nicely.
For example, currently, most compositors remove the global and then
destroy it later after N seconds expire. With this, the compositor could
experiment with destroying the output global once all its resources are
destroyed.
this allows setting a Gtk.GestureStylus to a state, where it can be
used to handle non-stylus devices (e.g. mice).
This might be useful for applications that handle stylus input, but
want to allow falling back to a mice, if the user is unable to provide
stylus input.
this is to prevent gdk from causing a segfault, when getting event axes
for events that don't have them (i.e. attempting to get pressure from a
mice input device).
GDK_TOUCH_END deserves the same treatment than GDK_BUTTON_RELEASE, since it's
subject to the same circumstances (popping up a menu on long press would be
immediately dismissed on release if we handled them there). Ideally, we would
want to match releases that we obtained a press for while grabbed, but as
the popup is also dismissed on GDK_BUTTON_PRESS/GDK_TOUCH_BEGIN, there's no
use for this tracking.
And GDK_TOUCH_CANCEL sounds weird as a reason to dismiss popups, just like
crossing events would.
Related: https://gitlab.gnome.org/GNOME/mutter/-/issues/2512
Even though button 1 (or touch down) presses do most often have
an effect in one way or another (starting drag, moving focus,
starting selection, ...), there is one situation that they do
immediately nothing: When clicking on the entry does not move
the text caret around. Dragging might start a selection, but
the entry did not do anything just yet, and an immediate
button/touch release should remain at "did nothing".
And that is precisely the hint that the Wayland IM context's click
gesture takes, clicks that do not scroll nor move the caret around,
having the GtkText not claim the gesture in that situation makes
the IM gesture able to do its thing without in-fighting.
This is typically not a problem when the GtkText is embedded in
another GtkEditable implementation (e.g. GtkEntry), since the
IM gesture is inactive and capturing from the parent widget, so
gets a pass that it otherwise doesn't get when both gestures are
in the same widget. This makes it work regardless of GtkText not
being a child of a composite widget, like NautilusQueryEditor
and AdwRowEntry.
Closes: https://gitlab.gnome.org/GNOME/gtk/-/issues/5351
Everybody (including myself) gets this wrong,
so accept 'simple' and 'none' as shorthands
for the official IDs 'gtk-im-context-simple'
and 'gtk-im-context-none'.
clang complained that we may end up jumping
to the cleanup code without initializing data
in the jpeg code. Always initialize data to
NULL to prevent that eventuality.
GTK knows when a surface is modally blocked and automatically drops
button press and release events, so do not block input in advance
from WM_MOUSEACTIVATE.
This is largely adapted from commit 83027c68f1 ("11: Implement
inhibit_system_shortcuts API"), with similar rationale:
To implement the inhibit_system_shortcuts API on X11, we emulate the
same behavior using grabs on the keyboard.
To avoid keeping active grabs on the keyboard that would affect
other X11 applications even when the surface isn't focused, the X11
implementation takes care of releasing the grabs as soon as the
toplevel loses focus.
Note that Windows has low-level keyboard hooks that could help achieve
the expected behaviour. This is implemented by spice-gtk & gtk-vnc for
example, but correctness isn't obvious. I left a TODO comment.
This patch helps implementing remote desktop widgets with GTK4, since
currently on win32 backend Alt-Tab and such are always left to the
system unless there is keyboard grab (which can't be requested by the
client API anymore, afaict).
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
We are not normally using the gtk40-pot target to
generate the gtk40.pot file. On the off chance that
somebody does, lets make sure we pass the same
arguments to xgettext here as in the make-pot script
that is used on damn lies.
Noticed this while reviewing the gi-docgen docs for GtkAspectFrame while developing some java bindings.
It's my understanding that @self was intended; as it would cause gi-docgen to interpret it as a reference to
the the GtkAspectFrame pointer named 'self'.
This script is used to extract our strings for translators
on damn lines, and passing these flags to xgettext makes
it put a hint into the pot file about strings that are
used as printf format strings.
8455b9ac74 seems to have introduced a problem where we can wind
up focusing no widget at all if the `while (parent)` loop doesn't
find a widget it can successfully move the focus to. This 'fixes'
that by falling back to doing the previous thing if we make it
all the way through that loop without moving the focus. Thanks to
@coreyberla for a hint to improve the implementation.
Signed-off-by: Adam Williamson <awilliam@redhat.com>
There's 2 things broken here:
- The mask was calculated on top of the GDK button (i.e. skipping
4-7 buttons), so GDK_BUTTON4_MASK and GDK_BUTTON5_MASK were not
assigned. This is now calculated on the (continuous) BTN_ evcodes
so it is guaranteed that the next 2 physical buttons (i.e.
back/forward) get these two places in the mask assigned.
- Furthermore, these buttons would be pushed to places in the
modifier mask that they didn't belong to. It is now checked hard
that only the first 5 buttons enable a modifier flag.
Overall, this ensures that no event masks with bonkers values are
forwarded, and that no stale implicit grabs are left after additional
buttons are pressed.
Closes: https://gitlab.gnome.org/GNOME/gtk/-/issues/5301
This makes GtkSettings values on X11 match what we get on
other backends.
Reporting size settings in logical pixels (i.e for scale
== 1) is useful for properly supporting mixed-DPI setups.
As X11 doesn't support mixed-DPI setups anyway, XSettings
doesn't bother providing logical values. Thus we scale
from physical to logical values ourselves.
Fixes https://gitlab.gnome.org/GNOME/gtk/-/issues/5223
Fixes https://gitlab.gnome.org/GNOME/gtk/-/issues/5230
This reverts commit acd9c12667.
This commit breaks the build with GLib main on all platforms,
and defining _GLIB_EXTERN arguably invades the GLib namespace.
A different fix for msvc will have to be found.
Add a test that runs make-pot. This will only pass
if you've updated po/POTFILES.in and .skip after
moving source files around.
Unfortunately, it won't catch new source files that
are missing.
These are being replaced by GtkFileDialog.
This commit only moves the headers for GtkFileChooserWidget and
GtkFileChooserDialog to deprecated/, and keeps the implementations
in gtk/, since they will eventually be salvaged into a private
GtkFileChooserWindow.
It is getting replaced by GtkAlertDialog
This commit only moves the header to deprecated/,
and keeps the implementation in gtk/, since it will
eventually be salvaged into a private, dialog-free
widget.
These are being replaced by GtkFontDialog
and GtkFontDialogButton
This commit only moves the headers for GtkFontChooserWidget and
GtkFontChooserDialog to deprecated/, and keeps the implementations
in gtk/, since they will eventually be salvaged into a private
GtkFontChooserWindow.
These are being replaced by GtkColorDialog
and GtkColorDialogButton.
This commit only moves the headers for GtkColorChooserWidget
and GtkColorChooserDialog to deprecated/, and keeps the
implementations in gtk/, since they will eventually be
salvaged into a private GtkColorChooserWindow.
We need the padding inside the filelistcell, so that
its event controllers cover the whole area.
Introduce a .complex style class for columnviews that
achieves that, and make the filechooser use it.
The build breaks with a C4013 warning/error on Visual Studio because we don't
have a prototype defined for _gtk_get_datadir(), so include gtkprivate.h.
The vs2017-x64 CI did not catch this error because it is building GLib as a
fallback subproject, causing the msvc_recommended_pragmas.h header not to be
found, which is used to detect problems like this.
The tracker search engine implementation was not
setting all the custom attributes that we require
now.
The quartz search engine will need similar fixes.
These settings existed before, we keep using them.
This loses some information about sorting by multiple
columns, but it is sufficient to get the same primary
sort column back.
The "Show Time" setting does not take immediate effect (only after
changing folders) because it's set as a single call to
column_view_get_time_visible() on the FileChooserCell creation.
Instead create a bind a show-time property that gets updated
as the setting is changed.
Move the gestures to the individual cells, and
make them trigger the context menu via an action
that takes item position and coordinates.
The semantics are changed slightly: the menu actions
now operate on the clicked item, not on the selection.
Still to do: Fix up keyboard activation.
If the async query fails to reproduce a file info,
we still need to thaw the model, otherwise it ends
up frozen forever.
This was deduced by reading the code, I haven't
actually seen it happen.
We can use the new collation property of GtkStringSorter,
and get the benefit of sort key caching. This commit
also fixes an accidental leak of all sorters, and
removes the sorter from the location column - we never
show that column when individual colummns are sortable.
This reverts commit 34752a15a71597d00a8d08befc545ac1c178b81b.
Leaving out the drag source portion as that needs a total
reimplementation. The GtkDropTarget only required minor
modifications.
Put a filter model between the selection model and
the filesystem model, and make it filter on the
filechooser::visible attribute. This makes the filer
combo in the filterchooser and the 'show hidden files'
item work. But we need to prod the filter to trigger
a refiltering every now and then.
Provide the filtered-out and visible bits as a file attributes
under the names filechooser::filtered-out and filechooser::visible,
so that we can filter on it.
To track changes of the selected items in a selection
model, we need to listen to both ::selection-changed
and ::items-changed.
This fixes the open button not turning sensitive
when initially loading a new folder.
When a list item is activated, we activate the default widget.
Unfortunately, due to some other bug, sometimes the open button
is not made sensitive, and then default.activate falls back
to activating the focus widget (which is the item we are just
coming from). Boom
Soon GtkFileSystemModel will not be a GtkTreeModel implementation,
so preemptively remove any usage of this interface. Populate the
list store using the GListModel's 'items-changed' signal.
This has to be the shortest-living object in GTK history!
It helped us greatly during the transition to GtkColumnView, but
now we can remove it in favour of GFileInfo directly. Perhaps I
could have never introduced GtkFileSystemItem in the first place,
but we're 30 commits deep and it's too late to just redo the whole
thing that will get us exactly here anyway.
We now start a mini-series of commits that will ultimately remove
the GtkTreeModel implementation of GtkFileSystemModel.
As a first step, port GtkSearchEngineModel iter through the files
using GListModel API.
Now that most of the treeview usage is gone, remove the remaining
code that uses it - mostly event handling code, which for now won't
work, but will be fixed by next commits - and drop the tree view
entirely.
So far, GtkFileChooserWidget has relied on GtkTreeView's selection
management. This commit moves it away from GtkTreeView, and that's
a massive surgery - sorry :(
The most important aspect of this commit is that 'selection_model'
is now the main model we deal with. Changing between directories,
recent files, and search, all sets the selection_model's model.
Selections are entirely handled by GtkSelectionModel now.
React to column view's 'activate' signal, instead of treeview's
'row-activated'. It doesn't handle file sensitivity yet, but that
will probably be dropped later.
Move the entire location column, which only contains the location
renderer, to the column view. The code to generate locations from
the current folder is essentially intact.
This commit moves the icon loading code into a new private
widget called GtkFileThumbnail, which is bound to the GFileInfo
of the model, and asynchronously loads the file icon from that.
Replace the 'list' page of the main stack with another page, this
one containing a GtkColumnView. This, again, is the very minimal
code to achieve a column view - and validate the GListModel code
introduced in the previous commit - but there's a long way until
this column view covers the full range of features of the file
chooser.
The tree view still lives in an unused 'list2' page. From now on,
commits will "cannibalize" the treeview, each commit porting any
particular feature - be it a column, an event controller, etc -
to the column view, and dropping the corresponding feature from
the treeview.
This is a trivial implementation of the GListModel interface. It
does not do anything fancy, like filtering out hidden files, nor
sorting.
The purpose of this minimal implementation is to bootstrap the
initial work to port GtkFileChooserWidget to GtkColumnView.
On platforms like NixOS, the libX11 installation prefix may differ from /usr/share,
breaking the hardcoded placeholders. Let’s re-use the X11 path definition from imcontextsimple.
Arrange for double-click-followed-by-drag to do
select by words, not select-and-dnd. This matches
the behavior in GtkTextView better and feels
intuitive.
Fixes: #2024
Just relying on GAppInfo leads to suboptimal
results. Instead, call either the OpenURI portal
or the org.freedesktop.FileManager1 interface
directly, and only fall back to GAppInfo.
The wrapper code for the OpenURI portal is taken
from gio, with small adjustments.
Fixes: #5260
When getting the serial for primary/clipboard selections we used a
function that largely relied on a GdkEvent being passed. We have
another available function that looks up the most recent serial
given the ongoing touch/tablet input as well.
This is the second best, compared to actually knowing the
input/device from the event that was received by the UI an triggered
the clipboard operation, and is already in use in other places
(e.g. window dragging). It is valid for these situations too.
Closes: https://gitlab.gnome.org/GNOME/gtk/-/issues/5250
Add a new option --deprecations to the validate
command that will warn about use of deprecated types.
The list of current deprecations is unfortunately
hardcoded in the source, so this list will have to
be kept up-to-date.
Fixes: #5256
In overwrite mode, every typed character gets
handled as a delete+insert, but we should not
record these as two individually undoable
steps.
This matches how we handle overwrite mode in
GtkTextView.
Fixes: #4411
We can get spurious focus-out/-in pairs when
the editable label is in a popover that gets
a Wayland keyboard enter event as a result of
clicking the editable label.
A timeout isn't a great solution, but nothing
better is available right now.
Fixes: #4864
Only clear a queued move_focus if the widget
we are focusing is actually visible.
This was happening in some cases when popovers
are dismissed by clicking outside, and it was
causing us to miss proper focus updates that
were already queued.
This partially undoes changes from 3dbf5038fa.
That commit did two things:
1) Move the focus update to after-paint time
2) Change from grabbing focus to the visible parent
to calling move_focus (TAB)
The second part did have the unintended consequence
of moving focus laterally.
Fixes: #4903
GtkSingleSelection will only emit either of those signals if they
change. But it is possible that only one of those properties changes,
and in those cases we want to only notify for that property changing in
the dropdown, too.
We don't want to notify::selected or notify::selected-item if they
didn't change.
This will bring performance benefits on frequently changing lists.
In particular, if lists get filtered or reordered, but the selected item
stays in the list, not doing a notify::selected-item will avoid updates
in connected handlers like GtkDropdown (and its handlers), thereby
avoiding lots of unnecessary updates.
There is a widespread need to access the CSS foreground
color for custom drawing in snapshot functions, so make
it available after gtk_style_context_get_color was
deprecated with a new widget api.
Some of our tests use deprecated style context api.
Most of them should be ported to use global style
providers eventually. For now, ignore deprecations.
The notable exception here are the global provider apis,
which are needed in some form and don't have a replacement
yet. Move them to gtkstyleprovider.[hc], so we can wholly
deprecated gtkstylecontext.[hc].
Move the implementations from gtksnapshot.c to
gtk/deprecated/gtkrender.c and deprecated these
functions. We want to get rid of them.
These functions are still used in some of our widgetry,
so use G_GNUC_BEGIN/END_IGNORE_DEPRECATIONS around
them.
Use newlines rather than spaces to separate file paths (or uri's)
when serializing text/plain files. There isn't a matching
deserializer, so we can do this in isolation. Newlines
seem to make more sense when pasting into a text editor etc.
Fixes: https://gitlab.gnome.org/GNOME/gtk/-/issues/5240
It doesn't require one generally anyway, because only the root can
change scale and when that happens the root will queue a redraw.
But even if the root doesn't queue a redraw, render nodes (the only
thing discarded by queue_draw()) are scale-independant.
As documented on MSDN:
> Unlike the WM_LBUTTONUP, WM_MBUTTONUP, and WM_RBUTTONUP messages, an
> application should return TRUE from this message if it processes it.
The template use in the inspector was not properly
disposing all widgets. gtk_widget_dispose_template
will only unparent widgets that have been named
as template children, so we need to make the toplevel
elements in the ui file named children, or manually
dispose them. This commit does the former.
These are a family of pretty specialized widgets, and
are very rarely used. Instead of porting them away
from GtkTreeView and GtkComboBox, deprecate them.
This reverts commit 11829fe7d0.
The mkenums_simple function can't properly handle headers
in subdirectories currently, so go back to the template
version.
For the same reasoning as the preceding commit.
Also don't make GtkColumnView focusable. Its internal list view
is already focusable, which is enough to take care of the empty
view case.
The container view itself being focusable makes keyboard navigation
slower by adding a useless focus step.
It also means if an item gets removed, the focus jumps back to the view,
instead of jumping to the next item, as seen in nautilus bug report:
https://gitlab.gnome.org/GNOME/nautilus/-/issues/2489
Instead of making the GtkListBase container itself focusable, override
the .grab_focus() vfunc. This way, calling gtk_widget_grab_focus() on
the view container keeps working sucessfully, but focuses the focus
item directly instead.
This is particularly useful to have because applicaiton authors do
not have direct acess to this class's children, so they can't call
gtk_widget_grab_focus() on them directly.
We connect to the inserted-text signal for the entry's buffer.
During the lifetime of the entry, the buffer changes. This is
literally the example used for GSignalGroup in the docs.
MinimumIncrement is an AT-SPI-ism that has no counterpart in the ARIA
specification, so it should not live inside public API. Additionally,
it's not really a useful method because it collapses two values on the
adjustment API.
The only method in the GtkAccessibleRange interface should be the
set_current_value(), which allows ATs to control the current position in
a ranged widget.
The AT-SPI implementation can now use all the accessible properties,
including the VALUE_TEXT one, mapped to the Text property on the
AtSpi.Value interface.
Empty/zero bounds are sent by the Wayland compositor if there are no
valid bounds to report, e.g. if there are no connected monitors. Report
this to GTK, which uses this to clamp calculated sizes, as INT_MAX, so
that clamping isn't done until there are actual valid bounds to clamp
to.
This fixes clients sometimes shrinking to their minimum size during
hotplugs or after having suspended the session.
We shouldn't assume there is always a monitor to derive bounds from.
If there is no monitor, pass empty bounds, as this matches what
xdg_toplevel.configure_bounds do in this case.
This is an experiment to see if I can keep up with
doing post-release version bumps, so git snapshots
will always have a different version from released
tarballs.
This commit also marks the beginning of the 4.10
development cycle, as 4.8 has been branched.
GTK4 gdk/broadway: correct gdk_broadway_device_query_state() to return pointer coordinates relative to the upper left corner of surface
See merge request GNOME/gtk!5053
Signal handlers ust return their preferred action and that one must be
unique.
Shout at them if they don't do that, before gdk_drop_status() does
tesame thing.
Related: gnome-build-meta#554
Related: gnome-builder#1799
"left of right" should be "left or right".
There's a small (subjective?) English nit in there as well: I believe
that buttons are placed (for example) "on the right" rather than "at the
right".
As far as I'm aware, these only exist with `gdk_wayland_surface_` names
for historical reasons, before these types were split.
This way, those functions will be able to access members of the
`GdkWaylandToplevel` struct. And it just saves a few lines of code.
There is nothing particularly specific to drawables
in there (and we don't have that concept anymore),
so just name the source file to match the header.
Easier for everybody.
Doing reset() on the text widgets after commit and delete_surrounding
is still too eager for some IMs (e.g. those that expect being able
to commit text while keeping a preedit buffer shown).
However, reset() is more of a "synchronize state" action on Wayland,
and it is still desirable to do that after changes that do come from
the IM (e.g. requesting the new surrounding text and cursor/anchor
positions). Notably here, the text_input protocol may still come up
with a preedit string after this state synchronization happens.
Shuffle the code so that the text widgets do not reset() the IM
context after text is deleted or committed, but the Wayland IM does
apply its practical effects after these actions happen. This keeps
the Wayland IM fully up-to-date wrt text widget state, while not
altering the ::commit and ::delete-surrounding-text behavior for
other IM context implementations.
Closes: https://gitlab.gnome.org/GNOME/gtk/-/issues/5200
Fixes: 5b78fe2721 (gtktextview: Also reset IM context after IM...)
Fixes: 7c0a395ff9 (gtktext: Also reset IM context after IM...)
Fixes: 52ac71b972 (gtktextview: Shuffle the places doing IM reset)
Fixes: 9e29739e66 (gtktext: Shuffle the places doing IM reset)
Move the autocleanup declarations into their
respective headers.
While we are at it, correct the autocleanup
declaration for GdkEvent to use gdk_event_unref,
not g_object_unref. Oops
The lookup order tests were relying on out
debug spew using g_log, so they can redirect
the output by setting a log writer function.
Rewrite this to use g_test_subprocess() and
parse stderr.
Introduce GDK_DISPLAY_DEBUG() and GDK_DEBUG() and
the helper function gdk_debug_message(). This is
meant to clean up the mess of our current debug
statements which wildly mix g_message, g_print
and g_printerr.
Check that the touchpad gesture event has a matching number of fingers before
updating the GtkGesture point tracking, instead of afterwards. Avoids pointless
tracking of these touchpad events when we know beforehand that the gesture
will never be activated by the touchpad events.
Related: https://gitlab.gnome.org/GNOME/gtk/-/issues/5199
The old wiki page has a couple issues:
* It's out-of-date, not having any notes for GTK4 specifically,
and it doesn't link to `gvsbuild`, which (I believe) is
the current official (and best) way to build GTK with MSVC.
* It's "locked", so it's harder for contributors to update. This
is OK, except for one spicy detail:
* It's not clear how to contribute/report issues on the locked wiki
pages, so out-of-date information falls off the radar.
Regardless :) the gtk.org GTK for Windows docs are a much better
springboard, in my opinion.
gi-docgen is supposed to be ran natively on the build machine, without
native: true dependency() searches for gi-docgen on the host system.
When it doesn't find it, it falls back to a subproject even if gi-docgen
is available on the build machine.
also make gtk_doc require introspection
Doing clever things with objcopy is faster and seems to be reliable on
x86_64 Linux, but also doesn't work on all toolchains and architectures:
in particular, Debian has had trouble with this on arm and mips.
In a distro build environment where we are compiling all of GTK every
time, the cost of potentially unreliable builds is higher than the cost
of using slower but more conservative GResource embedding.
Resolves: https://gitlab.gnome.org/GNOME/gtk/-/issues/5107
Signed-off-by: Simon McVittie <smcv@debian.org>
The X11 backend can mark modifiers like Shift as consumed even if they
aren't actually active, which seems to be something to do with making
shortcuts like `<Control><Shift>plus` and `<Control>plus` work as
intended regardless of whether the plus symbol is obtained by pressing
Shift and a key (like `+/=` on American, British or French keyboards)
or not (like `*/+` on German keyboards).
However, this can go badly wrong when the modifier is *not* pressed.
For example, terminals normally have separate bindings for `<Control>c`
(send SIGINT) and `<Control><Shift>c` (copy). If we disregard the
consumed modifiers completely, when the X11 backend marks Shift as
consumed, pressing Ctrl+c would send SIGINT *and* copy to the clipboard,
which is not what was intended.
By masking out the members of `consumed` that are not in `state`, we
get the same interpretation for X11 and Wayland, and ensure that
keyboard shortcuts that explicitly mention Shift can only be triggered
while holding Shift. It continues to be possible to trigger keyboard
shortcuts that do not explicitly mention Shift (such as `<Control>plus`)
while holding Shift, if the backend reports Shift as having been
consumed in order to generate the plus keysym.
Resolves: https://gitlab.gnome.org/GNOME/gtk/-/issues/5095
Bug-Debian: https://bugs.debian.org/1016927
Signed-off-by: Simon McVittie <smcv@debian.org>
The filetransfer protocol says to use
application/vnd.portal.filetransfer, but I used
application/vnd.portal.files when I implemented the
protocol. Oops.
This commit dds the correct mimetype, but we still
support the old one to preserve interoperatibility
with existing flatpaks using GTK 4.6.
Fixes: #5182
Some of the X keyboard layouts use compose
sequences of length one to make individual
keys generate multiple Unicode characters.
To support this use case, change the index
part of the table format to also include
an offset for length 1. Bump the table
version to indicate this change.
Fixes: #5172
For some of the a11y states, calling gtk_accessible_reset_state
can change the type of the state value from boolean or tristate
to undefined.
Handle that, instead of throwing criticals.
Related: !4910
The code in the fontrendering demo is a bit sloppy
and assumes that we always get a single run when
appending a sequence of 4 chars and 4 spaces.
That is not in general true, such as for Emoji.
Instead of working harder to handle Emoji here,
just give up and fall back to 'a'.
Fixes: #5166
We need to register the portal mime types before
the others to prefer them, doing this call async
messes up that ordering.
This is effectively reverting 69fb3648b2
When the IM commands the GtkText to delete text, the cursor position
would change, and so would the surrounding text. Reset the IM context
so that these updates are properly picked up by the IM.
Fixes backspace key behavior in the GNOME Shell OSK, since that relies
on the surrounding text being properly updated for the next iteration.
When the IM commands the GtkText to delete text, the cursor position
would change, and so would the surrounding text. Reset the IM context
so that these updates are properly picked up by the IM.
Fixes backspace key behavior in the GNOME Shell OSK, since that relies
on the surrounding text being properly updated for the next iteration.
Resetting the IM on IM updates is too eager and indeed the simple
IM context doesn't like that this happens in the middle of dead
key handling.
We however want to reset the IM after actual text buffer changes
(say, a committed string) moved the cursor position, altered the
surrounding text, etc. So that the IM implementation does know to
update its state.
Since there is going to be an actual IM reset anyways, it does
no longer make sense to try to preserve the old priv->need_im_reset
status during commit handling.
Fixes: 52ac71b972 ("gtktextview: Shuffle the places doing IM reset")
Closes: https://gitlab.gnome.org/GNOME/gtk/-/issues/5133
Resetting the IM on IM updates is too eager and indeed the simple
IM context doesn't like that this happens in the middle of dead
key handling.
We however want to reset the IM after actual text buffer changes
(say, a committed string) moved the cursor position, altered the
surrounding text, etc. So that the IM implementation does know to
update its state.
Fixes: 9e29739e66 ("gtktext: Shuffle the places doing IM reset")
Closes: https://gitlab.gnome.org/GNOME/gtk/-/issues/5133
Update the label size request when setting the digits property
by calling the update_label_request () util function.
That util function works by measuring the size request of the
label with the lower and upper values of the adjustment, then
taking the max. That way the size requisition is constant
regardless of the actual displayed value.
Since the util function internally works by setting the text
of the label, let it also set the text at last by taking in
account the current adjustment's value. Most of its callers
do that anyway.
Fixes https://gitlab.gnome.org/GNOME/gtk/-/issues/5156
If you've begun a user action and call `gtk_text_buffer_set_text`, you
get an unexpected warning:
```
Gtk-WARNING **: Cannot begin irreversible action while in user action
```
which can be fixed by doing the delete/insert yourself. But this is not
documented as occurring, so document it.
`apply_monitor_change()` already calls `update_scale()`.
Note that this only affects old compositor versions (see
`should_update_monitor()`) so it's just a minor cleanup.
We want to claim the event sequence in the click gesture when appropriate,
such as activating a row or clicking an editable cell, but this is currently
done too early, preventing other gestures for drag-and-drop and rubberband
selection entirely.
Fixes#3649Fixes#3985Fixes#4669
Do not perform coordinates transformation when gdk_event_get_position()
returns FALSE as it returns NaNs in that case and these coordinates
are not used anyway in further processing (closes#5134).
The way we explicitly set the font on the entry
conflicts with the placeholder text styling. But the
entry isn't normally empty, so placeholder text is
not that important here. Remove it and use a tooltip
instead.
When reordering notebook tabs, updating the sensitivity state of the
arrow buttons is necessary if the tab is moved to the beginning or
end of the tab list.
When notebook tabs are reorderable, pressing the notebook arrow buttons to
change the active tab results in tabs reordering unexpectedly.
Claim the event sequence after pressing an arrow button to avoid conflicts
with the motion/drag gesture used for reordering.
A typo resulted in the tab container widget being retrieved instead of
the tab widget. If an adjacent action widget was present, an infinite
loop occurred when switching tabs while a screen reader was enabled.
This function is probably not generally useful for a Gtk+/win32 user,
and it's only used internally by gdk-win32. It's time to deprecate it, I
believe.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Test that we can expand and collapse a row, and then
add another child below it, without crashing.
Adapted from the testcase in #4595.
This tests the fix in the previous commit.
When we collapse a node, we clear out the children,
but we were not disconnecting the signal handler on
the child listmodel, leading to bad outcomes when
that model is persistent and changing.
Fixes: #4595
Right now we only support system DPI awareness in GTK4. In that case
it makes sense to scale text with the DPI of the primary monitor, like
done in GTK3.
We plan to land support for proper fractional scaling in Gdk/Win32, so
in the future the "gtk-xft-dpi" setting will be gathered as intended,
i.e. for text magnification, as an a11y feature.
As I propose to deprecate gdk_win32_surface_get_impl_hwnd() next,
replace it with the alternative.
The main difference between the two functions is that
gdk_win32_surface_get_impl_hwnd() fails gracefully by returning NULL if
the surface is not of the win32 implementation.
All the surfaces should be native surfaces here, and the existing code
doesn't seem to deal with NULL case anyway.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
The function isn't used by Gtk itself anymore, and does not help much.
It creates extra issues for bindings, as it doesn't fit well with code
doing the same job for other objects.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Commit 59f6c50df8 set the memory limit to 100M,
which turns out to exclude some large, valid jpegs.
So, bump things to 300M, matching what was done
in gdk-pixbuf.
We were disabling the insert-emoji action when the
no-emoji input hint is set, but the Ctrl-. shortcut
was bypassing the action and kept working. Make
the shortcut activate the action instead.
Fixes: #5123
When some of the Emoji have been filtered out by
a search term, arrow keynav would behave oddly and
get stuck in invisible sections. Fix this by ignoring
any filtered out children when moving between
sections for arrow keynav.
Fixes: #5076
The function is gone since commit ea65abc7e2 ("Rewrite
GdkWin32Keymap (load table directly from layout DLL)")
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Currently, the wayland IM context sends zwp_text_input_v3.commit from
a number of places, and some of them with partial data. In order to
make client state updates "atomic" and complete, make the communication
happen over an unified notify_im_change() function that happens on
a narrower set of circumstances:
- The GtkIMContext is reset
- The GtkIMContext is just focused
- The gesture to invoke the OSK is triggered
- The IM context is reacting to changes coming from the compositor
Notably, setting the cursor location or the surrounding text do not try
to commit state on their own, and now will be flushed with the corresponding
IM update or reset. But also, these requests won't be prevented from
happening individually on serial mismatch, instead it will be the whole
state commit which is held off.
With these changes in place, all client-side updates are notified
atomically to the compositor under a single .commit request.
Fixes: https://gitlab.gnome.org/GNOME/gtk/-/issues/5106
Fixes: https://gitlab.gnome.org/GNOME/gtk/-/issues/5105
During text widget manipulation (inserting or deleting text via keyboard)
the IM context is reset somewhat early, before the actual change took place.
This makes IM lag behind in terms of surrounding text and cursor position.
Shuffle these IM reset calls so that they happen after the changes, and
ensure that the IM is actually reset, since that is currently toggled on
a pretty narrow set of circumstances.
Also, fix a bug during GtkEventControllerKey::im-update where the condition
on cursor position editability to reset the IM context was inverted.
[196/296] Linking target testsuite/gtk/builder.exe
/usr/lib/gcc/x86_64-w64-mingw32/11.2.1/../../../../x86_64-w64-mingw32/bin/ld: warning: --export-dynamic is not supported for PE+ targets, did you mean --export-all-symbols?
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
During text widget manipulation (inserting or deleting text via keyboard)
the IM context is reset somewhat early, before the actual change took place.
This makes IM lag behind in terms of surrounding text and cursor position.
Shuffle these IM reset calls so that they happen after the changes, and
ensure that the IM is actually reset, since that is currently toggled on
a pretty narrow set of circumstances.
I assume this was committed by mistake. It isn't used, and some
packaging systems will automatically remove it during `clean`.
Signed-off-by: Simon McVittie <smcv@debian.org>
We need to free the queued context list in dispose
if we didn't get to register the contexts, and we also
need to free the list properly when we do get to
register them.
This showed up in valgrind as leaked GList structs.
CI is mostly interested in GTK not introducing compiler warnings, other
submodules like Wayland might have their own and that shouldn't hinder
CI testing of GTK.
Disable -Werror for the wayland submodule, and let it be fixed independently
at some point.
When GTK_EVENT_CONTROLLER_SCROLL_DISCRETE is set, accumulate deltas also
for mouse scroll so a high-resolution mouse wheel click behaves in the
in the same manner as a low-resolution mouse wheel click.
Starting with the Wayland protocol wl_pointer >= 8, discrete axis
events have been deprecated in favour of high-resolution scroll event.
Add a listener for high-resolution scroll events and, for backwards
compatibility, handle discrete events as discrete*120.
Instead of calculating the discrete scroll deltas in
GtkEventControllerScroll, move that code to the event constructor and
access the precalculated values using gdk_scroll_event_get_deltas.
Refactor, no functional changes.
Starting with Linux Kernel v5.0 two new axes are available for
mice that support high-resolution wheel scrolling: REL_WHEEL_HI_RES and
REL_HWHEEL_HI_RES.
Both axes send data in fractions of 120 where each multiple of 120
amounts to one logical scroll event. Fractions of 120 indicate a wheel
movement less than one detent.
The 120 magic number is a copy of the Windows API, so this new
constructor can be used both in Linux >= 5.0 and Windows >= Vista.
gtk_tree_view_top_row_to_dy, which is called from GtkTreeView's
size_allocate function, changes the adjustment value. Since this
conflicts with the animation when changing the active row, bail
out until the animation is finished.
Fixes#4550
When a GtkTreeView scrolled horizontally, it was not possible to
select rows outside the initial area due to an erroneous comparison
between widget and bin window coordinates.
Original change to widget coordinates occurred in commit
a0de570e47
Commit adba0b97 fixed missed pointer crossings by using a helper function that
was already present and looked like did everything that was needed. However
this function was oriented to keyboard focus and it also did update the related
widget state. Doing these changes on pointer-based crossing was misuse, and
could cause weird interactions with keyboard focus management.
Fix this by using gtkmain.c gtk_synthesize_crossing_event() that is in fact
oriented to pointers.
Fixes: adba0b97 (gtkwindow: Synthesize pointer crossing events on state changes)
Closes: https://gitlab.gnome.org/GNOME/gtk/-/issues/5094
Instead of passing an event and figuring out coordinates from it, pass
directly the toplevel coordinates so that we can use this outside event
handling.
All callers have been updated to pass the coordinates, in practical effects
they were already based on the GtkNative.
The inner loop in gtk_paned_set_focus_child() tries to find the
topmost GtkPaned, however, if the `w` variable ends up becoming
NULL after bubbling up the entire GtkWidget hierarchy, this loop
never breaks.
Check for NULL in this loop.
Closes https://gitlab.gnome.org/GNOME/gtk/-/issues/5094
c68247f63b introduced a scroll multiplier,
intended to be significantly lower than the GTK 4.6 behavior but higher
than 1. However, it was _higher_ than 4.6, since 4.6 also had a permanent
1/10 multiplier in GDK, so the cited multiplier values were really 6.4 and
9.7.
We may have situations where velocity is 0/0, but are overshooting. Places where
this happens are mouse wheels, and continuous scroll that ended up still before
finish. In this situation we also want to run the animation for overshoot, so
check for the corresponding axes to also set up the kinetic scroll helper.
Closes: https://gitlab.gnome.org/GNOME/gtk/-/issues/4784
The expected configurability is not going to arrive yet from compositors, and
it is precipitate for GTK to gain any configurability. We do know a factor of 1
feels way too slow, and we do know a factor of page_size * pow (2 / 3) feels way
way too fast.
With the previous multiplier, gtk4-demo at its default size had a vertical textview
factor of 64.332901, and maximized on a 1920x1080 screen a factor of 97.585365.
Pick a magic multiplier that is both significantly below these values and above 1,
and stick to it.
Future work will add the configurability of smooth scroll events where it belongs.
At that point this commit may be reverted so we don't pile up on magic numbers again.
Related: https://gitlab.gnome.org/GNOME/gtk/-/issues/4793
Add missing #define g_memdup2() for gdksurface-broadway.c in case of enabled
broadway-backend as used otherwise.
Copy static would_drop() replacement for g_log_writer_default_would_drop()
from gtk-builder-tool.c to gtk-reftest.c
When widgets go mapped/unmapped, we repick but don't generate crossing
events. Since there could be stateful controllers that use those in
the previously picked widget (e.g. GtkEventControllerMotion), skipping
those breaks their state.
Ensure to send the relevant crossing events on every situation that
changes the pointer focus, so these controllers get a fair opportunity
to undo their state.
Closes: https://gitlab.gnome.org/GNOME/gtk/-/issues/2877
Even though the argument is non-nullable, GTK sometimes incurs in that
by itself by destroying the surface while the event is in flight. This
is the case of popping down a GtkDropdown. When this happens we simply
ignore the crossing event, but we should let it through instead, the
compositor did not send it in vain and we possibly still have pointer
state to undo.
Drop the surface checks, so that the event is propagated along GTK.
Following what was done for pinch/swipe events, give hold gestures their
own distinct sequence as well. Without this it was NULL, which was already
distinct to other touchpad gestures.
This delaying of the cancel event was made to avoid intermediate cancellation
for >=2fg hold gestures followed by pinch/swipe gestures, and it worked as
long as everything was considered to have the same sequence.
Since each pinch/swipe pointer gesture now gets its own sequence, this no
longer applies, nor works. This results in zoom/rotate/swipe gestures being
stuck since the sequence for the touchpad events changes mid-gesture.
Sticking to this pattern of giving touchpad gestures their own sequence,
these hold events cannot be assumed to coalesce with other touchpad gestures,
it is better to let it propagate altogether so that both the hold gesture
and the incoming gesture trigger coherent begin and end/cancel phases.
In the worst case, this results in "::begin, ::cancel, ::begin , ..." before
triggering a touchpad gesture, but the extra begin/cancel ought to be a safe
no-op in widgets.
Closes: https://gitlab.gnome.org/GNOME/gtk/-/issues/5003
The textbuffer test is calling into a function defined by the AT-SPI
accessibility backend. As of commit 4ddf1b70 we only build and run the
test on Linux, but the function in question isn't really
accessibility-related: it's just a serialization function.
Let people know that they will need to use GTK with the Nahimic service
disabled or OpenGL disabled or put their GTK application into the Nahimic
backlist, or try to use GLES, since there is a known issue in the Windows
nVidia graphics drivers and Nahimic that causes GL operations to fail,
causing crashes in operations such as window resizes.
This will close issue #4113--sadly, there is nothing we can do within
GTK to fix the issue.
If gtk_builder_expose_object() is called twice with the same name, it will
result in a g_critical(). This improves that situation by checking for the
object before exposing additional times.
This turns out to be handy in situations where templates are expanded
multiple times, such as application-side implementations of UI merging.
If we get an invalid TARGETS reply, we might not have a valid 'type',
which ends up as NULL and segs in the g_str_equal.
(This is probably fallout from my fix 506566b6a4, which I still
can't reproduce reliably, so the last one just moved the seg a bit
further along, and we still don't know who is sending a bad TARGETS).
This corresponds to:
https://bugzilla.redhat.com/show_bug.cgi?id=2062143
C API users can keep dealing with the implicit equivalence of
GdkFileList and GSList, but language bindings have no idea that one type
is another, and none of them exposes GSList as a type anyway, so they
will need a way to construct a GdkFileList.
Instead of making GdkFileList mutable, and re-implement GSList, we only
provide a constructors pair that lets you create a GdkFileList from a
linked list or from an array.
The gnome-runtime-images have been recently migrated to Quay. This is already reflected in the template.
Please note this MR has been created semi-automatically. If it doesn't make sense, feel free to close it.
Sysprof has a new -Dagent=true build option which allows installing a
/usr/bin/sysprof-agent program (simimlar to sysprof-cli). It provides a
P2P D-Bus API to the process which can control subprocesses. It's used by
IDE tooling to have more control across container boundaries.
However, we do not need it for GTK CI.
Rubberband does not work when initiated past the last row
(warning is printed "Could not start rubberbanding: No item).
Clamp y at the max height of the widgets in the listview
Rubberband does not work when initiated past the last row
(warning is printed "Could not start rubberbanding: No item).
Clamp y at the max height of the widgets in the gridview
Fixes: #3462
The function gtk_grid_view_get_items_in_rect() erroneously calculates
columns less than 0 and greater than n_columns when the user attempts
to rubberband all the way to the left or right respectively. This
causes the rubberband to persistent and creates unexpected behavior.
Limit the rows to a minimum of 0 and maximum of n_columns - 1.
Fixes: https://gitlab.gnome.org/GNOME/gtk/-/issues/3445
DnD under Windows needed 3 fixes to work with Gtk.DropTarget.
1. The droptarget_w32format_contentformat_map list never gets
filled so the gdk_win32_drop_read_async throws
"No compatible transfer format found".
This is an easy fix and done the same way in the win32 clipboard code.
2. After a drop no gdk_drop_emit_leave_event gets emitted.
This causes a second drop to trigger a bunch of assertion
'self->drop == drop' failed because the first drop is still active.
This is also an easy fix and done the same way by the macos backend.
3. Handling gdk_drop_status/gdk_drop_get_actions interaction.
In gtk_drop_target_do_drop the code
```gdk_drop_finish (self->drop, gdk_drop_get_actions (self->drop));```
calls the finish operation with the actions of the drop which triggers
```g_return_if_fail (gdk_drag_action_is_unique (action));```
in gdk_drop_finish. The code assumes that GdkDrop::actions gets
narrowed down by calling gdk_drop_status. This is hard to assure
because at the same time gdk_drop_get_actions is used by
gtk_drop_target_accept to figure out if a drag is accepted.
GdkDrop::actions serves a double purpose here as the supported source
actions and the currently agreed on action. Both the x11 and the
wayland backend get this wrong somewhat too. Under wayland/x11 when
a drag coming from a source that supports both MOVE and COPY is
first hovering a drop target that only supports COPY it is afterwards
no longer accepted by other drop targets only accepting MOVE.
Under x11 this is permanent for this drag but with wayland the drag
recovers when hovering other widgets. The win32 backend now sets the
supported source actions before any enter/move/drop and narrows them
down in gdk_win32_drop_status.
The patch only touches the win32 backend and fixes all three issues,
for me restoring DnD under windows.
Closes#4498
There's a list user_widgets that contains all of the entries and
selections during authentication. This is only freed upon
finalizing the GtkMountOperation. It's possible (and true for the
GVFS SMB implementation) that a MountOperation can have the
gtk_mount_operation_ask_password_do_gtk () function called multiple
times (i.e. bad password). The user_widgets list grows with now
invalid pointers to old widgets (causing unexpected behavior and
seg faults).
Free the user_widgets list upon dialog destruction, we don't need it
anymore.
Fixes: https://gitlab.gnome.org/GNOME/gtk/-/issues/5059
Fixes: https://gitlab.gnome.org/GNOME/gtk/-/issues/5058
It is already explicitly assumed that anonymous authentication will
be used when available, but it is not clear to the user since neither
of the check buttons are selected. Select the Anonymous check button
by default.
When computing a transform value, there is nothing
to do, but we still need to copy the matrix from
src to dest, since it depends on the other transforms
in the array whether we are using the src or the
dest in the end.
This fixes cases like
-gtk-icon-transform: perspective(100px) matrix(1,2,...);
which would otherwise end up with a zero matrix.
This serial should be that from a button press/touch down/etc, use
the last implicit grab here, which will presumably be from the same
device that triggered the event.
Closes: https://gitlab.gnome.org/GNOME/gtk/-/issues/5048
Functions already exist for providing a unique drag action for gdk_drop_finish().
Reuse these functions in the drag_enter/motion callbacks, since they require
a unique action as the return value.
Fixes#3187
The XDND suggested action is a relic from when the source would control
the action for a drop. With the new GtkDropTarget the target decides
the action (not the source). That means the all of the returned
results from the ::enter and ::motion handlers will be unexpectely
ignored. Prefer to use the preferred action over the x11 suggested action.
Fixes: https://gitlab.gnome.org/GNOME/gtk/-/issues/4259
gdk_x11_drop_update_actions() sets actions to
drop_x11->suggested_action when !drop_x11->xdnd_have_actions
and then sets it again to drop_x11->suggested_action if it is.
If xdnd_have_actions is true use xdnd_actions.
For default popover arrow position and default height-for-width layout mode,
natural_width is calculated first with for_size=-1 and orientation=HORIZONTAL,
at the end of gtk_popover_measure() natural_width won't be added with tail_height.
Then to measure with for_size=natural_width and orientation=VERTICAL, obviously
for_size shouldn't be substract with tail_height.
The wrong logic will force content in popover gets less width and then text labels
in popover may get wrapped unnecessarily.
The new content-fit property was wrongly suggesting to manually set
widgets' overflow property, but that property is not really intended to
be set by external code. This commit removes those suggestions and
directly set picture's overflow to be hidden.
It allows to specify the resize mode of the paintable inside the
GtkPicture allocation. This also deprecates the keep-aspect-ratio
property.
Fixes#5027.
We were modifying the removed value before passing
it to the items-changed signal, so we always ended
up with removed == 0 in our signal emission, instead
of passing the original value on, as we should.
Pointed out in !4870
The PangoWeight enum agrees with the numeric values
we use here, so we can do this without a switch and
support numeric weight values at the same time.
Flatpak CI is failing because of unknown option "print-backends".
print-backends was renamed to print in c4d350c260
and subsequently was removed in a4aa6d79ad
(replaced by print-cups and print-cloudprint as auto options)
The width of the left gutter and the height of the top gutter
are now used while computing the child allocations for e.g.
anchors, otherwise - if such a gutter is present - the
widget would be at the wrong position.
Closes#5016
In a list with a visible scrollbar, the scrollbar usually becomes
invisible when the numbers of items is less than the required amount
to scroll. If, however, the list is emptied all at once,
the scrollbar remains. This happens because when there's an empty
list gtk_list_view_size_allocate() returns early before the scrollbar
adjustment is updated.
Given that the list is empty, simply reset the adjustment values
to zero.
Fixes: https://gitlab.gnome.org/GNOME/gtk/-/issues/4370
GtkCheckButton is not derived from GtkToggleButton anymore.
This caused some issues in GtkPrinterOptionWidget which
did not port handling of the button.
Print backends loaded in GtkPrintUnixDialog's load_print_backends()
are not freed later as done in e.g. GtkPageSetupUnixDialog.
This commit destroys and unref those print backends.
Closes#5019
Don't return to the main loop, instead force a run of the paint idle.
The paint idle will know to skip all the phases that aren't requested.
This is critically important becuase gdksurface.c assumes the
FLUSH_EVENTS and RESUME_EVENTS phases are matched, and we cannot
guarantee that if we return to the main loop and let various reentrant
code change the frame clock state.
This would lead to bugs with events being paused and never unpaused
again or even crashes.
Fixes#4941
Something like letter-spacing: -0.5px make a lot of
sense. But we were handling the number as integer
somewhere, loosing the fractional part.
Fixes: #5034
Work harder to find examples for char variation
features, and pull the feature labels out of
the font if possible. This lets us show
meaningful names like "Localised @ and & symbols"
instead of "Stylistic Set 7" or even "ss07".
Add a GtkColumnView scrolling performance test similar to the one used
previously in https://gitlab.gnome.org/GNOME/gtk/-/issues/3334.
The test creates a table with 20 columns and 10,000 rows and scrolls it
to a random position every frame, while measuring the frame times.
There is a commandline flag to pick the cell widget between none (for
benchmarking raw column view scrolling) and various label types. There
is also a commandline switch to disable automatic scrolling in case a
manual assessment is desired. Finally, there's an argument for
controlling the number of columns.
I'm not sure this is API safe, but it is necessary if we want to support
section items and canvas items.
If it's deemed API-unstable, we have to copy this object and deprecate
this one.
This way, we no longer prescribe the use of either GtkListItem or
GtkListItemWidget.
This means we can use it in other places, such as for custom section
header objects or with my Canvas ideas.
With recent updates to GLib, I now see cases where we can hit a state that
has finalized before notify (which will bump the ref count back up). This
is evident in GNOME Text Editor when showing a language submenu from a
popover, and then dismissing the popover and subsequently the tab.
With the previous commit, we at least get a warning like this, which helped
track down the issue.
Gtk-CRITICAL **: gtk_action_observable_unregister_observer: assertion 'GTK_IS_ACTION_OBSERVABLE (observable)' failed
GLib-GObject-CRITICAL **: g_object_ref: assertion '!object_already_finalized' failed
This patch fixes both of those criticals.
Fixes#5009
The menu/action system tends to be incredibly re-entrant, and while fixing
the misuse during finalization cycles should be a priority, this can help
protect just a bit more.
Related #5009
If we take the early return we don't unscale this at the bottom of the
function, causing wrong coordinates in HiDPI screens.
This bug also affects GTK3 (I noticed this running Firefox tests on X).
The GdkToplevelSize struct already has the concept of "bounds", which
means the largest size a window should reasonably have. It's practically
the equivalent of the monitor the window is intended to be mapped on,
with the "struts" (e.g. panels) cut out. It's used by GTK to use this
information to calculate a default window size that is "lagom" (swedish;
not too large, not too small).
This checks mainly that we do the right thing wrt PangoAlignment
weirdness.
0.25 and 0.75 are set to 0.0 and 1.0 currently because of Pango
limitations (and no desire to manually move lines).
But if that were to be fixed, both the ref and the test should update in
the same way and things should just keep working.
Texts usually want the alignment of each row to match the xalign of
the text itself.
Derive the alignment of the PangoLayout from the xalign property of
the inscription. Because Pango doesn't provide float row alignment,
map left, center and right from the xalign in 1 / 3 steps.
We use "label" just like GtkLabel as the two widgets differ in the way
they are measured, but they should be styled the same.
If it turns out we change our opinion on this for specific cases, we
can add style classes later.
Use set_child_visible(FALSE) on those widgets and don't allocate them.
This should usually be the majority of items, so it's quite a worthwhile
addition.
Idea by Ivan Molodetskikh.
Related: #3334
Simplify the API to just return the requirements that the user
has asked for. The rest of the code was undocumented and previously
used as a buggy source for a default value from internal code.
Since the buggy code is now fixed, remove all unnecessary cruft.
There are two reasons for this:
* First, the refactored realize code now makes sure that no
context with unsupported version is ever created.
* Second, this code could bump into false possitives and negatives, since
the user is not requested, nor expected to set_required_version
in any specific order relative to set_allowed_apis. Therefore,
some version could be rejected or accepted based on a set of
allowed apis that the user has not yet correctly configured.
Mimic the behavior of the egl context creation by stablishing
some sane logic for the api and version used. Split the decision
of the type of context (api, legacy) and the creation of a context
of a certain version and all its properties.
By setting and then getting the required version in a context, the code
was not respecting user requirements. Instead, simply get the requested
version by the user clipped by the requirements (display version)
It is useful for backends to get user set preferences while
ensuring the correctness of the result, which will be always
greater or equal than the minimum version provided
GtkGestrureDrag::drag-end can be emitted when the pointer has just
crossed the drag threshold and we have not started the rubberband yet.
This happens if another gesture has claimed the event sequence earlier
in the current event propagation chain.
In such situation, our ::drag-end calls gtk_list_base_drag_update(),
which proceeds to start the rubberband. That's obviously wrong.
Additionally, it also tries to get modifiers from an event it we are
already denied, which obviously fails with criticals:
`gdk_event_get_modifier_state: assertion 'GDK_IS_EVENT (event)' failed`
Thus, if there is no rubberband when we receive ::drag-end, do nothing.
We haven't had any scalable directories in this list.
Add some. Since we seem to have settled on including
just actions and status as subdirectories for each
size, add scalable/actions and scalable/status.
Fixes: #4960
This allows inverting the default text-direction in an application for
debugging, testing, and QA purposes. IDEs such as Builder may automate this
to encourage more application developers to test with a text-direction
different than their own.
If we have a <lookup name="foo" type="SomeInterface"> a runtime warning
would be emitted and the expression would fail to be created. This is
because the interfaces will likely be a GObject as well, meaning we check
the object type branch instead of the interface.
Instead, we need to use the fundamental type like other parts of the
expression system use.
Add "stylus" to the list of substrings in a device name that cause it to be recognized
as a GDK_SOURCE_PEN device (previously "wacom", "pen" and "eraser"). Some devices
just use "stylus" in their name, and are otherwise recognized as
GDK_SOURCE_TOUCHSCREEN instead.
Fixes#4394.
When loading cursors at scale, we expect the
cursor images to have a size of scale * size.
If we don't find such images, load them at their
unscaled size and scale them up ourselves.
Without this, cursors will appear in unexpected
sizes depending on scales and themes.
Related: #4746
On Wayland it is a protocol violation to upload buffers with
dimensions that are not an integer multiple of the buffer scale.
Until recently, Mutter did not enforce this. When it started
doing so, some users started seeing crashes in GTK apps because the
cursor theme ended up with e.g. a 15x16 pixel image at scale of 2.
Add a small sanity check for this case.
Not updating shadow size unconditionally would lead to shadow size not
being set on map, which would lead mutter to think that we are a Window
without extents and then become confused when we suddenly set some.
Make sure that doesn't happen by always having shadows set on map, just
like GTK3.
Fixes#4136
If a context is not realized, calling gtk_at_spi_context_to_ref() will
return a null ref, because its path has not been initialized yet. This
was already done for all other cases in get_parent_context_ref(), but
was missing for the GtkStackPage case.
Fixes: https://gitlab.gnome.org/GNOME/gtk/-/issues/4944
Meson knows all private dependencies itself when passing the library as
first positional argument, no need to specify them manually. Also
simplify backend specific files by simply requiring gtk4, just like
unix-print already did.
This should fix generated gtk4-uninstalled.pc, see Meson bug report:
https://github.com/mesonbuild/meson/issues/10415
It appears that we mess up accounting for blinking
cursors sometimes, and can hit blink_cb when there
is a nonempty selection.
Instead of asserting, warn and stop blinking.
Related: #4767
Ensures modal window is raised above parent; does return focus to
parent window when transient child window is closed.
That solves problems with overlapped modal window and focus loss in
complex multi window UI.
This brings back a subset of what quit-mnemonic.ui tested for, but
trying a lot harder to trigger the label overdrawing its allocation,
which will cause the text to be cut off when clipping is happening.
It should not be an issue at all with GTK4, but keeping that test around
is a good idea.
Instead of asserting only in debug builds (which are generally not
shipped in distributions) we should deliver a critical log-level message
so that these can be found sooner when not developing with jhbuild,
Flatpak, etc.
Also assert that we've setup the state correctly when realizing the
GskGLRenderer object.
Fixes#4625
I can't quite figure out what this test was meant
to test, and how to make it do so in a way that
does not fall afoul of rendering issues in the GL
renderer and rounding differences in pango.
Can't win with reftests.
There were several mistakes here.
The width of subtextures was set to the width of
the main texture, the data size wasn't properly
calculated, and the preconditions were inverted.
Yay us!
Now that we use event controllers we can forward keybindings from the
external entry to the filechooserwidget at the bubble phase.
Fixes#4905
References:
* commit 1fb075dbca
* commit 686116ba61
This can happen if the group can be resolved even when doing the initial
registration of an action as observer will not yet be in the GSList of
watchers (and therefore has no weak references).
Fixes a warning like the following:
g_object_weak_unref: couldn't find weak ref
These were getting created with possible non-zero values and then inserted
into a hashtable where the readers may not know the state of the group.
Ensure those values are set to zero until we assign them below.
Those property features don't seem to be in use anywhere.
They are redundant since the docs cover the same information
and more. They also created unnecessary translation work.
Closes#4904
Previously, there was an issue with glitching after showing/hiding a
popover that was not also destroyed. This was due to the popover having
an update_freeze_count of zero after hiding the surface.
That resulted in it's toplevel continuously dropping frames such as during
high-frame-rate scrolling in textviews. This problem is much more visible
on high-frame-rate displays such as 120hz/144hz.
With this commit, we freeze the frame clock of the popup until it is
mapped again.
Having the initial layout set to VK_IMAGE_LAYOUT_GENERAL causes issues
when going from the final layout to the initial layout since the image
layout is expected to be the general layout. Setting the initial layout
to undefined doesn't have this restriction.
When a non-existing file is selected in the file chooser
for print-to-file, we weren't updating the button label
to show the new filename. Fix that.
Also, use newer file chooser api.
The popover menu previously always pops up in the center of each
row regardless of where the mouse cursor is currently positioned.
Make the popover popup at the current mouse position. If the popover
is triggered by the keyboard (i.e. SHIFT+F10), then align it with the
start of the row.
After right clicking multiple rows, or after adding / removing rows
(i.e. new network locations), right clicking the row will crash
nautilus.
This happens because the popover may become orphan but still expect
a parent.
Reposition the popover menu instead of reparenting it.
After disconnecting a network mount in places (when there's 2 or more
mounts), right clicking another mount crashes the application.
Set row_for_action to NULL when successfully unmounted.
In GTK 3 we used to move the popovers around using set_relative_to();
this is gone in GTK 4 and the apparent direct replacement is setting
the target widget as the new parent.
But this requires a lot of careful handling least the popover become
orphan, which gets us ready to crash at any moment.
Since we only care about positioning the popovers relative to a row,
let's use the set_pointing_to() instead of reparenting. Now, the
sidebar is always the parent.
This commit adds a new meson option -Dupdate_screenshots=true.
When it is enabled, and -Dgtk_doc=true is also used, then the
build will generate images to include in the API docs from
ui files in docs/reference/gtk/images.
Note: we still keep a copy of the images in git, in order to
allow building without a display connection. To update the
images in git, the generated images need to be copied back
from the builddir to the srcdir.
This is a hot path when scrolling a ColumnView, and
g_param_spec_pool_lookup () was taking a measurable part in this hot
path. Instead, notify using pspecs to avoid the name lookup.
Related: https://gitlab.gnome.org/GNOME/gtk/-/issues/3334
freeze/thaw_notify () showed up on the perf trace for rapid ColumnView
scrolling. Track the three properties manually to make it a little
faster.
Related: https://gitlab.gnome.org/GNOME/gtk/-/issues/3334
This looks like a leftover excess invalidation from when the surrounding
code was refactored to not just be called on parent changes but also
when repositioning inside the same parent in commit
507016cafc
Ivan Molodetskikh found this problem in
https://gitlab.gnome.org/GNOME/gtk/-/issues/3334#note_1445873 which
contains a longer analysis of this problem and the performance
reductions it causes.
Related: #3334
Remove the clipping to the widget area that
GtkWidgetPaintable imposes, so we can see shadows
and other out-of-bounds rendering. This is particularly
useful for toplevel windows with client-side decorations.
This allows consumers greater control over the label without the need
to expose each of the label properties as part of GtkCheckButton interface.
Specifically, motivation for this commit is to be able to wrap the label.
Closes#4698
The fixed-size format we use currently can only handle up
to 32768 bytes of string data. If a compose file contains
more, reject it with a warning.
Fixes: #4873
Even if the FileChooserNative instance drops out on us while we're still
waiting for the portal to answer, we should keep the data and pointers
alive until the sequence of asynchronous operations is running. The code
already tries to do that, by acquiring a strong reference to the
GtkFileChooserNative instance, but it's also freeing data as soon as the
dialog is hidden, while asynchronous callbacks that will look at the
fields on that data are still in flight.
To avoid that, we defer freeing the data until the asynchronous
callbacks are invoked, and we keep a reference on the dialog while we're
emitting signals on it.
Fixes: #4883
The enum values are not compatible, and moreover, there is an extra
GTK_WRAP_NONE that PangoWrapMode doesn't have - thus,
pango_wrap_mode_to_string() will assert.
As far as I can tell, Orca does not read the wrap-mode key in the
dictionary for text attributes, anyway.
Fixes: #4869
if the loop for determining max width grows too big, print an error and
abort assuming that a satisfactory value was reached.
This will cause wrong layout and might cause widgets to overlap, but it
will not infloop.
It actually works around and doesn't really fix the primary cause of the
following bugs, but good enough to close them:
Fixes: #4252Fixes: #4517
If we get consecutive preedit string updates that announce a NULL
string, we still do end up issuing ::preedit-changed with those.
Ignore changes from NULL to NULL, it is the other combinations which
must issue this signal.
It looks like os.add_dll_directory() works in a LIFO order, so we call
os.add_dll_directory() from the end of the list of directories in %PATH%
so that the directories are searched in the correct order.
...when we are using Python 3.8.x or later. Python 3.8.x or later on Windows
require one to call os.add_dll_directory() on every directory that contains
dependent non-system DLLs of a module that are not bundled/installed with the
module.
Since we are very likely running programs that rely on dependent items in
%PATH%, make things easier for people by calling os.add_dll_directory() on
all the valid paths in %PATH% in api.py, so that the test will run
successfully on Windows with Python 3.8.x or later.
adwaita-icon-theme has more appropriate icons for showing/hiding text now.
use those, and in the process fix the fact GtkPasswordEntry has been using
them the other way around.
The root accessible object is registered asynchronously, as it needs to
call a method on the AT-SPI registry daemon. This means we need to defer
registering the GtkAtSpiContext on the accessibility bus and in the
cache until after the registration is complete.
Fixes: #4825
Direct access of the fields of the union trips compiler warnings with
GCC 12, such as:
../gtk/gtkimagedefinition.c:135:13: error: array subscript
‘GtkImageDefinition {aka union _GtkImageDefinition}[0]’ is partly
outside array bounds of ‘GtkImageDefinitionEmpty[1]’ {aka
‘struct _GtkImageDefinitionEmpty[1]’} [-Werror=array-bounds]
When changing folders, we were making the select
button insensitive when there is no folder selected.
However, the select button should be usable to
select the current folder.
Fixes#4020
It is very irritating when the entry completion popup
appears not in response to user input in the entry.
In particular, when that happens right as the dialog
is shown.
To prevent that, temporarily disable completion
when setting the entry text programmatically.
When changing folders, we were making the select
button insensitive when there's no files around.
That doesn't make sense in save mode when we don't
want to select a file but create one.
Fixes: #4851
This allows the user to navigate via tab the links in a label and exits
the widget after the last link, when moving forward, and first link,
when moving backward.
This also ensures that ellipsised links arn't focused.
Fixes: https://gitlab.gnome.org/GNOME/gtk/-/issues/4681
The `has-tooltip` property gets set to `false` for label with links if no
link is selected. This makes sure to only change the property to `true`
but never to `false`.
Instead of populating the properties right away (when the widget might
not have been allocated yet, and hence cannot know the right values),
the widget should queue an allocation, where it will populate the
values.
For reasons that only apply to the old serial handling, asking for
the surrounding after IM changes resulted in lazy handling of
commit() afterwards.
With the recent interpretation of serials, this problem became more
apparent, since it is in fact very likely that the last interaction
step after an IM change is notifying of the changed surrounding
text after the IM change was applied.
Make handling of surrounding text similar to caret position changes,
always commit() after the state change, but skip through non-changes.
This makes the compositor state fully up-to-date after an IM change.
The gesture as connected currently on the child GtkText is easily overridden
by the parent editables (and gently done so in the attempt to consume all
clicks).
Connect this gesture to the parent editable widget in these cases, so the
gesture can cohabit with the click-consuming one. It's not part of the same
group, but it won't be abruptly cancelled.
Fixes: https://gitlab.gnome.org/GNOME/gtk/-/issues/4795
Since GdkTimeCoord stores only axis values, prior to this change,
if a device didn't report GDK_AXIS_X or GDK_AXIS_Y, the history
attached to merged motion events wouldn't contain any positional
information.
Commit 6012276093 already addressed
this issue for devices without tools by storing the event position
in GdkTimeCoord using GDK_AXIS_X and GDK_AXIS_Y and augmenting the
GdkTimeCoord's axis bitmask accordingly.
This change generalizes that workaround to all devices. Note that
if a device DOES report values for GDK_AXIS_X and GDK_AXIS_Y, those
values won't be overwritten.
Closes#4809
We now collect this information during node
construction, so use it here.
The concrete change here is that we now avoid
offscreens for container nodes with multiple children,
as long as they don't overlap. In particular, this
avoid offscreens for ellipsized dim labels.
This fixes two issues with the offscreen rendering code for nodes with
bounds not aligned with the pixel grid:
1.) When drawing to an offscreen buffer the size of the offscreen buffer
was rounded up, but then later when used as texture the vertices
correspond to the original bounds with the unrounded size. This could
then result in the offscreen texture being drawn onscreen at a slightly
smaller size, which then lead to it being visually shifted and blurry.
This is fixed by adjusting the u/v coordinates to ignore the padding
region in the offscreen texture that got added by the size increase from
rounding.
2.) The viewport used when rendering to the offscreen buffer was not
aligned with the pixel grid for nodes at coordinates not aligned with
the pixel grid. Then because the content of the offscreen buffer is not
aligned with the pixel grid and later when used as textures sampling
from it will result in interpolated values for an onscreen pixel. This
could also result in shifting and blurriness, especially for nested
offscreen rendering at different offsets.
This is fixed by adding similar padding at the beginning of the
texture and also adjusting the u/v coordinates to ignore this region.
Fixes: https://gitlab.gnome.org/GNOME/gtk/-/issues/3833
When a window is minimized by user action, the `showAndMakeKey` method is not executed when idle. This prevents the window from being un-minimized immediately.
And allow programmatic minimization of a window by un-minimizing them in `_gdk_macos_toplevel_surface_present`
Closes#4811
macos: prohibit fullscreen transition if in transtion
This prevents performing additional fullscreen transitions while
a transition is already in progress.
Closes#4808
See merge request GNOME/gtk!4612
When given an invalid atom, gdk_x11_get_xatom_name_for_display can
return NULL and trigger a seg in gdk_x11_clipboard_formats_from_atoms.
Check for NULL.
Why I'm seeing a bad atom there is probably a separate question.
https://bugzilla.redhat.com/show_bug.cgi?id=2037786
Currently the GtkIMMultiContext may stick to a delegate GtkIMContext
that no longer applies after the multicontext is dissociated from
any widget.
Handle set_client_widget() so that it can handle changes between
widgets from 2 different display, but also so the delegate is made
NULL whenever the context has a NULL widget.
Doing so, any new client widget results in a new delegate IM context
lookup from the right GdkDisplay and GtkSettings, which avoids any
mix up.
Fixes: https://gitlab.gnome.org/GNOME/gtk/-/issues/4805
The call to gdk_win32_clipboard_request_contentformats() can return NULL even
without an error condition being hit (such as when the system clipboard is
empty), so check whether the returned GdkContentFormat pointer is not NULL
before calling gdk_clipboard_claim_remote(), which expects it to be not NULL,
otherwise we face a warning from that funtion and the subsequent
g_object_unref().
This at least partially fixes issue #4796.
We may well be using an EGL context that does not support Desktop (W)GL on
Windows, such as in the case of using libANGLE. So, check whether WGL is
supported for this running instance before trying to query WGL extensions.
This will get rid of warning messages from libepoxy.
Otherwise a stray scroll controller may prevent others from getting hold
events, even if it always propagates scroll events and does absolutely
nothing.
We only need a C compiler and not the whole toolchain,
and gst-plugins-bad was split into libraries and plugins.
pkg-config -> pkgconf.
This should speed the CI setup up a bit.
As documented:
> Overlay children whose alignments cause them to be positioned
> at an edge get the style classes “.left”, “.right”, “.top”,
> and/or “.bottom” according to their position.
Likely accidental regression in b7ee2cbc28
Fixes https://gitlab.gnome.org/GNOME/nautilus/-/issues/2099
WebKit's GTK 4 port can give us textures with an internal format of
GL_RGBA with GL_UNSIGNED_NORMALIZED and a bit-depth of 8. This fixes
warnings for every GdkGLTexture created/delivered to the GskGLRenderer.
The format is essentially the same as GL_RGBA8 since it is normalized
between 0.0..1.0 for 8-bit components.
Fixes#4783
When surface depth switches from non-high-depth to high-depth (or vice
versa) the current surface has to be destroyed before a new one can be
created for this window. eglDestroySurface however was getting passed a
GdkDisplay, rather than the EGLDisplay it expects. As a result the old
surface did not get destroyed and the new surface could not be created
causing rendering to freeze.
Fixes: https://gitlab.gnome.org/GNOME/gtk/-/issues/4773
If using the opacity CSS property the renderer cannot optimize these
handles without the use of offscreens due to the use of both a border
and rgb render node.
Instead, we can apply the alpha to the color values and get the same
effect in a way that the GL renderer can optimize without the use of
offscreen textures for a sizeable reduction in runtime overhead.
The pixel distance could be small enough between tick() calls that
this kind of checks might potentially become a problem. Rely only on
the calculated velocity to trigger the STOPPED phase, and use a lower
threshold to avoid cutting the animation too early.
Related: https://gitlab.gnome.org/GNOME/gtk/-/issues/4725
In order to properly accumulate scroll velocities, we need to keep
the kinetic scroll helpers after we have possibly stopped them
in the process of initiating a further scroll flick.
So, instead of stopping (and destroying) those helpers on scroll-begin,
keep them until the next scroll-end if a scroll was initiated before
kinetic scroll finished. This way we can fetch the last velocity when
calculating the extra kick.
In order to ensure the helpers don't live beyond what it is expected,
we now also remove them after a finished hold event.
Fixes the accumulation of scrolling velocity on consecutive scroll
sequences.
Do not depend on the kinetic scroll helpers existing or not before
exiting the animation, as we may want to keep those a little bit
longer after stopped.
We may want to fetch the last velocity obtained, even though we
preemptively called stop() on a kinetic scroll helper. Keep this
velocity so it can be queried later on.
On the "scroll" signal, the widget uses
gtk_event_controller_scroll_get_unit() to get the
scroll unit.
When the unit is GDK_SCROLL_UNIT_WHEEL, the
behavior is unchanged: the widget scrolls a
certain number of pixels at each wheel detent
click. This number of pixels is determined by the
window dimensions in get_wheel_detent_scroll_step().
When the delta unit is GDK_SCROLL_UNIT_SURFACE, the
widget directly adds the delta to the number of
scrolled pixels no matter the window dimensions.
Add a new GdkScrollUnit enum that represent the
unit of scroll deltas provided by GdkScrollEvent.
The unit is accessible through
gdk_scroll_event_get_unit().
This moves a lot of the texture atlas control out of the driver and into
the various texture libraries through their base GskGLTextureLibrary class.
Additionally, this gives more control to libraries on allocating which can
be necessary for some tooling such as future Glyphy integration.
As part of this, the 1x1 pixel initialization is moved to the Glyph library
which is the only place where it is actually needed.
The compact vfunc now is responsible for compaction and it allows for us
to iterate the atlas hashtable a single time instead of twice as we were
doing previously.
The init_atlas vfunc is used to do per-library initialization such as
adding a 1x1 pixel in the Glyph cache used for coloring lines.
The allocate vfunc purely allocates but does no upload. This can be useful
for situations where a library wants to reuse the allocator from the
base class but does not want to actually insert a key/value entry. The
glyph library uses this for it's 1x1 pixel.
In the future, we will also likely want to decouple the rectangle packing
implementation from the atlas structure, or at least move it into a union
so that we do not allocate unused memory for alternate allocators.
This removes the sharing of atlases across various texture libraries. Doing
so is necessary so that atlases can have different semantics for how they
allocate within the texture as well as potentially allowing for different
formats of texture data.
For example, in the future we might store non-pixel data in the textures
such as Glyphy or even keep glyphs with color content separate from glyphs
which do not and can use alpha channel only.
This allows the gskglprograms.defs a bit more control over how a shader
will get generated and if it needs to combine sources. Currently, none of
the built-in shaders do that, but upcoming shaders which come from external
libraries will need the ability to inject additional sources in-between
layers.
If the max_entry_size is zero, then assume we can add anything to the
atlas. This allows for situations where we might be uploading an arc list
to the atlas instead of pixel data for GPU font rendering.
We were missing the surface offset (e.g. shadows) at the time of expressing
the text caret location in surface coordinates. Add this offset so the
coordinates are as expected by the compositor.
Fixes: https://gitlab.gnome.org/GNOME/gtk/-/issues/4668
These are meant to always redirect events to the grabbing surface,
even for other surfaces of the same client. We weren't doing that
(instead letting the event go through unmodified), fix this handling
so GTK sees the events consistenty.
If a grab is held on a toplevel surface tree, and events happen on a
different surface tree from another toplevel/window group, we rewrite
these events so they look like generated on the window group that
holds the grab, but it missed that coordinates would fail to be
translated, so these would stay unchanged and "pointing" to random
parts of the toplevel that is holding the grab and handling the events.
Since off-surface coordinates are not specially meaningful, and in
fact impossible to obtain in some backends, just fake the coordinates
making it sure that all rewritten events point outside the surface.
The grabbing window will still handle the events, but the coordinates
in these will be harmlessly moot.
Fixes: https://gitlab.gnome.org/GNOME/gtk/-/issues/4760
The simplify and validate commands can function
without a display connection, only preview absolutely
needs one. Allow this, by using gtk_init_check().
The trickery we do with objcopy and ld to speed up
resource inclusion does not seem to work right on
32bit Arm, so just skip it there.
Fixes: #4757, #4748, #4752
When showing the native file chooser, we need to ensure we clear the
sorted surfaces in the display so that we don't risk delivering events
correctly on the next frame.
We had code to do it and it never actually got used correctly. This ensures
that the popup services are attached to the parents so that they get proper
stacking orders when displayed. Additionally, it fixes popups from being
shown as their own windows in Exposé.
If we are clicking through the shadow of a window, we need to take special
care to not raise the old window on mouseUp. This is normally done by the
display server for us, so we need to use the proper API that is public to
handle this (rather than CGSSetWindowTags()). Doing so requires us to
dispatch the event to the NSView and then cancel the activcation from
the mouseDown: event there.
If we closed a key window in response to events, we need to denote another
window as the new key window. This is easiest to do from an idle so that
we don't clobber notification pairs of "did resign"/"did become" key
window.
We have a sorted set of surfaces by display server stacking, so we can
take the first one we come across that is already mapped and re-show it
to become key/main.
If we have server-side decorations we might need to request a layout in
response to the resize notification. We don't need to do this in other
cases because we already handle that in the process of doing the resize
(and that code is that way because of delayed delivery of NSNotification).
If we are using NSWindow titled windows, we don't end up waking up the
frame clock when the window is resized on the display server. This ensures
that we do that after getting a notification of resize.
Ensure that resolution of the subproject occurs via the dependency
interface, not the "poke at subprojects manually" interface, and make
that actually work via --wrap-mode=forcefallback.
There's no need to mark it as not-required and then manually invoke
subproject(), since fallback should work correctly and it is always
needed.
However, if fallback was performed (or forced) it would error out since
get_variable() was instructed to only use pkg-config while the relevant
variable was exported by the subproject as an internal fallback
dependency.
There are cases we might want to consume a NSEvent without creating a
GdkEvent or passing it along to the NSApplication for processing. This
creates a new value we can use and check against to propagate that without
having to do out parameters at the slightly odd invalid pointer value for
a GdkEvent (similar to how MMAP_FAILED is done).
This can get in the way of how we track changes while events are actively
processing. Instead, we may want to delay this until the next main loop
idle and then check to see if we have a main window as the NSNotification
may have come in right after this.
This seems to be a problem since:
https://gitlab.gnome.org/GNOME/gtk/-/merge_requests/3565
To demo the problem, the video demo in gtk4-demo is currently set to
autoplay, but it doesn't autoplay on load as expected because the
"prepared" notification doesn't fire until the user explicitly presses
play.
Similarly if the demo is tweaked to disable autoplay then on loading a
video (or an audio-only ogg) the duration is not known or shown until
the user presses play.
In LibreOffice we want to know what the size of the video is to position
it before the user can interact with it to set it to play. We can
workaround this to some degree by listening to "invalidate-size" on the
GtkMediaStream object which updates for videos, but that doesn't wor
for audio-only streams.
So restore listening to media-info-updated but ignore -1 (which I see
for audio-only where I get -1 and then a useful value) and 0 of the
original report.
see also:
https://gitlab.gnome.org/GNOME/gtk/-/merge_requests/3550GNOME/gtk!4513
We were looking at GtkWidget:has-focus from
event controller signal handlers here, but
the widget property is only changed after
the event controllers.
Update the :has-focus property of the focus
widget when the active status of the window
changes.
We change the property after generating the
GDK_CROSSING_ACTIVE crossing events.
This one can be used for both premultiplied and non-premultiplied alpha
formats, since alpha is always 255. It is useful for opaque PNG upload
on both cairo and GL renderers.
That way, all permutations are possible. Previously it was only useful
in the cairo renderer, which required rgba8 → premultiplied bgra8, while
the GL renderer required rgba8 → premultiplied rgba8. Now both are
available.
This was only useful when building for AArch32 without -mfpu=neon, on
AArch64 or with -mfpu=neon gcc is smart enough to do the auto-
vectorisation, leading to code almost as good as what I wrote in
1fdf5b7cf8.
It appears that NVIDIA does not implement EGL_EXT_swap_buffers_with_damage
on their EGL implementation, but does implement the KHR variant of it.
This checks for a suitable implementation and stores a pointer to the
compatible implementation within the GdkGLContextPrivate struct.
We want to ensure that we recalculate the sort order of windows before
processing the motion. Generally this would be done in response from the
display server in GdkMacosWindow, but I've seen it possible to race there.
We need to handle the case where we might be racing against an incoming
configure event due to how notifications are queued from the display
server. Rather than calling configure (and possibly causing other things
to move around) this just queries the display server directly for the
coordinates that we care about.
Additionally, we can display:NO as we are in control of all the display
process now using CALayer.
We failed to handle the toplevel with transient-for case here which could
cause our X/Y calculations to be off in other areas such as best monitor
detection.
We do actually need the parent frame clock here because it is the way we
ensure that we get layout called for our popup surfaces at the same time
as the parent surface.
This doesn't appear to happen much, but if it does it is nice to setup
the window placement initially. Generally, transient-for is set after
the creation of the toplevel rather than here.
The GdkMacosBuffer object already has storage for tracking the damage
region as it is used in GdkMacosCairoContext to manually copy regions from
the front buffer to the back buffer. This makes the GdkMacosGLContext also
use that field so that we can easily drop old damage regions when the
buffer is lost. This happens during resizes, monitor changes, etc.
This helper is useful to ensure we are consistent with how we keep a
window clamped to the workarea of a monitor when placing windows on
screen. (This does not affect snap-to-edges).
Currently, we have all the plumbing in place so that GTK consumes the
startup notification ID when focusing a window through the xdg-activation
protocol.
This however misses the case that a window might be requested to be
focused with no startup ID (i.e. via interaction with the application,
not through GApplication or other application launching logic).
In this case, we let the application create a token that will be
consumed by itself. The serial used is that from the last
interaction, so the compositor will still be able to do focus prevention
logic if it applies.
Since we already do have a last serial at hand, prefer xdg-activation
all the way over the now stale gtk-shell focusing support. The timestamp
argument becomes unused, but that is a weak argument to prefer the
private protocol over the standard one. The gtk-shell protocol support
is so far left for interaction with older Mutter.
If _gdk_macos_surface_move_resize() was called with various -1 parameters
we really want to avoid changing anything even if we think we know what
the value might be. Otherwise, we risk messing up in-flight operations that
we have not yet been notified of yet.
This improves the chances we place windows in an appropriate location as
they don't et screwed up before window-manager placement.
We need to bring the application to the foreground in multiple ways, and
this call to [NSApp activateIgnoringOtherApps:YES] ensures that we become
foreground before the first window is opened. Otherwise we end up starting
applications in the background.
Fixes#4736
If we are double buffering surfaces with IOSurface then we need to copy
the area that was damaged in the previous frame to the back buffer. This
can be done with IOSurface but we need to hold the read-only lock so that
we don't cause the underlying IOSurface contents to be invalidated.
Additionally, since this is only used in the context of rendering to a
GdkMacosSurface, we know the life-time of the cairo_surface_t and can
simply lock/unlock the IOSurface buffer from begin_frame/end_frame to have
the buffer flushing semantics we want.
To ensure that we don't over damage, we store the damage in begin_frame
(and copy it) and then subtract it from the next frames damage to determine
the smallest amount we need to copy (taking scale factor into account).
We don't care to modify the damage region to swapBuffers because they
already have the right contents and could potentially fall into another
tile anyway and we'd like to avoid damaging that.
Fixes#4735
This can be used to lock a surface for reading to avoid causing the
surface contents to be invalidated. This is needed when reading back from
a front-buffer to the back-buffer as is needed when using Cairo surfaces
to implement something similar to BufferAge.
Previously, a single CVDisplayLink was used to drive updates for all
surfaces across all monitors. It used a 'best guess' rate which would
allow for updates across monitors of mixed rates. This is undesirable for
situations where you might have a 144hz monitor as it does not allow for
reaching up to that frame rate.
Instead, we want to use a per-monitor CVDisplayLink which will fire at the
rate of the monitor down to the level of updates we require. This commit
does just that.
When a surface crosses onto a new monitor, that monitor is used to drive
the GdkFrameClock.
Fixes#4732
Using the mode allows better detection of refresh rate and refresh
interval for the CVDisplayLink bridge to GdkFrameClock. Using it can help
ensure that our 144hz displays can actually reach that rather than falling
back to just 60hz.
This will also need future commits to rework the displaylink source to be
per-monitor.
When the fingers are placed on the touchpad, we get a scroll event with
the phase NSEventPhaseMayBegin. We can use this to synthesize an is_stop
event. This results in the scrolledwindow stopping scroll with stop
gestures.
This can cause another warning as well, however, which should be addressed
from #4730.
Fixes#4733
Windows can end up on different monitors despite having a parent or
transient-for ancestor. We want them to be driven by the CVDisplayLink
for the best-monitor, and so this needs to be unshared.
Currently we're using a display link that is for all active displays which
is just the display server trying to find some timings that try to overlap
as many as possible.
That was fine for a prototype, but we really need to do better for
situations with mixed frame rate (such as 60hz and 120hz promotion
displays). Additionally, the 144hz external monitor I have will never
reach 144hz using the current design.
This is just the first step in changing this, but the goal is to have
one of these attached to each GdkMacosMonitor which we can then use to
thaw surfaces specific to that monitor.
We will eventually be needing additional feedback from the display server
which would be nice to keep away from the rest of GdkMacosDisplay for
cleanliness sake. Particularly for feedback from mission control and other
environment factors that requires private API for proper integration.
This may come from different sources at around the same time, e.g.
a hold gesture while on overshoot. Avoid doing that if an
animation is already set.
Fixes: https://gitlab.gnome.org/GNOME/gtk/-/issues/4730
Instead of performing keyboard layout substitution whenever we find a matching
entry in the registry, first try to load the original layout and only attempt
substitution when that fails.
See #4724
When large viewports are passed to gsk_renderer_render_texture(), don't
fail (or even return NULL).
Instead, draw multiple tiles and assemble them into a memory texture.
Tests added to the testsuite for this.
CI currently fails with "fatal error LNK1318: Unexpected PDB error; OK (0) ''"
Google tells me it might be related to hitting a memory limit. Let's try
disabling debug for now.
There may be various reasons that an application could need access to the
underlying NSWindow that is being used to display the GdkMacosSurface
contents. This provides a minimal API to do that without exposing our
implementation details through public API.
As our rendering system is likely to change over time, we very much want
to keep GdkMacosView, GdkMacosLayer, GdkMacosTile, and GdkMacosWindow all
private implementation details which are subject to change.
As this is public API, we are a bit long-winded with the name so it is
clear what is being accessed without polluting symbol names with things
like "ns" as we used to.
When using server-side-decorations, we need to avoid potential cycles with
compute-size as it may not have the new sizing information yet. We can
just short circuit during "live resize" to get that effect.
Fixes poor window resizing from top-left on titled windows.
This doesn't give us appropriate results if we use the window delegate.
Instead, we need to adjust the frame at the same time we change the
style mask so that we end up in the same location.
Previously we had issues on macos where the overshoot would keep showing.
To fix this we need to actually use discrete events instead of the
generated deltas from macOS in the scroll wheel case. Additionally, we need
to drop the kinetic momentum events from macOS and rely on the gtk kinetic
events which are already happening anyway. We also need to submit the
is_stop event for the GDK_SCROLL_SMOOTH case when we detect it.
To keep the discrete scroll events correct, we need to alter the hack in
gtkscrolledwindow.c to use the same path as other platforms except for
when a smooth scroll event is in place. In the future, I would imagine that
this falls into the boundary of high-precision scrolling and would share
the same code paths as other platforms.
With all of these in place, kinetic scrolling with overshoot appears the
same on macOS as other platforms.
When creating new windows, it is better if we create them with a slight
offset to where they were created before so that they are visible to the
user separately from what they might be overshadowing.
This broke with the previous fixes for initial window positioning. We need
the initial positioning so that tails will be displayed correctly when the
popover surface is displayed.
If the size changes, we need to relayout the tiles. Otherwise we can keep
using what we had before. Generally, that shouldn't happen, but the
previous check was failing in a number of ways.
It looks like, particularly on the M1, we might need to double buffer the
contents of the IOSurface<->OpenGL texture bindings. This doesn't appear
to show up on the Intel macbooks I've tried, but I've seen it in the wild
on an M1.
If we have a 2x scale laptop with a 1x scale external display, we would
need to create a new IOSurface for the external display once it crosses
a boundary, otherwise we won't have something capable of displaying
correctly on the second monitor.
This provides a major shift in how we draw both when accelerated OpenGL
as well as software rendering with Cairo. In short, it uses tiles of Core
Animation's CALayer to display contents from an OpenGL or Cairo rendering
so that the window can provide partial damage updates. Partial damage is
not generally available when using OpenGL as the whole buffer is flipped
even if you only submitted a small change using a scissor rect.
Thankfully, this speeds up Cairo rendering a bit too by using IOSurface to
upload contents to the display server. We use the tiling system we do for
OpenGL which reduces overall complexity and differences between them.
A New Buffer
============
GdkMacosBuffer is a wrapper around an IOSurfaceRef. The term buffer was
used because 1) surface is already used and 2) it loosely maps to a
front/back buffer semantic.
However, it appears that IOSurfaceRef contents are being retained in
some fashion (likely in the compositor result) so we can update the same
IOSurfaceRef without flipping as long as we're fast. This appears to be
what Chromium does as well, but Firefox uses two IOSurfaceRef and flips
between them. We would like to avoid two surfaces because it doubles the
GPU VRAM requirements of the application.
Changes to Windows
==================
Previously, the NSWindow would dynamically change between different
types of NSView based on the renderer being used. This is no longer
necessary as we just have a single NSView type, GdkMacosView, which
inherits from GdkMacosBaseView just to keep the tedius stuff separate
from the machinery of GdkMacosView. We can merge those someday if we
are okay with that.
Changes to Views
================
GdkMacosCairoView, GdkMacosCairoSubView, GdkMacosGLView have all been
removed and replaced with GdkMacosView. This new view has a single
CALayer (GdkMacosLayer) attached to it which itself has sublayers.
The contents of the CALayer is populated with an IOSurfaceRef which
we allocated with the GdkMacosSurface. The surface is replaced when
the NSWindow resizes.
Changes to Layers
=================
We now have a dedicated GdkMacosLayer which contains sublayers of
GdkMacosTile. The tile has a maximum size of 128x128 pixels in device
units.
The GdkMacosTile is partitioned by splitting both the transparent
region (window bounds minus opaque area) and then by splitting the
opaque area.
A tile has either translucent contents (and therefore is not opaque) or
has opaque contents (and therefore is opaque). An opaque tile never
contains transparent contents. As such, the opaque tiles contain a black
background so that Core Animation will consider the tile's bounds as
opaque. This can be verified with "Quartz Debug -> Show opaque regions".
Changes to Cairo
================
GTK 4 cannot currently use cairo-quartz because of how CSS borders are
rendered. It simply causes errors in the cairo_quartz_surface_t backend.
Since we are restricted to using cairo_image_surface_t (which happens to
be faster anyway) we can use the IOSurfaceBaseAddress() to obtain a
mapping of the IOSurfaceRef in user-space. It always uses BGRA 32-bit
with alpha channel even if we will discard the alpha channel as that is
necessary to hit the fast paths in other parts of the platform. Note
that while Cairo says CAIRO_FORMAT_ARGB32, it is really 32-bit BGRA on
little-endian as we expect.
OpenGL will render flipped (Quartz Native Co-ordinates) while Cairo
renders with 0,O in the top-left. We could use cairo_translate() and
cairo_scale() to reverse this, but it looks like some cairo things may
not look quite as right if we do so. To reduce the chances of one-off
bugs this continues to draw as Cairo would normally, but instead uses
an CGAffineTransform in the tiles and some CGRect translation when
swapping buffers to get the same effect.
Changes to OpenGL
=================
To simplify things, removal of all NSOpenGL* related components have
been removed and we strictly use the Core GL (CGL*) API. This probably
should have been done long ago anyay.
Most examples found in the browsers to use IOSurfaceRef with OpenGL are
using Legacy GL and there is still work underway to make this fit in
with the rest of how the GSK GL renderer works.
Since IOSurfaceRef bound to a texture/framebuffer will not have a
default framebuffer ID of 0, we needed to add a default framebuffer id
to the GdkGLContext. GskGLRenderer can use this to setup the command
queue in such a way that our IOSurface destination has been
glBindFramebuffer() as if it were the default drawable.
This stuff is pretty slight-of-hand, so where things are and what needs
flushing when and where has been a bit of an experiment to see what
actually works to get synchronization across subsystems.
Efficient Damages
=================
After we draw with Cairo, we unlock the IOSurfaceRef and the contents
are uploaded to the GPU. To make the contents visible to the app,
we must clear the tiles contents with `layer.contents=nil;` and then
re-apply the IOSurfaceRef. Since the buffer has likely not changed, we
only do this if the tile overlaps the damage region.
This gives the effect of having more tightly controlled damage regions
even though updating the layer would damage be the whole window (as it
is with OpenGL/Metal today with the exception of scissor-rect).
This too can be verified usign "Quartz Debug -> Flash screen udpates".
Frame Synchronized Resize
=========================
In GTK 4, we have the ability to perform sizing changes from compute-size
during the layout phase. Since the macOS backend already tracks window
resizes manually, we can avoid doing the setFrame: immediately and instead
do it within the frame clock's layout phase.
Doing so gives us vastly better resize experience as we're more likely to
get the size-change and updated-contents in the same frame on screen. It
makes things feel "connected" in a way they weren't before.
Some additional effort to tweak gravity during the process is also
necessary but we were already doing that in the GTK 4 backend.
Backporting
===========
The design here has made an attempt to make it possible to backport by
keeping GdkMacosBuffer, GdkMacosLayer, and GdkMacosTile fairly
independent. There may be an opportunity to integrate this into GTK 3's
quartz backend with a fair bit of work. Doing so could improve the
situation for applications which are damage-rich such as The GIMP.
There are situations where our "default framebuffer" is not actually
zero, yet we still want to apply a scissor rect.
Generally, 0 is the default framebuffer. But on platforms where we need
to bind a platform-specific feature to a GL_FRAMEBUFFER, we might have a
default that is not 0. For example, on macOS we bind an IOSurfaceRef to
a GL_TEXTURE_RECTANGLE which then is assigned as the backing store for a
framebuffer. This is different than using gsk_gl_renderer_render_texture()
in that we don't want to incur an extra copy to the destination surface
nor do we even have a way to pass a texture_id into render_texture().
The GtkFileCHooserNativeQuartz injects a NSComboBox into the NSSavePanel
(which is displayed in a remote process since 10.15 whether or not you are
a sandboxed application). The style has changed and we need more space
here to not clip part of the combobox out of view.
I tried every size from 22 to 30 and this seemed to look the most natural
without skewing the location of the text within the combobox.
Previously, the popover would cause the window to go into the :backdrop
state which is not what we want for consistency with other platforms. This
fixes that by walking up the surface chain when we get notified of
loosing or acquiring "key" input from the display server.
If the rendering operation is over an opaque region, we can potentially
avoid clearing a large section of the framebuffer destination. Some cases
you do want to clear, such as when clearing the whole contents as some
drivers have fast paths for that to avoid bringing data back into the
framebuffer.
One may be using IJG libjpeg or libjpeg-turbo to build GTK, and their
build files may or may not generate pkg-config files for us. To make
things easier, we can make use of CMake's built-in support for finding
IJG libjpeg or libjpeg-turbo.
The CMake build files for libtiff may or may not generate pkg-config
files for us, so we can use Meson's CMake support to help us find
libtiff, as CMake has built-in support for finding libtiff.
Add a variable in meson.build that covers Visual Studio-like compilers,
so that we can use it to help us find depedencies using CMake rather
than via pkg-config, where applicable.
Change the existing use case for finding libpng accordingly.
We might have panels with controls in them where the window is running in
another process. The control could have a wrapper window which we would
see from this process. This can happen with the GtkFileChooserNative, but
any NSSavePanel in macOS 10.15+ is out of process (not just sandboxed
applications).
This significantly cleans up how we handle various move-resize, compute-
size, and configure (notification of changes) in the macOS GDK backend.
Originally when prototyping this backend, there were some bits that came
over from the quartz backend and some bits which did not. It got confusing
and so this makes an attempt to knock down all that technical debt.
It is much simpler now in that the GdkMacosSurface makes requests of the
GdkMacosWindow, and the GdkMacosWindow notifies the GdkMacosSurface of
changes that happen.
User resizes are delayed until the next compute-size so that we are much
closer to the layout phase, reducing chances for in-between frames.
This also improves the situation of leaving maximized state so that a
grab and drag feels like you'd expect on other platforms.
I removed the opacity hack we had in before, because that is all coming
out anyway and it's a bit obnoxious to maintain through the async flows
here.
This fixes GTK's NSWindow for toplevels so that they are allowed to enter
fullscreen. We were already handlign the state transitions from the
setStyleMask: halper, but we didn't previously tell the window we are
allowed to transition into that.
There is a bit of a mismatch here in that GTK doesn't have any such flag
that determines if a window is "allowed" by policy to enter fullscreen
since window managers on Linux are free to do that at will.
This makes it easier to figure out those values (which are mentioned in
the GtkApplication documentation) rather than working that out from the
way they're generated (or documented as being generated).
If we have GStreamer on macOS we likely have support for CGL to get an
OpenGL context we can use. This provides the missing pieces to get
accelerated video playback in gtk4-widget-factory working.
This more than halves the total runtime of this function since the
previous commit, from 8.36% to 4.02%, and is most likely memory
bandwidth limited on this specific board now.
I tried to do a SSE2 version as well, but couldn’t find any equivalent
of the LD4/ST4 ARM instruction.
On x86 on a Kaby Lake CPU, this makes it go from 6.63% of the total
execution time (loading some PNGs using the cairo backend) down to
3.20%.
On ARM on a Cortex-A7, on the same workload, this makes it go from 57%
to 8.36%.
We want our tracking area to be limited to the input region so that we
don't pass along events outside of them for the window. This improves the
chances we click-out of a popover with a large shadow.
This still doesn't let us pass-through clicks for large shadows on top-
level windows though.
We only should be asserting in static functions. Furthermore, this function
did not need to have GDK_BEGIN_MACOS_ALLOC_POOL as nothing is being
allocated there which would cause pooling to get used.
This needs to handle the boundary case where the value is exactly equal
to the edge of a rectangle (which gdk_rectangle_contains_point() does not
consider to be containing). However, if there is a monitor in the list
that is a better match, we still want to prefer it.
When using an external mouse on MacOS, the scrolling behavior is
reversed from the user's scrolling preference. Additionally, it is
noticeably sluggish.
This commit fixes both issues by negating the deltas and multiplying
them by 32 before constructing a new scroll event. 32 seems to be the
"traditional" scaling factor according to [Druid], but I'm not sure
where that value actually comes from. Regardless, scaling the deltas by
this amount makes scrolling feel a lot more responsive in the GTK demos.
Scrolling with a trackpad is not affected by either issue because it
triggers a different code path that uses more precise deltas, and
already negates them.
[Druid]: https://linebender.gitbook.io/linebender-graphics-wiki/mouse-wheel#external-mouse-wheel-vs-trackpad
We currently list everything as a dependencies, regardless of whether
it actually is; this is a source of confusion for users that read the
GTK documentation.
Gi-docgen has a new "related" key in the project configuration which
allows us to list libraries that are merely related to the namespace
we are documenting; the "dependencies" key is used to document the
actual namespace dependencies, now.
This was causing us to draw the same background content twice which is a
significant amount of bits to flip in the GPU for maximized windows,
especially twice.
This updates GtkPopover to use the new GtkNative abstraction for
reporting opaque regions of the window, in hopes that it can speed
up compositors for things like animated lists, menu transitions,
and more.
Fixes#4689
This switches to using the new GtkNative machinery for updating the
opaque region. Some amount of local calculation is still required for
determining when we should apply shadows, and this inherits what was
done previous for that.
Related #4689
This abstracts the machinery to update the opaque region for a GtkWindow
into GtkNative so it may be used from other native impelementations such
as GtkPopover.
Related #4689
Some Windows keymaps have bogus mappings for the Ctrl modifier. !4423 attempted
to fix this by ignoring the Ctrl layer, but that was not enough. We also need to
ignore combinations of Ctrl with other modifiers, i.e. Ctrl + Shift. For example,
Ctrl + Shift + 6 is mapped to the character 0x1E on a US keyboard (but it should
be treated as Ctrl + ^). Basically, always ignore Ctrl unless it is used in
conjunction with Alt, i.e. as part of AltGr.
Related issue: #4667
If any of the APIs that assumes that the entry is set already is used
before having one already set, things break pretty badly.
Fixes a downstream issue reported at https://github.com/gtk-rs/gtk4-rs/issues/873
`free` is defined in `stdlib.h`, see for example
<https://pubs.opengroup.org/onlinepubs/009604499/functions/free.html>. Without
this include compilation can fail with the following error:
```
../gdk/loaders/gdkjpeg.c: In function ‘gdk_save_jpeg’:
../gdk/loaders/gdkjpeg.c:264:7: warning: implicit declaration of function ‘free’ [-Wimplicit-function-declaration]
free (data);
^
../gdk/loaders/gdkjpeg.c:264:7: warning: incompatible implicit declaration of built-in function ‘free’
../gdk/loaders/gdkjpeg.c:264:7: note: include ‘<stdlib.h>’ or provide a declaration of ‘free’
../gdk/loaders/gdkjpeg.c:302:67: error: ‘free’ undeclared (first use in this function)
return g_bytes_new_with_free_func (data, size, (GDestroyNotify) free, NULL);
^
../gdk/loaders/gdkjpeg.c:302:67: note: each undeclared identifier is reported only once for each function it appears in
../gdk/loaders/gdkjpeg.c:303:1: warning: control reaches end of non-void function [-Wreturn-type]
}
^
```
We don't want to risk having something really weird come out if we have a
WCG colorspace, so instead only do the performance hack on systems where
the output is likely reasonable.
We will want to eventually just be drawing in the appropriate colorspace,
but that is not available yet.
When using software rendering w/ cairo, assume we're drawing in
the best-monitor's colorspace rather than RGB to avoid colorspace
conversions on every frame.
Instead of relying on cairo_t to perform drawing from our backing
image surface to the Core Graphics context, we can convert the
cairo_image_surface_t into a CGImageRef without having to copy
data if we are certain of the alignment of the image up front.
Without this, there are many situations, based on the size of the
window that could cause cairo to take a slow path and malloc/copy
the data to ensure that alignment.
The previous commit titled "macos: align image surface rowstride to
16-bytes" ensures that this invariant is true so that our drawing
code can assume we can reference the framebuffer from the
cairo_image_surface_t using a CGDataProvider.
Since GdkMacosCairoContext and GdkMacosCairoSubview are coordinating,
we can also setup the transformation/scale early when drawing the
cairo_image_surface_t instead of when copying it to Core Graphics.
Furthermore, the CGImageRef is created with an RGB colorspace so
that we are not performing colorspace conversion to the output
device. We don't get color matching between displays, but we don't
expect that anyway, particularly with the software renderer.
When creating a cairo_image_surface_t we want both the framebuffer pointer
and each row to be aligned to 16-bytes so that Core Graphics will use more
optimal paths.
However, cairo_image_surface_create() will not guarantee that the rowstride
is aligned to 16-bytes so we must do that ourselves.
We need to avoid conflating the managing of frame callbacks from
the freeze/thaw mechanics and ensure we don't perform extra thaw
requests at the wrong time.
Some keymaps on Windows contain bogus mappings for Ctrl+key for certain
keys, e.g. Ctrl+Backspace = Delete, or Ctrl+[ = 0x1B. These are never
used on Windows, so we should ignore them.
Fixes#4667
GTK's old key symbol list is missing a few symbols like the per mille
sign that is included in some keyboard layouts. This commit updates
gdkkeyuni.c to match libxkbcommon's current key symbol list.
From the GCC manpage:
> Wimplicit-fallthrough=5 doesn't recognize any comments as
> fallthrough comments, only attributes disable
> the warning.
So, check for the =5 version after checking for the simple version. This
way we get -Wfallhrough with clang and -Wfallthrough -Wfallthrough=5
with GCC, which works.
Hold gestures are used to bring existing gestures on touchpad
semantically closer to touchscreen gestures.
Touchpad gestures observe hold gestures with a matching amount of
fingers and emit their begin and end signals when fingers are detected
or removed on/from the touchpad.
When a hold cancel event is detected, it is required to wait a few
milliseconds until the next event(s) are received to avoid emitting
multiple begin signals.
Part-of: <!3454>
Handle hold events:
- GDK_TOUCHPAD_GESTURE_PHASE_BEGIN: scroll-begin is emitted.
- GDK_TOUCHPAD_GESTURE_PHASE_END: A hold gesture ends only when all
fingers are lifted from the touchpad without movement, so
scroll-end is emitted right away.
- GDK_TOUCHPAD_GESTURE_PHASE_CANCEL: A hold gesture is cancelled when
some fingers are lifted/put down or movement is detected. In this
case, scroll-end is emitted after a small timeout only if
GDK_SCROLL wasn't detected.
Part-of: <!3454>
Since the addition of GdkEventSequence in touchpad events, these
are now stored in the gesture using that sequence. This bit of touchpad
gesture handling was however missing to be updated, still looking up
the special NULL sequence.
Use the last sequence here, which will be the one coming from touchpad
gesture events.
Despite touchpad gestures having a sequence, these must use the logical
pointer focus. Avoid using the sequence for GtkPointerFocus lookups with
those events, in order to ensure those events make it all the way to the
intended target.
This is fallout from adding GdkEventSequence information to touchpad
gestures.
This change is done for 2 reasons:
- The logic to request this phase when compressing scroll events is
slightly broken. If there are multiple scroll events that are
coalesced into one, the surface frame clock will not get this request.
The worst case is having >= 2 scroll events on every frame, as the
compressed event will be left in the queue, and be further compressed
on future events.
- Even scroll events aside, this phase is requested in oddly specific
places that are not enough to cover all events, others do rely on
unrelated GdkFrameClock activity that happens to flush the events
as well.
Unify this phase request so it explicitly happens on the arrival of any
event. This ensures that events (compressed or not) will be handled
promptly after arrival.
Currently when the widget is realized after the focus in event the input
method isn't activated as enable is never sent. The call trace is
gtk_text_focus_changed ->
gtk_im_context_focus_in ->
gtk_im_context_wayland_focus_in
which returns early as self->widget is NULL since it's set up in
gtk_text_realize() via gtk_im_context_set_client_widget(). Handle that
case by invoking gtk_im_context_focus_in() from gtk_text_realize() too.
A case where the above happens is a GtkSearchEntry in a GtkSearchBar.
E.g. in gtk4-demo when starting the demo and then hitting the search
button right away.
Remember the current context on focus-in even though the text-input
isn't set up yet. This helps in the case where the text-input is not yet
created but a widget already got focused. Without that the enable()
invocation in text_input_enter() woulnd't be invoked leaving the input
method disabled.
This fixes
gtk4-demo --run=search_entry
which would initially not show the on-screen keyboard when e.g using
phoc/sway as compositor.
Tools like gtk4-launch can't set surface on the activation token so
don't require it. If the compositor requires it we can't do anything
about it anyway. This avoids a critical:
(gtk4-launch:23497): Gdk-CRITICAL **: 17:07:24.704: gdk_wayland_surface_get_wl_surface: assertion 'GDK_IS_WAYLAND_SURFACE (surface)' failed
Fixes: be4216e051 ("gdk/wayland: Support the xdg-activation wayland protocol")
Signed-off-by: Guido Günther <agx@sigxcpu.org>
As per Benjamin's suggestions, cleanup the previous implementation on
initializing the GLES context on Windows, so that we use more items that are
already in GDK proper and integrate two functions into one.
Instead of first trying to explicitly ask for a WGL 4.1 context, ask for
the WGL context version that matches what is reported via
epoxy_gl_version(), so that we get the maximum WGL version that is
supported by the graphics drivers, and make sure any WGL contexts that
are shared with this (initial) WGL context are created likewise.
We can try to do a default-bog-standard 3.2 core WGL context creation
if the need arises, but let's leave that alone for now.
The EGL context that we are actually creating must have matching OpenGL/ES
versions and allowed GL API set with the previously-created EGL context
that will be shared with it so that they can interoperate together, if
applicable.
This will fix the situation by making sure that we request for the
OpenGL/ES version and OpenGL API set that match with what we have in our shared
EGL context. Otherwise, the newly-created EGL context assumed a OpenGL/ES 2.0
context that supported desktop OpenGL, which may not be what we wanted, such as
in the case of libANGLE.
We are now able to create EGL contexts properly on Windows, but not GLES. This
tries to fix things by doing the following:
* Record the GL context type in a more proper fashion, using an Enum. This
makes things a bit cleaner.
* Force GLES-3.0+ contexts, since libANGLE requires this to properly work with
the shaders-its 2.0 contexts don't work well with our shaders.
Since now we have the shaders working on Windows under GLES with libANGLE using
a 3.0+ context, drop the check to fall back to the Cairo renderer when GLES is
being used.
We only save the size when we transition from floating to fixed, so that
we can restore the size to the one prior to being fixed.
However, we should not restore to this size whenever we see a 0x0 size
from xdg_toplevel, as it can do that any time it doesn't care about the
size, e.g. when the surface is floating and just changing state.
Fix this by only using the saved size when transitioning from fixed to
floating, not when staying floating while previously floating.
Closes: https://gitlab.gnome.org/GNOME/gtk/-/issues/4634
Calling functions inside a g_assert() means those functions will be
compiled out when building with G_DISABLE_ASSERT.
This fixes the release job in the CI pipeline.
Make GtkIMContextIME ignore ASCII control characters just like other
IMContext implementations (e.G. GtkIMContextSimple). Fixes bogus
characters appearing in text input fields (old bug 676077).
The actual code that does the IM context code handling on Windows now uses the
native Windows APIs to handle keystrokes, so this patch is no longer needed, as
it was found that it instead caused issues.
Pointed out in issue #2865.
This reverts commit fd6ce9975e.
gdk_wayland_toplevel_inhibit_idle() contained a contradictory assert
that always fail. More specifically, in the branch that is supposed to
create the idle inhibitor, there is an assertion that it must already
exist and that the refcount must be greater than zero. This causes a
crash on WMs/DEs that use the ZWP idle inhibit manager protocol such as
KDE Plasma and Sway. Fix this by just asserting that the refcount is
zero instead.
GtkBuilder uses GMarkup, which defines a boolean attribute value as:
- yes/no
- true/false
- 1/0
The ITS file for GtkBuilder UI definitions is only using the first pair,
likely because Glade only ever used those values. GTK's own tools, though,
will typically simplify the full yes/no and true/false strings to 1 and 0,
to minimise the parsing time.
Fixes: #4596
Various transforms are normalized with their next transform, and if they
end up being the identity transform will return NULL.
For example a translation by (x,y,z) and followed by (-x,-y,-z) will
result in NULL.
This makes the hotspot of DND surfaces work when using the Vulkan and
OpenGL renderers.
This bumps the CI image used to the newly built image. This is needed to
install a new enough libwayland-client.so needed for wl_surface.offset.
This is done by adding wayland as a meson subproject, building it
on-demand if the version in the system is not new enough. As
libwayland-client.so is pulled in implicitly when linking to gtk4, the
compile step needs LD_LIBRARY_PATH set to make ld find the right library
to link to.
For some users, GetKeyboardLayoutNameA() returns an alias instead of the
fully resolved keyboard layout identifier. In that case, we have to
query the registry to resolve the alias before we can look up the DLL
path.
See comments under https://gitlab.gnome.org/GNOME/gtk/-/issues/4610
Contrary to what you can read on the internet, SGCAPS keys don't work
by having capslock toggle the KBDCTRL bit, they actually have two
consecutive table entries, the first of which is for the normal
version and the second of which is for the capslocked version.
Background: SGCAPS is short for Swiss German caps because Swiss German
was the first layout to use this feature. For keys with the SGCAPS flag,
capslock has a different effect than pressing shift. For example:
Shift + ü = è, CapsLock + ü = Ü, CapsLock + Shift + ü = È
DLL loading failures should not happen under normal circumstances, but
we should at least try not to crash and and print better diagnostic
messages if they do happen.
See https://gitlab.gnome.org/GNOME/gtk/-/issues/4610
Previously, we treated CapsLock and KanaLock as part of the global
keyboard state, much like NumLock and ScrollLock, rather than using
the supplied modifier mask. This was because GDK does not have a
modifier mask for KanaLock, only for CapsLock, so it would not have been
possible to properly support it.
However, this approach ended up causing problems, with certain keyboard
shortcuts not registering when capslock was active. This was first
observed in Inkscape [0] and appears to affect shortcuts consisting of a
single key (like 'a') with no additional modifiers (wheareas shortcuts
like 'ctrl+a' work).
So now we are using the supplied GDK_LOCK_MASK instead, and dropped
support for KanaLock, which we probably don't need anyway (since regular
text input should be handled by the IME input module -- the keymap is
mainly for shortcuts and keybindings, where you don't really want
KanaLock).
[0] https://gitlab.com/inkscape/inkscape/-/issues/3082
The old code used repeated calls to `ToUnicodeEx` to populate
the translation table, which is slow and buggy. The new code
directly loads the layout driver DLLs from Windows.
See https://gitlab.gnome.org/GNOME/gtk/-/merge_requests/4338
Unref private spool_io of GtkPrintJob before setting it to a new one
in gtk_print_job_set_source_file() and gtk_print_job_set_source_fd()
to prevent a leak.
Fixes: #4627
We are no longer using PangoFT2 APIs in this demo, so make sure that we build
it on all builds since we already depend on a HarfBuzz/Pango version that
provide everything that we need here.
Drop the unnecessary pangofc-font.h include as a result.
Like what was done on gtk/language-names.c, acquire the language names via the
native Windows NLS APIs, eliminating a run-time dependency on iso-codes on
Windows.
Instead of relying on the iso-codes package, use the native Windows NLS APIs to
acquire the localized (translated) language names so that we do not need to
incur an extra runtime dependency on Windows. It's not coverering 100% of the
languages that we would like to support through this, but should cover much of
the things that are required.
We aren't really using PangoFT2 for [language|script]-names.c, and are
always using items from them, so make sure they are being built.
Also always include the pangoft2 dependency in gtk_dep if it is found.
<property name="text">Grumpy wizards make toxic brew for the evil Queen and Jack. A quick movement of the enemy will jeopardize six gunboats. The job of waxing linoleum frequently peeves chintzy kids. My girl wove six dozen plaid jackets before she quit. Twelve ziggurats quickly jumped a finch box.
Разъяренный чтец эгоистично бьёт пятью жердями шустрого фехтовальщика. Наш банк вчера же выплатил Ф.Я. Эйхгольду комиссию за ценные вещи. Эх, чужак, общий съём цен шляп (юфть) – вдрызг! В чащах юга жил бы цитрус? Да, но фальшивый экземпляр!
<property name="text">Grumpy wizards make toxic brew for the evil Queen and Jack. A quick movement of the enemy will jeopardize six gunboats. The job of waxing linoleum frequently peeves chintzy kids. My girl wove six dozen plaid jackets before she quit. Twelve ziggurats quickly jumped a finch box.
Разъяренный чтец эгоистично бьёт пятью жердями шустрого фехтовальщика. Наш банк вчера же выплатил Ф.Я. Эйхгольду комиссию за ценные вещи. Эх, чужак, общий съём цен шляп (юфть) – вдрызг! В чащах юга жил бы цитрус? Да, но фальшивый экземпляр!
Some files were not shown because too many files have changed in this diff
Show More
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
