It's useful to know if you're running the demos from an unstable version
of GTK+, so we should use a CSS fragment that updates the visual
identity of gtk-demo and gtk-widget-factory depending on the version of
GTK+ they are running against.
1. Include the broadway renderer (so we can test it properly fails on
Wayland or X11)
2. List all potential renderers, print useful information when Vulkan
is not compiled in instea dof omitting it
3. Improve docs
Instead of tracking offers in GdkWaylandSelection objects, track the
pending offer in the GdkWaylandSeat and pass it to the GdkDragContext
once we get an enter event.
A container node inside another container node doesn't make a lot of
sense, we can instead just use the parent container node and add the
child container node's children to it directly.
The header linux/input.h used by GDK is specific to Linux. It is
possible to get a few Linux headers on FreeBSD by installing v4l_compat,
but it is usually better to use the one shipped with FreeBSD.
We prefer dev/evdev/input.h to linux/input.h here, so it will always use
dev/evdev/input.h on FreeBSD regardless of v4l_compat.
https://svnweb.freebsd.org/changeset/ports/465644
Broke up a long line, added an empty one, indented another one, and re-aligned
a large amount of function parameter names that got misaligned in some past
refactoring.
https://bugzilla.gnome.org/show_bug.cgi?id=791939
This commit adds support the stable version of the xdg-shell protocol.
Support for the last version of the unstable series is left intact, but
will not receive new features.
The stable version is prioritized above the older version.
https://bugzilla.gnome.org/show_bug.cgi?id=791939
When pressing e.g. a window manager shortcut, which acquires keyboard grab,
Xorg would send FocusOut NotifyGrab then FocusIn NotifyUngrab. Currently
gdk would then deactivate the current surface, which makes accessibility
screen readers think that we have switched to a non-accessible application
and came back again, and thus reannounce the application frame etc. which we
don't want when e.g. just raising volume.
And actually, receiving FocusOut NotifyGrab does not mean losing the
X focus, it only means an application aqcuired a grab, i.e. it is
temporarily stealing keyboard events. On Wayland, this isn't even
notified actually.
This commit makes gdk only deactivate surfaces when there was an actual
focus switch to another window, as determined by has_focus_window (instead
of just has_focus), which happens either normally through FocusOut with
NotifyNormal, or during grabs through FocusOut with NotifyWhileGrabbed.
Fixes#85
gdk_win32_keymap_check_compose() shouldn't be called for
non-W32 displays (i.e. when using broadway or other backends
that could be made to run on Windows).
We no longer set the widget on construction, but instead require an
explicit call to gtk_widget_add_controller().
This way, the reference handling becomes explicit and bindable.
Because gtk_widget_add_controller() is (transfer: full), we don't
even need to unref the controller after adding it.
And we don't need to keep track of it, because controllers get cleaned
up by GtkWidget.
This is the first step towards refactoring how widgets deal with event
controllers.
In the future, the widget will treat controllers the same way it treats
child widgets:
1. The controllers will be created without a widget.
2. There will be gtk_widget_add/remove_controller() functions to add
or remove controllers.
3. The widget will hold a reference to all its controllers.
This way we will ultimately be able to automate controllers with ui
files.
gtk_get_current_event() returns a new reference to the event, it should
be freed across various return branches to avoid the event leak, or we
just fetch the little stuff we're interested in.
Now that all Cairo contexts are ported to managing cairo surfaces
themselves, the old fallback code that didi the managing is no longer
needed.
Also clarify the behavior of gdk_cairo_context_cairo_create() wrt the
vfunc by doing the early exit and the clipping outside of it.
* Remove DC refcounting (we trust GDK to always do
begin_frame/end_frame calls in pairs)
* Now that there's no GDK-provided double-buffer up the stack,
double-buffering is implemented here
(though it's disabled by default - in my tests it didn't provide
any visual improvements, but did decrease performance).
* For some reason delaying window resizes until the point where
we need to blit the double-buffer into the window leads
to visual glitches, so doulbe-buffered windows are resized
in begin_frame, same as non-double-buffered ones.
* New code to clear the paint region, for all drawing modes.
Hopefully, it isn't duplicated anywhere up the stack.
* GL has its own context now, so remove any GL-related comments.
* Layered windows are still used (because cairo actually works
better with them)
* A bit more code re-use for layered windows
* Some functions that were local to gdksurface-win32.c are made
usable for the whole backend
* Drag-indicator drawing is temporarily commented out to match
a similar change in X11 backend
We used to pass 2 regions to GdkDrawCotnext.end_frame() but code was
confusing what they meant. So we now don't do that anymore and only pass
the region that matters: The frame region.
This makes the previous gdk_draw_context_is_drawing() function public
under a new name.
I decided against the old name because we use the term "frame" for a
drawing operation, so I wanted to have this boolean flag reuse the term.
As they require a draw context and the draw context is already bound to
the surface, it makes much more sense and reduces abiguity by moving
these APIs to the draw context.
As a side effect, we simplify GdkSurface APIs to a point where
GdkSurface now does not concern itself with drawing anymore at all,
apart from being the object that creates draw contexts.
Previously, we got the damage, then computed the changed area, then
started a frame with that changed area.
But starting a frame computes the damage for us.
So now we start a frame, then get the damage area from that, then
compute the change area.
Also, split it into its own file - which was the original reason for
looking at this code, the rewrite was an unintentional side effect.
This changes the context to create surfaces on demand.
So whenever the compositor holds onto a surface while GDK wants to
render, it just creates a new surface. If the compositor releases
surfaces, we will retain one for the next frame to be rendered, but free
all extra ones.
This way, we should get to a stage where we have exactly as many
surfaces as needed and never allocate/free any.
Also, don't implement SurfaceClass.ref_cairo_surface() anymore. This
means calls to it will crash now. But as they only happen in the generic
GdkCairoContext implementation, we shouldn't be affected by that.
Plus, once all backends have been ported, that call is going away
anyway.
And make the GdkCairoContext as abstract.
The idea of this and thje following commits is to get rid of all
Cairo code in gdksurface.c (and $backend/gdksurface-$backend.c)
by moving that code into the Cairo context files.
In particular, the GdkSurfaceClass.begin_frame/end_frame()
functions (which are currently exclusively used by the Cairo code
should end up being moved to GdkDrawContextClass.begin/end_frame().
This has multiple benefits:
1. It unifies code between the different drawing contexts.
GL lives in GLContext, Vulkan in VulkanContext and Cairo in
CairoContext. In turn, this makes it way easier to reason about
what's going on in surface-specific code. Currently pretty much
all backends do things wrong when they want to sync to drawing
or to the frame clock.
2. It makes the API of GdkSurface smaller. No drawing code (apart
from creating the contexts) needs to remain.
3. It confines Cairo to the Drawcontext, thereby making it way
more obvious when backends are still using it in situations
where it may now conflict with OpenGL (like when doing the dnd
failed animation or in the APIs that I'm removing in this
branch).
4. We have 2 very different types of Cairo contexts: The X/win32
model, where we have a natively supported Cairo backend but do
double buffering ourselves and use similar surfaces and the
Wayland/Broadway model where we use image surfaces without any
Cairo backend support and have to submit the buffers manually.
By not sharing code between those 2 versions, we can make the
actual code way smaller. We also get around the need to create
1x1 image surfaces in the Wayland backend where we pretend
there's a native Cairo surface.
This does nothing but disallow passing NULL to gdk_surface_begin_paint()
and instead require this context.
The ultimate goal is to split out Cairo drawing into its own source file
so it doesn't clutter up the generic rendering path.
1. Remove set_icons property
2. Make it a GtkWidget subclass
3. Add gtk_fishbowl_set_creation_func()
4. Make the widgetbowl use the new benchmarking infrastructure of the
fishbowl
And of course, gsk_render_node_get_name() is gone, too.
The replacement is of course debug nodes.
As a side effect, GskRenderNode is now *really* immutable.
The shortcuts inhibitors hash table is created when we create a
GdkWaylandWindow implementation for a GdkWindow, and it's destroyed once
we finalize the instance. The fake "root" window we create for the
Wayland display does not have a backing native window, so the shortcuts
inhibitors hash table is set to NULL; this causes a critical error
message when calling g_hash_table_destroy() on it. The finalization of
the root window happens when we close a display connection.
We should use g_clear_pointer(), instead, as it's NULL safe.
Without this change, the displayclose test fails, as all warnings are
considered fatal.
Instead of hard-coding linker flags for a specific operating system and
a specific compiler, we can should cc.links to test them, so they can be
used on more operating systems and compilers.
All of the four platform-dependent backends are enabled by default. It
is usually a good default because it requires users to explicitly choose
backends they want to use. Rules in meson.build also automatically
disable unavailable backends for macOS, Windows, Linux, so users on
these 3 major platforms don't have to manually disable things when
running meson commands.
However, meson.build doesn't do the same thing for other Unix-like
systems, which is acceptable but not ideal. To make it easier to build
GTK+ on these systems, the Linux case, which enables X11 and Wayland and
disables Win32 and Quartz, is made the default for all operating systems
that are not Windows or macOS.
This commit also changes most 'host_machine.system()' calls to os_*
variables, which are easier to read and less likely to be used wrongly.
The shortcuts inhibitors hash table is created when we create a
GdkWaylandWindow implementation for a GdkWindow, and it's destroyed once
we finalize the instance. The fake "root" window we create for the
Wayland display does not have a backing native window, so the shortcuts
inhibitors hash table is set to NULL; this causes a critical error
message when calling g_hash_table_destroy() on it. The finalization of
the root window happens when we close a display connection.
We should use g_clear_pointer(), instead, as it's NULL safe.
Without this change, the displayclose test fails, as all warnings are
considered fatal.
.linked assumes the container is a GtkBox, which is documented as never
flipping children in RTL, so :first-child is always the left child, etc.
GtkBox does that by reordering its CSS nodes when the direction changes.
But most widgets don’t do that, so :first|last-child are 1st/last ADDED
and swap sides in RTL. GtkPathBar is so, and ignoring that in our themes
meant that in RTL, its left/right buttons got each other’s borders. Yuk!
This patch adds the groundwork for supporting widgets like that, via the
%linked_flippable placeholder, and applies that to override buttons in
filechooser .path-bar.linked > button
so that the correct borders get applied to those buttons when using RTL.
Note that I select only PathBars within a FileChooser because we also
have NautilusPathBar, which also uses widget.path-bar – but *does* flip
its nodes for RTL already, so letting that get affected broke it again!
https://bugzilla.gnome.org/show_bug.cgi?id=772817
See the previous commit.
There may be other cases of these being swapped by Gadget conversions,
but hopefully someone else will find and fix those before I have to…
Close https://gitlab.gnome.org/GNOME/gtk/issues/200
Otherwise, if the Popover is destroyed before the MenuButton, the latter
still had a non-NULL but invalid instance and tried to use it in dispose
Close https://gitlab.gnome.org/GNOME/gtk/issues/199
Use g_signal_connect_data() instead of g_signal_connect_object()
to make sure the callback gets disconnected when the data object
is destroyed. This avoids problems in garbage-collected bindings.
https://bugzilla.gnome.org/show_bug.cgi?id=789215
Instead of hard-coding linker flags for a specific operating system and
a specific compiler, we can should cc.links to test them, so they can be
used on more operating systems and compilers.
All of the four platform-dependent backends are enabled by default. It
is usually a good default because it requires users to explicitly choose
backends they want to use. Rules in meson.build also automatically
disable unavailable backends for macOS, Windows, Linux, so users on
these 3 major platforms don't have to manually disable things when
running meson commands.
However, meson.build doesn't do the same thing for other Unix-like
systems, which is acceptable but not ideal. To make it easier to build
GTK+ on these systems, the Linux case, which enables X11 and Wayland and
disables Win32 and Quartz, is made the default for all operating systems
that are not Windows or macOS.
This commit also changes most 'host_machine.system()' calls to os_*
variables, which are easier to read and less likely to be used wrongly.
Well, they don't require a redraw of the widget, because the widget
itself didn't change.
They require a redraw of the parent, because that now displays the
widget in a different position.
And this means we can keep the cache of the widget's render node.
My fishbowl numbers are through the roof^W water surface. Vulkan gets
4000 now.
Due to the few type checks in gtk_widget_get_display(), it was the
slowest part of a call to gtk_widget_query_size_for_orientation if the
in case of a cache hit.
We pulled out the bounds calculation for performance reasons, but the
caller can't know how to properly compute them. Inside gtk+, we can do
that but it's not good enough for public API.
The GVariant we are getting here might not be coming
from GTK+, but rather from some other source. Best to
be forgiving and deal with missing data without crashing.
This was causing the GTK+ portal backends to crash on
print requests from Qt.
It might happen otherwise that a change is recorded in between the
widget dispose and finalization, causing a crash when setting
the visible name for the GtkStack (as that will be NULL at that point)
This will be used in subsequent commits to fix the sign by which the
value is changed in response to directional scroll or keypress events.
The idea is: you have a movement to make – in the form of a delta that
follows widget directions, i.e. −1 means left or up, +1 means right or
down – and you want to know whether that delta needs to be inverted in
order to produce the intuitively expected directional change of :value.
The existing should_invert() is not sufficient: it just determines
whether to invert visually, but we need more nuance than that for input.
To answer that – while not doubling up the work for scrolls and keys – I
add a helper should_invert_move(), which considers other relevant state:
• A parallel movement on priv->orientation should just use the existing
should_invert(), which already worked OK for this case (not others).
• Movements on the other orientation now depend on priv->orientation:
◦ For a horizontal Range, always invert, so up (i.e. −ve in terms of
widget coords) always means increase value & vice-versa. This was
done in get_wheel_delta(), but move it here for use with keys too.
◦ For a vertical Range, ignore :invert as it’s only relevant to the
parallel orientation. Do not care about text direction here either
as RTL locales do not invert number lines, Cartesian plots, etc.
This returns TRUE if the delta should be inverted before applying to the
value, and we can now use this function in both scroll and key handlers.
https://bugzilla.gnome.org/show_bug.cgi?id=407242https://bugzilla.gnome.org/show_bug.cgi?id=791802
If widgets want to clip things, they now need to do it themselves.
By not taking care of clip, we avoid the need to track clip. And by not
tracking clip, we can avoid all unnecessary cache invalidations that we
were doing for render nodes whenever the clip changed.
And when you are scrolling, the clip changes *a lot*.
priv->button is a guint, but we assigned it to a local gint.
gtk/gtkmenushell.c:734:37: warning: comparison between signed and
unsigned integer expressions [-Wsign-compare]
if (button && (new_button != button) && priv->parent_menu_shell)
^
...from CellRenderer::start-editing, to point people in the direction of
info about the lifecycle of the Editable and how to do generic setup.
https://gitlab.gnome.org/GNOME/gtk/issues/154
Drop the line copied from .activate(), replace it with a description of
what this method actually does, and explain what a NULL result means.
https://gitlab.gnome.org/GNOME/gtk/issues/154
* Note in the intro that we're really thinking about temporary widgets
* Mention a gotcha regarding GtkEntry and how ::focus-out stops editing
* Give some examples of what you'd want to do in ::editing-done
* Be a bit more precise about what ::remove-widget represents
* Summarise the lifecycle between Renderer/Editable in .start_editing()
* Emphasise again there that this should be viewed as a temporary widget
https://gitlab.gnome.org/GNOME/gtk/issues/154
When filing a new merge request it's often hard to know who to ask for a
review; using the Git log doesn't always help — the person that touched
a file last may just be fixing the build or a compiler warning.
The `CODE-OWNERS` file format is something that GitHub uses in order to
pre-fill the list of reviewers:
https://help.github.com/articles/about-codeowners/
Ideally, in the future, we'll be able to use this file with a bot like
homu to automatically go through newly filed merge requests and
automatically ask the relevant people for reviews, instead of doing this
manually.
Keep in sync with the current tree.
The changes are mostly caused by updates in the internal hierarchy of
composite widgets, and the fact that the order in which the widget tree
is traversed is not exactly stable.
Instead of having a single massive test running through the a11y
directory, we can split off each individual file into its own unit.
Having individual units has several advantages:
- units are executed in parallel
- it's easier to identify the failing units
- logs for failed units are easier to read
We can't freeze the frame clock on commit, but only after-paint,
otherwise the frameclock will resume in the paint stage.
So freeze the frame clock at the end of the frame if we are waiting for
a frame callback.
Note; The diff is only lage because of indentation changes due to
avoiding early returns in favor of a branch.
GTK does use libintl directly (in gtkmain.c, for example) and thus
needs to be linked to it (if found and/or needed).
Previously we most likely were getting libintl from glib, but
that stopped for some reason. Either way, explicit linking is
the right thing to do here.
Instead of going through an ancillary script to strip away the
`WL_EXPORT` annotation from the generated code, we should bump up the
required version of Wayland, and use the `private-code` argument for
wayland-scanner, which does the right thing for us.
There where some problems (??) with ccache not detecting changes during meson
checks. Setting CCACHE_DISABLE during the meson execution makes ccache not use
the cache and pass things directly to the compiler.
Instead of connecting to / disconnecting from the frame clock, do it
inside the vfuncs next to changing the priv->realized boolean.
This removes a race between those 2 cases that could cause child
widgets' unrealize handlers to reconnect this widget to the frame clock
because it was still marked as realize when the widget had already
disconnected from the frame clock.
Fixes#168
But in turn, also allow it to work on widgets with their own surface.
This way, we can chain up from everywhere and won't have to export
gtk_widget_set_realized().
When using strncpy() with a buffer we need to account for the
terminating NUL character. GCC 8 started warning when using PPD_MAX_NAME
as the buffer length for strncpy() because the buffer we're copying into
has the same length — which means that the terminating NUL may be
skipped if the source string has a length of PPD_MAX_NAME.
The appropriate way to handle the case where we're copying a source with
a length bigger than of PPD_MAX_NAME is, as reported in the strncpy()
documentation, to copy `PPD_MAX_NAME - 1` bytes, and explicitly NUL
terminate the destination buffer. This has the additional benefit of
avoiding the compiler warning.
* Previous commit had misleading info. The code was
added to begin_paint() instead of end_paint(). Though
that did not affect its performance in any visible way.
* Company advised to move the code to an "after_paint" signal
handler, so that it works on all renderers, not just Cairo.
This change caused high fluctuation in FPS values in fishbowl
when it is put in a situation where it cannot achieve 60fps
(such as using Cairo renderer at ultra-high resolution).
This seems to be deliberate and not a bug.
There is no easily apparent way of being notified when frame updates
happene exactly, so we just query frame info at the end of each paint.
If we query too often (faster than DWM refresh rate), we just get
the same values twice in a row, but that is, hopefully, highly unlikely.
Now that queue_draw() isn't restricted to clip anymore, we don't need to
care about clip in the CSS engine either.
We do keep GTK_CSS_AFFECTS_CLIP around though because GtkWindow does
care for the window's size.
When asked for a nonexistent (positive) monitor number,
gdk_x11_display_get_monitor would (at best) return an uninitialized pointer,
instead of returning NULL.
That way, we can store the right region there: The actual painted area
instead of the exposed area (which is way too small).
Also, the GL context is the only user of this data, so storing it there
seems way smarter.
Even widgets with an empty allocation may still want to draw stuff.
Examples include shadows or child widgets with negative margins.
Fixes GtkEntry's progressbar not showing up anymore.
Deferring a bit further making those a standalone controller, make
binding activation happen on run_controllers(), so it happens by
default on widgets (unless the key event was consumed earlier)
without the need of a legacy event controller.
Non gesture controllers have no means to collaborate with other
controllers, thus should be considered standalone entities. It makes
no sense to propagate any further if scroll/key controllers handled
the event.
Use a key controller instead. Note that this currently
breaks the handling of Enter, since we activate key bindings
twice, causing us to switch to the label and back.
This is a GtkGesture done to deal with stylus events from drawing tablets.
Those have a special number of characteristics that extend a regular
pointer, so it makes sense to wrap that.
This may result on the later emission of crossing events, with one of the
sides being already unmapped/unrealized. The widget being unmapped will
result on repick and emission of a set of crossing events anyway.
This event controller is meant to replace usage from key-press/release-event
handlers all through. Optionally it can be set a GtkIMContext, so interaction
is carried by the controller.
1. Pass clip rectangles to gtk_snapshot_push_state() that point into
the state array.
2. g_array_set_size(len+1) the state array
3. Make that function realloc() the state array.
4. The clip rectangle now points into invalid memory
5. Use the clip array
This patch fixes things by moving step 5 to before step 2.
1b9aa1b708 ('a11y: drop the focus tracker') removed a bit too much. We
still have to emit window:activate/deactivate events. They are easy to
emit anyway.
Fixes#127
Overlays are drawings that get rendered on top of the inspected window.
The only overlay in existence so far is the highlight overlay, which is
used to highlight widgets and replaces the "draw" signal handler used
previously.
Instead of just notifying the inspector of what is going to be rendered,
allow the inspector to modify it.
This way, the inspector can overlay information it deems relevant over
the render node while still having access to what the actual widget
(without the inspector) would paint.
If you want to draw a widget to cairo today, you create a widget
paintable, snapshot it to a render node and then draw the render node to
cairo.
And yes, this is that complicated on purpose. Don't draw widgets to
Cairo.
This allows being more specific about the size.
It's useful in particular when the resulting render nodes might be
too small for the size, not only when they are too large. For the
latter case, using a clip node would be enough.
It also requires adding a clip node when rendering the resulting
paintable, but that should be optimized out by GtkSnapshot when not
necessary.
This is actually not just a mechnaism to protect against too many
signals, but it's also a method to getting those signals at the wrong
time.
For every size/content change, a widget needs to invalidate twice:
Once when it queues a resize/redraw (going valid => invalid) and once
when the new size/content is actually assigned (going invalid => valid).
However, one of those invalidations might be inconvenient for the
listener. GtkImage for example does not like receiving
invalidate-contents signals when new contents are assigned, but is fine
with them when the old ones go invalid. And it will not try to draw the
paintable in between anyway.
So by bypassing the 2nd emission if nothing was changed, we can make
GtkImage happy.
This error message is printed into the journal if a GTK app can't
connect to eithre Wayland or X11. Make it at least mention who is not
capable of connecting to a server.
Unrelated, we might want to improve our error reporting when a GTK app
can't start, so that debugging issues with system startup / login get
easier to resolve.
When the clip changes that is passed to a snapshot function, we need to
create eventual cached render nodes because they might not have drawn
their whole area before.
Fixes issues with redrawing when scrolling.
When the max cost for finding a path gets to high, the diff can now be
aborted.
Because render nodes have a fallback method (by just marking the whole
bounds of the nodes as different), we use this to improve performance
of diffs.
This brings fishbowl (which is basically a container node with N images
that change every frame) back to close to previous performance.
Now that we have the full render nodes available, there is not much
benefit in fine-grained control over multiple rectangles.
In particular, it's causing pain with complex regions.
There might be a benefit in clipping to the region's rectangles in cases
like widget-factory where the whole diff is made up of the 2 rectangles
of spinner and the pulsing progress bar, but it needs a good heuristic
for where this is useful.
Instead of calling gdk_surface_invalidate_region(), just
gdk_surface_queue_expose() and rely on the renderer computing the diff
from the previous rendering.
... and diff the previous node with the current one to determine the
clip region.
This doubles the work necessary to track clip regions, but the following
commits will clean that up.
It doesn't need to be exported anymore.
As a side effect, the inspector no longer has any information about the
render region, so remove the code that was taking care of that.
Now that we don't clip the created render nodes anymore, we don't have
to compute the clip region beforehand.
So snapshot the render nodes before initializing the renderer.
This includes a copy of the diff(1) algorithm used by git diff by Davide
Libenzi.
It's used for the common case ofcontainer nodes having only very few
changes for the few nodes of child widgets that changed (like a button
lighting up when hilighted or a spinning spinner).
This requires a bunch of refactorings:
1. Don't pass the current clip region to gtk_widget_snapshot()
so we don't create full widget contents
3. Have a widget->priv->draw_needed that we invalidate on every
queue_draw() call and set on every snapshot()
2. In queue_draw(), walk the widget chain to invalidate the
render nodes of all parents
... and gsk_render_node_can_diff(). Those are vfuncs to compute a region
containing all the pixels that differ between the two nodes.
This is just the plumbing that chains into node classes. No node
implements it yet.
delete_range_cb is set to be called before the text suppression done by
the gtktextlayout (otherwise it does not work properly). But at that
point the cursor position is not yet up to date. We thus need to move
the accessibility cursor notification to after the actual text
suppression, by using another callback.
This fixes cursor position in brltty screen reading.
(cherry picked from commit fa6994d033)
The second parameter of the text-changed::delete event is to be the length,
not the end position. This fixes spurious text removals in brltty
screen reading.
(cherry picked from commit 209f908a03)
Like other widgets, this returns a floating reference, so
(transfer full) is wrong. Just omit the annotation as others do,
thus implying (transfer none).
Close https://gitlab.gnome.org/GNOME/gtk/issues/156
This is meant as an input to the font chooser.
We don't want the user to select a language, but
rather have fonts presented as they would work for
the current language. Therefore, do away with the
lang/script combo on the tweak page.
Really exclude the portions in the gtkfontchooserwidget.c that are built
when HarfBuzz and PangoFT2 are built, and update the Meson files to
exclude such sources as well from the main GTK SO/DLL and from the
gtk4-demo program.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
Remove g_auto*() usage from these sources and use the traditional
g_free(), as g_auto*() are GCCisms (or CLangisms).
Also, don't include unistd.h unconditionally and stop including
langinfo.h and dirent.h, since they seem to be unused.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
The demos are now built as GUI programs, which will require the presence
of WinMain() on Visual Studio builds, unless we specify the entry point.
Pass the /entry:mainCRTStartup linker flag on Visual Studio builds for
the demo programs so that they can link properly.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
For some font features, we can figure out affected
glyphs, and show before/after. For some others, we
hardcode typical sequences.
Still to do: figure out how to find ligatures and
show them.
We now properly use GdkFrameTimings and can therefor check for dropping
even a single frame in the history (of 16 frames or 1/4 of a second).
Once that happens, we immediately stop adding new items to the bowl.
A side effect is that the number of icons is now a *lot* smaller.
This ensures that the frame clock gets updated with correct presentation
times even if nothing was drawn.
This is necessary for benchmarking but would also be relevant for videos
that want to sync to the frame clock but draw frames a lot less.
This commit ensures that each GdkSurface impl remembers the
cursor that GDK sets for it, and that this cursor is set
each time WM_SETCURSOR is called for that sufrace's HWND.
This is needed because W32, unlike X, has no per-window cursors -
the cursor on W32 is a global resource, and we need to keep track
of which cursor should be set when pointer is over which surface
ourselves (WM_SETCURSOR exists exactly for this reason).
This commit also makes GDK remember the surface that has an implicit
grab (since implicit grabs are gone from the upper levels of the toolkit),
and ensures that crossing events are correctly synthesized and the grab
is broken when surface focus changes. This fixes a bug where opening
a new window (by clicking something in some other, pre-existing window)
will make that new window not get any mouse input due to the fact
that the mouse-button-down event from that click caused an implicit
grab on the pre-existing window, and that grab was not released afterward.
Instead of now-unused GdkWin32Cursor class (a subclass of GdkCursor),
add a stand-alone GdkWin32HCursor class that is a wrapper around
HCURSOR handle.
On creation it's given a display instance, a HCURSOR handle and a boolean
that indicates whether the HCURSOR handle can or cannot be destroyed
(this depends on how the handle was obtained).
That information is stored in a hash table inside the GdkWin32Display
singleton, each entry of that table has reference count.
When the GdkWin32HCursor object is finalized, it reduces the reference
count on the table entry in the GdkWin32Display. When it's created,
it either adds such an entry or refs an existing one.
This way two pieces of code (or the same piece of code called
multiple times) that independently obtain the same HCURSOR from the OS
will get to different GdkWin32HCursor instances, but GdkWin32Display
will know that both use the same handle.
Once the reference count reaches 0 on the table entry, it is freed
and the handle (if destroyable) is put on the destruction list,
and an idle destruction function is queued.
If the same handle is once again registered for use before the
idle destructior is invoked (this happens, for example, when
an old cursor is destroyed and then replaced with a new one),
the handle gets removed from the destruction list.
The destructor just calls DestroyCursor() on each handle, calling
SetCursor(NULL) before doing that when the handle is in use.
This ensures that SetCursor(NULL) (which will cause cursor to disappear,
which is bad by itself, and which will also cause flickering if the
cursor is set to a non-NULL again shortly afterward)
is almost never called, unless GTK messes up and keeps using a cursor
beyond its lifetime.
This scheme also ensures that non-destructable cursors are not destroyed.
It's also possible to call _gdk_win32_display_hcursor_ref()
and _gdk_win32_display_hcursor_unref() manually instead of creating
GdkWin32HCursor objects, but that is not recommended.
If GtkExpander:sensitive was FALSE, the arrow still got the normal fg
colour, which made it look clickable, in contrast to the adjacent label.
Fix this by adding selectors to catch the applicable :disabled states.
Note: Needing these may indicate an oops in generic styles elsewhere,
but I couldn’t see any, so let’s just get it looking right for now.
Close https://gitlab.gnome.org/GNOME/gtk/issues/146
Add a new W32 backend-specific message filtering mechanism.
Works roughly the same way old event filtering did, but without
events (events are GDK/X11 concept that never really made sense
on W32), so there's no functionality for 'altering' events being
emitted. If an event needs to be emitted in response to a message
do it yourself.
Implemented like this, it should give better performance than
if we were to use GLib signals for this, since W32 sends a LOT
of messages (unlike X11, which doesn't send events as often)
all the time, and invoking the signal machinery on *each* message
would probably be bad.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
Rename GdkWin32Selection to GdkWin32Clipdrop, since GdkSelection
is mostly gone, and the word "selection" does not reflect the
functionality of this object too well.
Clipboard is now handled by a separate thread, most of the code for
it now lives in gdkclipdrop-win32.c, gdkclipboard-win32.c just uses
clipdrop as a backend.
The DnD source part is also put into a thread.
The DnD target part does not spin the main loop, it just
emits a GDK event and returns a default value if it doesn't get a reply
by the time the event is processed.
Both clipboard and DnD use a new GOutputStream subclass to get data
from GTK and put it into a HGLOBAL.
GdkWin32DragContext is split into GdkWin32DragContext and GdkWin32DropContext,
anticipating a similar change that slated to happen to GdkDragContext.
OLE2 DnD protocol is now used by default, set GDK_WIN32_OLE2_DND envvar to 0
to make GDK use the old LOCAL and DROPFILES protocols.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
The Fedora base image we use for our CI doesn't always keep Meson up to
date with our requirements, so it's better if we just install Meson via
Python's pip.
Examples are not like demos; the latter are installed, and provide a
Flatpak manifest for CI pipelines and GNOME Builder. We should not be
using a single configure time option to gate building both.
With the previous approach we would spend most of the time waiting for
the swapchain to be filled again because it seems the compositor takes
care of 2 images at once from time to time.
This is not visible in profiles because waiting for a frame is a
read/poll/whatever operation that does not take CPU. It's only
noticeable because the app becomes less responsive.
We forgot to account for the case where we lookup for HarfBuzz manually
under Visual Studio builds, so only set HAVE_HARFBUZZ (and thus
HAVE_PANGOFT, since PangoFT2 depends on HarfBuzz) after we did the
fallback check for HarfBuzz.
Also, check for hb.h instead of harfbuzz/hb.h to be inline with the
pkg-config case, as the sources also include the HarfBuzz header by
using #include <hb.h>, not #include <harfbuzz/hb.h>
https://bugzilla.gnome.org/show_bug.cgi?id=773299
This makes the code compile again, though obviously there have been
some substantial changes in how IM contexts work, so it's possible
that IME IM context doesn't work now.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
Check for freetype2 version, because pangoft works with any version
(pangoft availability does not indicate that ft2 is new enough), unlike
GTK.
On Windows, since pangoft is optional, we check for the presence of
freetype2 .pc file first after finding that we have pangoft, and then
check for FT_Get_Var_Design_Coordinates() manually by looking for the
freetype headers and .lib first, and then looking for the presence of
that symbol, since freetype2's Visual Studio build system does not
generate a .pc file for us.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
The main buildscript expects 'print_backends' list to be defined.
Since printbackends is os_unix-only, we need to define this list
ourselves for other OSes.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
Turn the GtkFontChooserLevel field into flags, and
add flags for OpenType variations and features. The
motivation for this is to make font-features in the UI
opt-in, since applications need to support them by
applying the pango attribute.
Meson warns when doing that, as it's not really portable.
Since we're using platform-specific linker flags on Darwin, we can also
do the same on Linux; the syntax is GCC-specific, so we're going to need
Clang users to test it.
Adding the offset node broke serialization in 2 ways:
1. We store the enum value in the node, so make sure to not change it
for existing values
2. The offset node was missing in the deserialization lookup table
Instead of fiddling around with scale in the iconhelper (and getting it
wrong), create a GtkScaler around the paintable that takes care of the
scaling.
This is the equivalent snapshot function to pango_cairo_show_layout().
Not to be confused with gtk_snapshot_render_layout(), which is the
equivalent to gtk_render_layout().
This is a special case of the transform node that does a 2D translation.
The implementation in the Vulkan and GL renderers is crude and just does
the same as the transform node.
Nothing uses that node yet.
When drawing onto a recording surface, source surfaces get cached.
But if we g_free() the surface data after we're done, that cache is
gonna point at invalid data...
If G_ENABLE_CONSISTENCY_CHECKS is defined (i.e. if our buildtype is
'debug'), add a opengl debug callback that prints all debug messages
with a severity higher than SEVERITY_NOTIFICATION as a warning to the
console.
Turns out that GCC errors out when building the GLib test suite, as it
now checks for overflows in allocator functions, and we're testing for
those.
This would not be an issue for GTK, but since we're building GLib as a
subproject, we get failures for those as well.
Until we can find out how to disable errors for subprojects, or fix the
GLib test suite not to trip up warnings in GCC, we're going to live
without compiler warnings treated as errors for a while.
This way, we can postpone the actual rendeing of the node until the
renderer. This allows the renderer to choose the right scale to
render at, so it can decide to use 2x scale for hidpi on its own.
Last but not least, it makes all nodes independent of the context they
are created in, because they do not need to know at snapshot time what
they will ultimately be rendered into.
Set the display for each event that we put.
Also reorganize the dnd_event_put() function a bit, giving it a surface
directly instead of setting it by implication.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
dest_surface is going to always be NULL for source contexts.
Previously we used to put the root window there to pass this check,
but root windows are gone (and root surfaces never existed to begin
with), so we have to adapt.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
This affects gdk_device_query_state() for the virtual device. It has
no window, and is forced to query the display itself, and display
defaults its scale to 1 even for HiDPI desktops. Use the same
"query scale of a NULL monitor" trick that we use in other places
to get the global desktop scale.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
A side effect of vkQueuePresentKHR is the Vulkan implementation calling
wl_surface_commit() on the corresponding Wayland surface. Thus, before
this, we must synchronize the surface state (e.g. opaque region, window
geometry, etc) that changed since last time. Prior to this commit this
was done after calling vkQueuePresentKHR(), causing the surface state to
always correspond to the previous buffer state. As of this commit this
is now done before calling vkQueuePresentKHR(), thus before
wl_surface_commit().
A side effect of eglSwapBuffers* is the EGL implementation calling
wl_surface_commit() on the corresponding Wayland surface. Thus, before
swapping buffers, we must synchronize the surface state (e.g. opaque
region, window geometry, etc) that changed since the last buffer swap.
Prior to this commit, this was done after eglSwapBuffers*, causing the
surface state to always correspond to the previous buffer state. As of
this commit this is now done before swapping the buffers, thus before
wl_surface_commit().
If you want transparent region, you can just render them transparent.
If you want input shaping, use gdk_surface_input_shape_combine_region().
Also remove gtk_widget_shape_combine_region().
... and its implementation in the X11 backend.
GDK does lots of work trying to reduce the region in expose events
so that when the server sends multiple expose events, touching the
same area we can make sure to only redraw stuff once. However:
(1) this is only relevant of there's tons of delay and multiple
expose events get sent
(2) we coalesce multiple events into a single expose event anyway
(3) we do this on the frame clock
But most importantly:
(4) Since the invention of compositing, servers caches all contents
anyway
When building GTK through the CI infrastructure, it would help to have
some ways of testing it; for instance, if we want to verify that theme
changes are useful, or if we want to run the result without necessarily
build it locally.
This is where flatpak comes in handy. By having the CI build a flatpak
buundle, and storing it as an artifact, of the GTK demos, we can easily
point developers and designers to an installable binary that won't break
their system, nor will require development tools and environments to
run.
If a surface is unmapped by the client while gdk is processing updates,
(for example Firefox un-mapping its window on Expose events), the
windowing backend resources might be lost (for example with Wayland)
which can cause a crash in end_paint().
Make sure we drop the cairo surfaces as well when hiding the surface,
that will avoid the crash in gdk_surface_impl_wayland_end_paint() when
trying to attach the staging cairo surface to a released wl_surface,
these will be recreated when needed when the surface becomes visible
again and there is no need to keep such buffers around for a surface
which is not visible anyway.
https://bugzilla.gnome.org/show_bug.cgi?id=793062
xvfb doesn't like C.utf8 and returns XLocaleNotSupported.
While (afaik) C.utf8 and C.UTF.8 should be the same thing, and the former
is returned by locale -a on Fedora, switch to C.UTF-8 to make xvfb happy.
This makes gtk+:gdk tests pass.
When the widget gets finalized it clears the widgetnode and gtk_css_widget_node_get_widget
returns NULL. Guard against gtk_css_widget_node_get_widget() returning NULL like in other
places.
See https://gitlab.gnome.org/GNOME/pygobject/issues/28#note_82862
This is an automated change doing these command:
git sed -f g gtk_widget_set_has_window gtk_widget_set_has_surface
git sed -f g gtk_widget_get_has_window gtk_widget_get_has_surface
git sed -f g gtk_widget_set_parent_window gtk_widget_set_parent_surface
git sed -f g gtk_widget_get_parent_window gtk_widget_get_parent_surface
git sed -f g gtk_widget_set_window gtk_widget_set_surface
git sed -f g gtk_widget_get_window gtk_widget_get_surface
git sed -f g gtk_widget_register_window gtk_widget_register_surface
git sed -f g gtk_widget_unregister_window gtk_widget_unregister_surface
git checkout NEWS*
This is an automatic rename of various things related
to the window->surface rename.
Public symbols changed by this is:
GDK_MODE_WINDOW
gdk_device_get_window_at_position
gdk_device_get_window_at_position_double
gdk_device_get_last_event_window
gdk_display_get_monitor_at_window
gdk_drag_context_get_source_window
gdk_drag_context_get_dest_window
gdk_drag_context_get_drag_window
gdk_draw_context_get_window
gdk_drawing_context_get_window
gdk_gl_context_get_window
gdk_synthesize_window_state
gdk_surface_get_window_type
gdk_x11_display_set_window_scale
gsk_renderer_new_for_window
gsk_renderer_get_window
gtk_text_view_buffer_to_window_coords
gtk_tree_view_convert_widget_to_bin_window_coords
gtk_tree_view_convert_tree_to_bin_window_coords
The commands that generated this are:
git sed -f g "GDK window" "GDK surface"
git sed -f g window_impl surface_impl
(cd gdk; git sed -f g impl_window impl_surface)
git sed -f g WINDOW_IMPL SURFACE_IMPL
git sed -f g GDK_MODE_WINDOW GDK_MODE_SURFACE
git sed -f g gdk_draw_context_get_window gdk_draw_context_get_surface
git sed -f g gdk_drawing_context_get_window gdk_drawing_context_get_surface
git sed -f g gdk_gl_context_get_window gdk_gl_context_get_surface
git sed -f g gsk_renderer_get_window gsk_renderer_get_surface
git sed -f g gsk_renderer_new_for_window gsk_renderer_new_for_surface
(cd gdk; git sed -f g window_type surface_type)
git sed -f g gdk_surface_get_window_type gdk_surface_get_surface_type
git sed -f g window_at_position surface_at_position
git sed -f g event_window event_surface
git sed -f g window_coord surface_coord
git sed -f g window_state surface_state
git sed -f g window_cursor surface_cursor
git sed -f g window_scale surface_scale
git sed -f g window_events surface_events
git sed -f g monitor_at_window monitor_at_surface
git sed -f g window_under_pointer surface_under_pointer
(cd gdk; git sed -f g for_window for_surface)
git sed -f g window_anchor surface_anchor
git sed -f g WINDOW_IS_TOPLEVEL SURFACE_IS_TOPLEVEL
git sed -f g native_window native_surface
git sed -f g source_window source_surface
git sed -f g dest_window dest_surface
git sed -f g drag_window drag_surface
git sed -f g input_window input_surface
git checkout NEWS* po-properties po docs/reference/gtk/migrating-3to4.xml
Rename all *window.[ch] source files.
This is an automatic operation, done by the following commands:
for i in $(git ls-files gdk | grep window); do
git mv $i $(echo $i | sed s/window/surface/);
git sed -f g $(basename $i) $(basename $i | sed s/window/surface/) ;
done
git checkout NEWS* po-properties po
This renames the GdkWindow class and related classes (impl, backend
subclasses) to surface. Additionally it renames related types:
GdkWindowAttr, GdkWindowPaint, GdkWindowWindowClass, GdkWindowType,
GdkWindowTypeHint, GdkWindowHints, GdkWindowState, GdkWindowEdge
This is an automatic conversion using the below commands:
git sed -f g GdkWindowWindowClass GdkSurfaceSurfaceClass
git sed -f g GdkWindow GdkSurface
git sed -f g "gdk_window\([ _\(\),;]\|$\)" "gdk_surface\1" # Avoid hitting gdk_windowing
git sed -f g "GDK_WINDOW\([ _\(]\|$\)" "GDK_SURFACE\1" # Avoid hitting GDK_WINDOWING
git sed "GDK_\([A-Z]*\)IS_WINDOW\([_ (]\|$\)" "GDK_\1IS_SURFACE\2"
git sed GDK_TYPE_WINDOW GDK_TYPE_SURFACE
git sed -f g GdkPointerWindowInfo GdkPointerSurfaceInfo
git sed -f g "BROADWAY_WINDOW" "BROADWAY_SURFACE"
git sed -f g "broadway_window" "broadway_surface"
git sed -f g "BroadwayWindow" "BroadwaySurface"
git sed -f g "WAYLAND_WINDOW" "WAYLAND_SURFACE"
git sed -f g "wayland_window" "wayland_surface"
git sed -f g "WaylandWindow" "WaylandSurface"
git sed -f g "X11_WINDOW" "X11_SURFACE"
git sed -f g "x11_window" "x11_surface"
git sed -f g "X11Window" "X11Surface"
git sed -f g "WIN32_WINDOW" "WIN32_SURFACE"
git sed -f g "win32_window" "win32_surface"
git sed -f g "Win32Window" "Win32Surface"
git sed -f g "QUARTZ_WINDOW" "QUARTZ_SURFACE"
git sed -f g "quartz_window" "quartz_surface"
git sed -f g "QuartzWindow" "QuartzSurface"
git checkout NEWS* po-properties
In certain cases, we might create large cairo nodes, resulting in
surfaces so large, cairo can't handle them. Fix this by limiting the
cairo node to the current clip region.
Save the child info using g_object_set_qdata and just use the widget's
built-in child list for everthing else. This is especially simple for
GtkGrid since it has never supported reordering its child widgets.
This allows to override the role declared to the atk stack. For
instance,
<accessibility>
<role type="static"/>
</accessibility>
allows to tell the accessibility stack that a label is just a message in
a message box.
Fixes#109
is_touchpad_device() for XI2 was hardcoded to look for libinput only.
Extend it slightly to correctly identify other Xorg touchpad drivers.
https://gitlab.gnome.org/GNOME/gtk/issues/97
This allows widget to attach their streams to GdkWindow(s)
The idea is to allow attaching a stream to windowing system(s) so the
stream can make use of its resources, in particular GL contexts.
I am however unsure what to attach to:
- GtkWindow
- GdkWindow
- GtkWidget
- GskRenderer
Each of these provide advantages and disadvantages.
So I'm very much open to better suggestions.
This way, we can support external libraries providing implementations of
GtkMediaFile.
We also add a media backend called 'nomedia' that can be enabled to not
compile any support for GtkMediaFile. This is useful when people want to
statically compile GTK into an application that does not use media.
For now, this option is the default.
We also support a new environment variable GTK_MEDIA that allows
selecting the implementation to use.
GTK_MEDIA=help can be used to get info about the available
implementations.
This is necessary so that bidnings work properly and don't make
gdk_gl_texture_release() a function on GdkTexture.
It also allows code to identify what type of texture they are dealing
with.
Finally, we can now decide to add getters later without screwing
anything up, if we want to allow people to access GL textures directly.
The logic for this in 3 got lost in the move from gadgets to widgets. We
must update the sensitivities when :wrap, :value, or the bounds change.
Close: https://gitlab.gnome.org/GNOME/gtk/issues/88
This used to test that "windowed widgets" and "non-windowed widgets"
handle alpha correctly, but none of the mentioned widgets are windowed
at all anymore. For the rest, this is more easily and dynamically
testable by simply using the inspector.
If @menu_label == NULL, we create a default page->menu_label. This took
@tab_label.get_label() and passed that to page->menu_label.set_text().
This is wrong because we set the plain text of the menu_label from the
rich text of @tab_label. So, if @tab_label used mnemonics or markup, our
menu_label got the raw underline or markup tags shown in it as raw text.
As we call set_text() on the menu Label, the fix is to be symmetric: use
@tab_label’s get_text() as source, as that strips underlines and markup.
It’s not worth making the default Label ‘inherit’ :use-underline/markup;
that’s a slippery slope, and users wanting such things can just create a
fully fledged GtkLabel to pass as @menu_label to suppress the default.
https://bugzilla.gnome.org/show_bug.cgi?id=705509
That's kinda weird but allows us to delete the texture case from
GtkIconHelper and GTK_IMAGE_TEXTURE from the GtkImageType enum.
And it doesn't cause any other problems because the cell renderer
can't deal with paintables - otherwise it would mirror GtkImage and have
a "paintable" property instead.
If set, the image will draw its contents while keeping their aspect
ratio. That means empty areas show up on the top/bottom or left/right.
Also move the special-case snapshotting code for icons to
GtkIconHelper. That's where it belongs.
This only implements the vfuncs, but does not actually emit signals
yet.
It's also not useed for anything other than snapshot() so far, this
will come in later commits.
... instead of going via surfaces.
A side effect is that the pixbuf property is no longer readable because
we have no good way to get the pixbuf back out of the texture, but I
don't think this matters a lot.
If people want to read the pixbuf property, we need to add some code to
make that work.
Instead of loading them into surfaces (which we want to get rid of), we
load into textures.
In fact, we introduce a new paintable subclass called a GtkScaler that
takes care of tracking scaling.
This also ideally gets rid of an extra conversion once renderers learn
to render textures directly.
Use that instead of hacks to guess when an image is considered invalid
according to https://drafts.csswg.org/css-images-4/#invalid-image
Also add a GtkCssImageInvalid that implements the behavior of invalid
images according to the CSS spec so thjat image implementations can
refer to that image.
This makes demos be compiled/linked before tests. And that means that
while hacking, I can already run widget-factory when the tests are
still linking.
This is in preparation for accepting the image type paintable.
It's a bit incovenient because we need more code to track width/height
ourselves (as the paintable no longer does it for us), but it's not too
hard.
GtkIconHelper does not track invalidations on the paintable.
This is a neat trick to get around the circularity between GDK, GSK and
GTK that we inherit with the GdkPaintable interface.
GdkPaintable uses GtkSnapshot
GtkSnapshot creates GskRenderNodes
GskRenderNodes use GdkTextures
GdkTexture will soon implement GdkPaintable
This causes a loop that spans GDK, GSK and GTK and this is the easiest
way to resolve it without breaking bindings (at least that's the idea).
This adds a new GtkStyleAnimation called GtkCssDynamic (for lack of a
better name) that is spawned whenever at least one dynamic value is part
of the GtkCssStyle.
Nobody's doing that. And CSS Images are no longer using Cairo anyway.
If we wanted to support querying them (hint: we don't) we should be
using GdkPaintable.
The GitLab cache is kept across jobs, whether they succeeded or not:
this means that if a compiler check fails during the Meson
configuration, the small compiler program gets cached and restored the
next time the job is run, thus failing again.
the -gtk-scaled() change in the previous commit makes it so
that we now print out the scale factors. Update the expected
output of affected tests to take that into account.
We slightly expand the syntax of -gtk-scaled to allow
specifying an explicit scale after each image, and then
we create a single-image GtkCssImageScaled with the
preferred scale in compute().
And then clip the texture using the current (maybe also rounded) clip.
This way, the result is correct. We don't necessarily have to do the
offscreen drawing in any case, but got the safe route for now.
Realization is done as a side effect of calling
_gtk_entry_completion_resize_popup(), but if this is done before the
GdkScreen of the GtkWindow is set up correctly, it may result in the
widget being unrealized when the screen is updated. This may happen
when the file dialog parent window is not using the default GdkDisplay.
To avoid this issue, realize the popup after the screen has been
properly set up.
Fixes#83 in gtk4
I've rebuilt the new Docker image we use for CI to include GStreamer in
the dependencies.
We really need to have the Docker registry hosted on gnome.org, to avoid
pointing people at Dockerhub.
This is a constructor, and it needs a transfer annotation.
Sadly, the resulting introspection representation is going to be a less
than satisfactory `Gdk.gl_texture_new()`, because there is no such thing
as a GdkGLTexture in the public API.
Now that subtitle's default value "Searching" for OPERATION_MODE_SEARCH
is duplicated as it should be, we cannot reassign other strings to it
anymore, as that resulted in the original dupe of "Searching" leaking.
Fix this by only assigning the dup'd "Searching" after trying to get
more specific values, not before. We therefore need to set it to NULL
during its declaration, and that means we needn't in the final else.
Having a FileChooserDialog in location-entry mode then pressing
<primary>f to move to search mode would crash with an invalid free().
In that case, FileChooserWidget.get_subtitle() returned a static string
straight from gettext. This crashed when the GBinding from :subtitle to
FileChooserDialog’s HeaderBar:subtitle shortly tried to free the string.
Fix by duplicating the string before returning it, like all other paths.
https://bugzilla.gnome.org/show_bug.cgi?id=791004
…the wayland registry.
Wnen _gtk_im_module_get_default_context_id calls
match_backend (context_id) and the default GdkDisplay
is wayland, match_backend() should return TRUE only if
gdk_wayland_display_query_registry (display, "gtk_text_input_manager")
returns TRUE.
A problem with textures is that they can become too big for GPU memory,
which would require tiling. But for tiling we only need to download
the pixels needed by the tile.
Similarly, there might be interest to not upload full textures if a
renderer knows it only needs a small part.
Both of these methods require the ability to specify an area of the
texture to be downloaded. So change the download vfunc to include
this parameter now before we add even more textures later.
A private gdk_texture_download_area() function has also been added, but
nobody is using it yet.
The border is now drawn on the frame node, not using an internal border
node, so we are no longer able to align the label to vertically overlap
the border. The property no longer performs its original purpose, & nor
is it a useful candidate for giving a new role, so no point keeping it.
https://bugzilla.gnome.org/show_bug.cgi?id=778886
When generating introspection data, we instantiate types without
calling gtk_init, so make sure that extension points are registered
before the type is trying to implement them.
So we can use that one when translating event coordinates. Also adapt
the widgetfocus demo to ensure this works.
We should probably at some point delete either the int or the double
version.
GdkMemoryTexture is a texture implementation for holding data in memory
(read: GBytes). You specify the GdkMemoryFormat that data is in and off
you go.
Renderers can use this to add uploads in various different formats and
don't need to fallback to GDK doing the conersion on the CPU.
Supported formats can be extended if we need new ones, for now I just
added the relevant ones for Cairo and GdkPixbuf.
The constructor is also private still, because I'm not sure we want to
export GdkMemoryFormat.
Wrappers that do from_cairo_surface() and for_pixbuf() do exist though.
Put GdkGLTexture into its own file and rename the API to
gdk_gl_texture_foo() instead of gdk_texture_foo_for_gl().
Apart from naming, no actual code changes.
GIO has this facility, so we should use it.
At the same time, make sure the immodules directory
exists, even if we don't install any modules there
outselves.
Now that GtkExpander subclasses GtkContainer instead of GtkBin, it needs
its own guard against adding more than one child.
Also, the documentation should no longer describe adding a child as if
it is descended from GtkBin.
We can't use gtk_widget_get_allocation for either non-anchored widgets
(which happens with the child widget when the expander is unexpanded)
nor toplevel windows since that will include the window decorations.
Fixes#70 in gtk4
We are creating these using g_object_new, so the _new function is never
called, resulting in a NULL mb_charset. Fix this by moving the
initialisation into the _init function.
An alternative GskTextNode constructor that does no text measuring. That
way, we can measure the text before and check if the node will be
outside of the current clip anyway.
This was just testing that text nodes do alpha correctly, but the test
even breaks if the default font is different from the one that was used
to create the reference image, so drop it for now.
gtk_im_context_set_client_widget() allows passing NULL as widget to signal that
the widget no longer exists. The xim implementation didn't handle that
case which led to the test suite on gitlab-ci failing.
The gir XML file contains description of types and functions from
gtk/gtk-a11y.h. Indicate that this header should be included in addition
to gtk/gtk.h in applications written in C. #56
The header got included without config.h being included first which resulted in the
wrong _GDK_EXTERN macro being used. As a result some symbols weren't exported
and starting a DnD action would crash in the linker.
This patch adds config.h includes in all places where clang complained about
_GDK_EXTERN redefinitions.
See #32 for more info.
Do the mouse cursor un-obscuring in the ::motion handler instead of in
the ->event handler. We don't get rid of the GtkWidgetClass::event
handler altogether that way, but it's a step in the right direction.
Add an extension point called gtk-im-module, which requires
the type GtkIMContext. Simplify the loading by using GIO
infrastructure. Drop the locale filtering for now, I don't
think it is really necessary nowadays.
Convert existing platform modules to gio modules.
Sill to do: Drop the conditional build machinery.
Either always include them, or never.
GRAB_BROKEN events cause the ::cancel handler of the gesture to be
emitted, which is where we also call gtk_button_do_release, so this
GtkWidget::event handler is unnecessary.
The given coordinate needs to be trough-relative, since that's what the
slider is relative to. Also use the trough's content size and not the
outer size.
GtkTextView::measure should include the height of the text-layout in its
minimum and natural heights. This fixes scrolling when a text-view has a
scrolled-window ancestor that is not its immediate parent.
Release builds should only disable cast checks, to match what we used
to do back with the Autotools builds.
The Autotools build had an "--enable-debug=no" option, but that was
rarely used; Meson has debug, debugoptimized, release, and plain build
types, but we considered the "release" build types as the old "disable
all debugging code", which is not really accurate.
Disabling assertions and preconditon checks should be left to people
with constrained environments and/or packagers; they are supposed to
use the "plain" build type, and override the CFLAGS themselves.
The code assigning the display to the debug_flags struct gets only
called when the default display changes, which never happens
when there already is one.
This makes it call the change callback in case a display is already
there.
The same fix was applied to gtk3 in !26 where calling gdk_init()
before gtk_init() would trigger this case. With gdk_init() gone
in master this is less likely to happen, but still possible
if gdk_display_open() is called before gtk_init().
See https://gitlab.gnome.org/GNOME/pygobject/issues/166
The Wayland backend was already not supporting this setting
since it is an XSetting that is not backed by a GSetting.
Drop this setting altogether, since we will stop supporting
general-purpose modules.
This exists to exit early for invisible lines. It attempts to use the
LineDisplay’s direction to create a corresponding PangoLayout. However,
the dir is not yet set by this point, & the display was new0()d, so its
dir is always 0 == TEXT_DIR_NONE. Thus, we always create an LTR layout.
Whatever the original intent, this outcome seems to be OK, so let’s make
the code say what it means, rather than using a misleading conditional.
https://bugzilla.gnome.org/show_bug.cgi?id=779099
With autotools the schemas were compiled into each test suite directory
and the tests set GSETTINGS_SCHEMA_DIR to the test build directory.
With meson's gnome.compile_schemas() we can not define a target directory
so just make sure it is built in the gtk directory and set GSETTINGS_SCHEMA_DIR
to the gtk build directory when running the tests.
This makes the gtk+:gtk suite pass when no gtk is installed on the system.
All remaining users of that vfunc now implement snapshot using cairo
render nodes (win32 and radial).
Also, GtkCssImageClass.snapshot is now NULL, so if a subclass doesn't
implement it, it will now crash.
Previously it would try to call the draw vfunc.
This way, we avoid a 1px border at the bottom of the actual searchbar
widget and move it instead to the child of a GtkRevealer. Since we can
now use widgets with 0px height, we finally get rid of the 1px border
that was drawn even if the searchbar child was hidden.
Don't use the current layout size as minimum size anymore, that doesn't
make sense. Also move the code from size_request() from gtk2 into the
now current measure() function.
There's no reason to use a separate file until the format of the file
changes though, as this just means that GTK+ 3.x and GTK+ 4.x
applications would end up showing different bookmarks in the file
chooser.
https://bugzilla.gnome.org/show_bug.cgi?id=793425
Drop the 'enable-' prefix, and always enable all platform-specific
backends. We can disable them depending on the platform. This way,
the documentation printed by `meson configure` remains accurate.
Instead of having separate options for each print backend, we can use
the same approach as the input method modules: a single option, with a
comma-separated list of print backends.
We can call it 'included-immodules', and simplify its logic by always
attempting to split the value, to avoid turning an array into a string
and then back into an array again.
The annotation (allow-none) is wrong. Since
gtk_tree_view_is_blank_at_pos() also calls
gtk_tree_view_get_path_at_pos(), the same fields should have the same
annotations.
Due to the recent changes introduced in glibc 2.27 "%OB" is the
correct format to obtain a month name as used in the calendar
header. The same rule has been working in BSD family (including
OS X) since 1990s. This change is simple but makes GTK+ 4.x require
glibc >= 2.27. If this requirement cannot be fulfilled then we must
cherry-pick the full commit cbf118c from gtk-3-22 branch.
Closes: #9
The internal known_globals hashtable is used to carry accounting for
interfaces that depend on others (as ordering is not guaranteed), extend
its usage so it also keeps track of unimplemented interfaces (here at
least).
The API call will then use this to allow querying the globals offered by
the compositor, it will be useful to determine whether we can use
text-input protocols or should fallback to other IMs.
The CI runner is pretty slow to set up (takes about 6 minutes to get
through the system dependencies needed to build GTK), and does not work
with dependencies as subprojects.
Until we figure out how to make it work, and make it work a bit faster,
we should drop CI and rely on Continuous for a while longer.
We can revert this commit as soon as we find out how to make things
work.
The other method annotations were removed in commit c306e448b3.
There is no introspected ABI change, as g-ir-scanner would just ignore
the annotation.
This eliminates the last warning when building GTK on Linux.
When building dependencies as subprojects we need to tell the
introspection scanner where to find the introspection data; this
means using GIR targets from the subproject.
Functional revert of commit 9c4892f291.
Fixes introspection scanner warnings like:
Warning: Gtk: gtk_drag_finish: Methods must belong to the same
namespace as the class they belong to
That is, the gtk_drag_* functions cannot be methods as they have a
"GdkDragContext" as the instance parameter, and that is not a valid
type for the Gtk namespace.
This is not an introspected ABI change, as the generated introspection
data ignores the annotation.
Signed-off-by: Emmanuele Bassi <ebassi@gnome.org>
https://bugzilla.gnome.org/show_bug.cgi?id=692152
Instead of using `--include-uninstalled` in the scanner arguments, we
can tell Meson to use an introspection target, and it'll do the
appropriate thing for us.
Since part of our type system is resolved at run time, we need to use
run time facilities to get diagnostic messages, like deprecation
warnings for properties and signals.
We should mention the G_ENABLE_DIAGNOSTIC environment variable in the
migration guide, to let developers know how to catch deprecations and
changes while porting their code.
Contrary to what the comments in this function might suggest, it does
not actually do anything about child positions, child child sizes. So,
packing doesn't matter and we don't need to iterate over all child
widgets twice.
And remove the non-NULL checks for minimum_size and natural_size since
these are non-NULL by definition since this function is only called from
measure().
Drop the public filtering API. The x11 backend already has
the ::xevent signal as replacement. The win32 backend needs
a similar signal to replace filtering.
Reshuffle header inclusions in the x11 backend a little bit
to avoid a cyclic inclusion between gdkprivate-x11.h and
gdkdisplay-x11.h that is otherwise causing problems.
- Make the rules for including headers explicit
- Make the symbol visibility rules explicit, and drop the
old "leading underscore" hack
- Drop Private data structure declarations and priv pointers
from public headers
- Mention G_DECLARE_* macros
- Mention `#pragma once`
Calling gtk_menu_item_get_label on a GtkSeparatorMenuItem would
otherwise create a GtkLabel child, increasing the vertical size request
to that of the child label.
Remove all the old 2.x and 3.x version annotations.
GTK+ 4 is a new start, and from the perspective of a
GTK+ 4 developer all these APIs have been around since
the beginning.
Load random strings from /usr/share/dict/words instead of reusing the
same 10 words all the time. That way, we get closer to the real world
use case of having a full alphabet.
The standard Vulkan SDK ships with a pkg-config file, like a modern
library should.
We should fall back to finding the library and header only for platforms
where pkg-config is not really a thing.
Based on a patch by: Daniel Stone <daniels@collabora.com>
https://bugzilla.gnome.org/show_bug.cgi?id=793181
The GDK_POINTER_MOTION_HINT_MASK enumeration value is gone, but we're
still keeping around the "is_hint" field in GdkEventMotion, even though
every backend sets it to `false` — except for the core X11 device
manager.
Avoid the ugly priv->tool_box==NULL check in ::add (and ::remove) by
just not using template xml for this small class. Also, make sure the
GtkBin child is properly set and implement remove to also properly
remove it. Remove the manual widget margins and add some CSS for it.
Also switch to simply using a GtkCenterBox.
This happens when deserializing testcases and it really confuses
valgrind into thinking we're longjmp()ing.
And deserializing rendernodes is slow anyway, so who cares about a few
more malloc()s.
GdkContentFormatsBuilder is currently not introspectable, as it does not
have a GType. We can turn it into a boxed type, but we need to implement
memory management for it.
The current gdk_content_formats_builder_free() function returns a newly
constructed value, so we cannot use it as a GBoxedFreeFunc; additionally
copying a GdkContentFormatsBuilder contents would make it a bit odd, as
you could get multiple identical GdkContentFormats out of the copies.
A simple approach is to model the GdkContentFormatsBuilder API to follow
the GBytes one: use reference counting for memory management, and have
a function to release a reference, return a GdkContentFormats, and reset
the GdkContentFormatsBuilder state.
For language bindings, we can provide a get_formats() function that
returns the GdkContentFormats instance and resets the builder instance,
leaving the reference count untouched.
For C convenience we can keep gdk_content_formats_builder_free(), and
make it a wrapper around gdk_content_formats_builder_get_formats(), with
the guarantee that it'll free the builder instance regardless of its
current reference count.
https://bugzilla.gnome.org/show_bug.cgi?id=793097https://blogs.gnome.org/otte/2018/02/03/builders/
The result won't be visible anyway. This also prevents problems with
widgets that create some resource the size of the widget, like
GtkGLArea. It also keeps us from snapshotting revealers with
size 0.
GDK has a lock to mark critical sections inside the backends.
Additionally, code that would re-enter into the GTK main loop was
supposed to hold the lock.
Back in the Good Old Days™ this was guaranteed to kind of work only on
the X11 backend, and would cause a neat explosion on any other GDK
backend.
During GTK+ 3.x we deprecated the API to enter and leave the critical
sections, and now we can remove all the internal uses of the lock, since
external API that uses GTK+ 4.x won't be able to hold the GDK lock.
https://bugzilla.gnome.org/show_bug.cgi?id=793124
The main GDK thread lock is not portable and deprecated.
The only reason why gdk_threads_add_timeout() and
gdk_threads_add_timeout_full() exist is to allow invoking a callback
with the GDK lock held, in case 3rd party libraries still use the
deprecated gdk_threads_enter()/gdk_threads_leave() API.
Since we're removing the GDK lock, and we're releasing a new major API,
such code cannot exist any more; this means we can use the GLib API for
installing timeout callbacks.
https://bugzilla.gnome.org/show_bug.cgi?id=793124
The main GDK thread lock is not portable and deprecated.
The only reason why gdk_threads_add_idle() and
gdk_threads_add_idle_full() exist is to allow invoking a callback with
the GDK lock held, in case 3rd party libraries still use the deprecated
gdk_threads_enter()/gdk_threads_leave() API.
Since we're removing the GDK lock, and we're releasing a new major API,
such code cannot exist any more; this means we can use the GLib API for
installing idle callbacks.
https://bugzilla.gnome.org/show_bug.cgi?id=793124
We need to have two lists: one, with the list of sources that need to be
introspected; and one with the list of sources that contain only private
symbols.
This reduces the amount of source files that the introspection scanner
needs to traverse, and thus the build time.
After commit ffef28a7e8,
gtk-icon-browser was spewing critical warnings when
changing sections. Avoid that by respecting the return
value of gtk_tree_model_get_iter.
BTN_STYLUS3 is defined by the Linux 4.15 kernel and is sent when the
third button on a stylus is pressed. At the moment, only Wacom's "Pro
Pen 3D" has three stylus buttons. Pressing this button triggers a button
8 event to be sent under X11, so we use the same mapping here.
https://bugzilla.gnome.org/show_bug.cgi?id=790033
We can't just unconditionally create a larger texture here, since the
incoming cairo surface might have a device scale that doesn't fit our
scale_factor. Instead, look up the surface device scale and use that.
We have a couple of Python 3.x scripts that parse C files, and since C
does not have any encoding, we need to force one ourselves, to avoid the
case when we're running the build in a non-UTF-8 locale.
https://bugzilla.gnome.org/show_bug.cgi?id=792497
GtkGesture is a GtkEventController. gtk_event_controller_dispose() calls
_gtk_widget_remove_controller(). That NULLs the pointer-to-Controller in
our EventControllerData but does not delete said ECData from our GList.
Subsequently, if that same Widget gets unparent()ed, that method calls
unset_state_flags(), which leads to doing reset_controllers() if we are
insensitive. Now, unlike most most other loops over the GList of ECData,
reset_controllers() does not skip nodes whose pointer-to-Controller is
NULL. So, we call gtk_event_controller_reset(NULL) and get a CRITICAL.
This surfaced in a gtkmm program. The Gesture is destroyed before the
Widget. The Widget then gets dispose()d, which calls unparent()… boom.
I didn’t find an MCVE yet but would hope this logic is correct anyway:
The simplest fix is to make the loop in gtk_widget_reset_controllers()
skip GList nodes with a NULL Controller pointer, like most other such
loops, so we avoid passing the NULL to gtk_event_controller_reset().
In other, live cases, _gtk_widget_run_controllers() loops over the GList
and removes/frees nodes having NULL Controllers, so that should suffice.
But this clearly was not getting a chance to happen in the failing case.
https://bugzilla.gnome.org/show_bug.cgi?id=792624
This state flag is used in several places in GTK+, for example to
ignore RESIZE_INC hints if tiled. Setting it is also necessary for
backwards compatibility with applications that changed their behaviour
when tiled, such as GNOME Terminal and its MATE fork.
Signed-off-by: Simon McVittie <smcv@debian.org>
https://bugzilla.gnome.org/show_bug.cgi?id=789357
Commit c415bef5de introduced support for the new _GTK_EDGE_CONSTRAINTS
atom. If the compositor supports that atom, however, we were always
setting the tiled state, even if no actual tiling information is
available, where the correct action is to completely remove any traces
of the tiled state.
Fix that by correctly removing the tiled state when compositor supports
_GTK_EDGE_CONSTRAINTS Xatom.
https://bugzilla.gnome.org/show_bug.cgi?id=788516
Otherwise, gtk_widget_get_window returns NULL and we can't successfully
perform a grab via the later gdk_set_grab call. This fixes the entry
completion in the file chooser not working.
Filter models rely on views taking a ref on every node
they care about. GtkIconView was not doing that. Amazingly,
this has never shown up in a bug so far, until I spotted
the fallout in gnome-font-viewer.
Test that filter models propagate ::row-changed if there is
an external reference on the node, and not otherwise. This
is showing up in buggy icon view behaviour, where the icon
view is not redrawing if the content changes in a model that
is below a filter model.
We have to explicitly set some of these to their default values so
expand-set is TRUE and the boxes stop propagating their expansion state
up the hierarchy.
The inspector may hold on to render nodes and textures
beyond the lifetime of the widget (and thus the GL
resources). To handle this situation, allow the widget
to explicitly release the GL resources, and make
the texture available on the clent-side as a cairo
surface. This lets the recorder still show the content
after the widget is gone.
This was causing us to leak, in the following scenario:
1) gtk_widget_destroy is called on a GL area
2) dispose is run and clears the context
3) the GL area is unrealized, but the context is already cleared,
so we leak all the GL buffers
Handle the situation that a GL texture might remain
in use (e.g. by a slow frame, or by the recorder)
In that case, we can't modify it but must use a
new one. Keep a pool of GL textures for this eventuality.
g_list_model_get_item is transfer full, so we need
to drop the references we get from it. This was showing
up while testing the GL texture cache in GtkGLArea.
We already check right before this one whether child->pack != packing
and if so, we continue to the next iteration. So, no need to check again
whether the inverted condition child->pack == packing is true, because
it is.
For the one update_layout_width call in size_allocate, we can just use
the passed-in allocation width instead of a separate (relatively slow)
gtk_widget_get_width call.
Just append a texture node.
Note that this is not 100% done yet. The GL area really
needs to keep a pool of textures, and only reuse them
once the GdkTexture object is gone.
The query function for cursor sizes and capabilities
are not very interesting. At least, they are not used
in GTK+, and all backends but X11 just hardcode
made-up values anyway. So, lets drop them.
We were failing to change the sort order for the
default sort column in some cases. Fix that, and
add a testcase for this issue.
https://bugzilla.gnome.org/show_bug.cgi?id=792459
Add a testcase for the previous fix
We rely on log messages here. Since logging is per-display
now, we need to set a display on our custom icontheme object
to get the expected log messages.
We no longer emit this signal. You can use various gestures
and event controllers instead. If you need to catch raw
motion events, use the generic ::event signal.
This replaces the use of ::button-press-event. There's two
issues with this commit:
1) We don't have a good way to do the equivalent of
gdk_event_triggers_context_menu with gestures
2) We have to defer to and idle to avoid ordering
issues with the treeviews own gestures
This function is misnamed - it is only ever relevant for
windows. And with the ::hide-on-delete property , it is
no longer necessary to use the signal for this simple case.
These signals are no longer used in GTK+ and have a (not quite
perfect yet) replacement with GtkEventControllerMotion.
If you need to catch the raw events, the generic ::event
signal still works.
This lets us replace the ::enter/leave-notify-event handlers.
Not that there is a FIXME here - we currently can't get the
crossing mode from the event controller, so we have to fall
back to gtk_get_current_event().
This lets us replace the ::enter/leave-notify-event handlers.
Not that there is a FIXME here - we currently can't get the
crossing mode from the event controller, so we have to fall
back to gtk_get_current_event().
This signal is unused in GTK+ and configure events are handled
internally by GtkWindow.
If you need to catch configure events, the generic ::event
signal still works.
This signal is not used in GTK+, and has a suitable
replacement with GtkGesture and its subclasses.
If you need to catch the raw touch events, the generic ::event
signal still works, too.
This signal is not used in GTK+, and has a suitable
replacement with GtkEventControllerScroll.
If you need to catch these events, the generic ::event
signal still works, too.
The event is not useful at all, so we are better off
with a signal that doesn't have it, and it is only
relevant on toplevel windows, so we don't need it on
GtkWidget.
With this commit, delete events no longer go through the
::event, ::delete-event, ::event-after widget signals,
but just cause the ::close-request signal on GtkWindow to be
emitted.
I don't think there is a way to get a delete event
on this popup - there's no window decorations, no close
button, etc. So no need to handle ::delete-event.
These functions are entirely trivial, their documentation
is much longer than their implementation, and it contains
an example that is annotated as "don't do this"...
Add a way to toggle debug output on and off
from the inspector. For now, we don't add a
log viewer here, since that has the risk of
deadlock until we've the logging completely
separated by display, and also requires us
to install a log writer function, which
libraries are not supposed to do.
Add a setter for per-renderer debug flags, and use
them where possible. Some places don't have easy access
to a renderer, so this is not complete.
Also, use g_message instead of g_print throughout.
As far as possible, use per-display debug flags.
This will minimize the debug spew that we get from
the inspector if it is running on a separate display.
Since gtk_widget_get_allocation doesn't return x/y values relative to
the GdkWindow anymore, we need to manually translate the widget
coordinates here.
Using get_preferred_size here does not work since it computes the
minimum height for the minimum width, but we want to know the minimum
height for the current width.
event_widget is not modified anymore after the assignment from
handle_pointing event and we need the event's user data set for the
_gtk_window_check_handle_wm_event call.
So the slider does not overlap the value label. Since the value label is
allocated at the widget edge in gtk3, the correct fix here would
probably be to remove the 12px padding we apply to the entire scale and
instead apply it only to the trough.
We should be smarter in picking a good device eventually,
but for now, we just allow to explicitly choose one. To
see a list of all devices, use GDK_VULKAN_DEVICE=list
To specify which device to use, use GDK_VULKAN_DEVICE=<number>
This is a bit of filechooser internals that gets shared with
nautilus, which is fine, but it shouldn't be part of our
public API. There are no other users than nautilus.
When the toolbar style is both-horiz, and the item
is not important, we were not centering the icon in the
same way as in gtk3. The reason is that we overlooked
the expand child property being set to TRUE in this case.
We should not tie the sensitivity of the select button
to the tweak action, since there may be fonts which are
selectable, but not tweakable.
Instead, enable the select button when a font is selected,
as it should be.
These can't be returned as part of the font description,
so we need new api for them. For now, this is just readonly
properties. Maybe these should be writable too, eventually.
Add a button the dialog's header bar that lets us
switch to a second page where we can customize
the selected font.
Make the font chooser widget export an action that the
dialog can use for the button. This has some advantages:
- we can export not just the toggle state, but also enabled
- we can reuse the same enabled state to make the select
button insensitive when no font is selected
To determine whether a font is selected, listen to changes
of the list selection. And ensure that the font chooser is
in an initial state when mapped, even if we close the dialog
from the tweak page.
Both GtkWidget and GtkContainer had similar docs regarding hfw/wfh
geometry management. Move these just to GtkWidget. Also make sure the
examples compile, port everything from gtk_preferred_* to measure and
replace some occurrences of "container" with "widget" where container
was just used to refer to a widget with child widgets.
This test was not updated to using a draw func instead
of the ::draw signal yet. At the same time, make it use
::size-allocate instead of ::configure-event.
Instead of creating a new one for every ident, name and string, just
create one GString and reuse it. This means the GString we keep around
will grow to the maximum size of any ident, name or string we parse,
which is still not terribly large.
Commit 4ee02725b4 made the :hover apply to
the title node, not the arrow node, but the selectors it added were not
caught by the recent commits fixing the specificity of title > arrow.
We don't want a pointer that is moved off a scrollbar
to trigger a row when it gets released. To avoid this,
require an explicit opt-in to handling unpaired-releases.
1) Add a --compare cmd line switch that lets people compare the normal
render path and the render_texture path.
2) Add a -o cmd line switch that lets people render the given .node
file to a texture and save that texture to the given png file name.
Putting a combobox in an expander was causing the combo arrow
to go sideways. Increase the specificity with which we address
the expander arrow to avoid that.
The documentation and annotations for some of the print API is defined
in platform-specific source files, so we need to ensure we're passing
those files to the introspection scanner in order to avoid warnings, and
to get the appropriate introspected API.
We already ceil() the given float texture sizes here, so if they are
valid, the result should definitely be > 0. Textures with size 0 can't
be properly used, especially not as render targets, where they will
trigger an assertion failure later in a glCheckFramebuffer call.
Text nodes will almost always end up using the exact same texture and
the same program. So, in that case we can simply add vertex data for all
the characters we need to draw and use just one draw call.
Render nodes can end up with bounds < 1 since they are floats, and the
implicit cast to int ends up creating a texture with 0 width or height.
Use ceil() instead in create_texture so we don't have to do that on the
caller side everywhere.
This way, we can also clip the created node bounds to the current clip
of the GtkSnapshot. This works as long as we don't modify the start and
end points, and happens all the time while rendering.
Clipping a color node is trivial so we do it here directly since that
might later save the entire clip node as well as freeing the fragment
shaders from coloring lots of pixels that will be clipped away.
The only time a style-updated indicates we need
to reload fonts is when it is synthesized by GtkSettings
in response to a fontconfig timestamp change, but
we are listening to those already, anyway.
Some emoji fonts (such as Emoji One), render Emoji sequences
such as some of the family variations using multiple individual
glyphs. This rendering is too wide and breaks our grid layout.
Therefore, we will just skip any sequence whose rendering is
more than twice as wide as a simple smiley.
Instead, pass the source window to gdk_drag_begin().
Also make Wayland use this window instead of the one under the pointer
(though those 2 Windows are most likely the same anyway).
The function was referencing itself in ways that bamboozled gcc:
static void
foo (void)
{
g_signal_handlers_disconnect_by_func (NULL, foo, NULL);
}
Well done, function, you get your own commit!
The code was asserting something that was not always holding
true. We can hit row == NULL here on page-up too. Handle that
case by moving to the first row.
https://bugzilla.gnome.org/show_bug.cgi?id=791549
Without selections, drags can't have them either.
Also included is removing the selection from GtkSelectionData.
Includes a bunch of crude cleanups to Wayland code that no longer has to
care about selection atoms.
VkImage contains a reference to the VkDeviceMemory and, because
the current code frees the VkDeviceMemory before destroying the
VkImage that references it, a warning is triggered by the validation
layers.
This is not critical, since we release both resources at the same
place. But the warning triggered by the validation layers sums up
adding 1 MB per second of extra debug logging, making the debugging
process much more painful.
This commit simply swaps the destruction order, and destroys the
VkImage first, then the now unused VkDeviceMemory.
There is a gtk_event_controller_scroll_set_flags() call that's meant
to be called after construction (eg. due to scrolledwindow relayouts
hiding/showing scrollbars). The property shouldn't be construct-only
for consistence.
In the motion compression phase the coalesced events will be saved
as a GdkTimeCoord on the motion event that shall be delivered.
For simplicity (and because history doesn't make much sense otherwise)
event history is only recorded while there are buttons pressed, this
also tidily ensures that those coalesced events would have the same
target widget on the gtk side than the delivered one, because of
implicit grabs.
Two warts remain. gdk_event_copy() should be unnecessary as
events should be considered static after delivery, so g_object_ref()
should be just as good. There's a few exceptional cases that the event
is copied and then modifier for later processing, those cases should be
reconsidered individually.
And gdk_event_free() could be likewise turned into g_object_unref(),
many callers remain though.
Now all events structs are private, it doesn't make as much sense
having GdkEventPrivate wrapping allocating events. This is a first
step towards removing it.
It won't stand true anymore that the GdkEventType argument is the
first field of the GdkEvent* structs. All callers have been updated
to use event->any.type instead.
Instead of just passing the GdkContentFormats, we are now passing the
GdkContentProvider to gdk_drag_begin().
This means that GDK itself can now query the data from the provider
directly instead of having to send selection events.
Use this to provide the private API gdk_drag_context_write() that allows
backends to pass an output stream that this data will be written to.
Implement this as the mechanism for providing drag data on Wayland.
And to make this all work, implement a content provider named
GtkDragContent that is implemented by reverting to the old DND
drag-data-get machinery inside GTK, so for widgets everything works just
like before.
Scrolling a path bar is of marginal usefulness - you need to
find a really deep place in your filesystem hierarchy in order
to scroll one or two places at best. And the code we had for
this was not working. And it was using legacy event handlers.
Instead of fixing it, remove it.
We now have a GdkX11Display::xevent signal that gets emitted for every
XEvent and allows you to interrupt processing via TRUE/FALSE return
values.
These return values to correspond to GDK_FILTER_REMOVE and
GDK_FILTER_CONTINUE respectively.
The GDK_FILTER_TRANSLATE case from gdk_window_add_filter() is now meant
to be handled via gdk_display_put_event().
This is in preparation for DND.
It moves a lot of code from gdkclipboard-x11.c to
gdkselectionoutputstream-x11.c to untangle it from GdkX11Clipboard
usage.
This code was doing horrible things, and the atk documentation
for the focus tracking feature says that this is deprecated and
not used anymore. So lets not do it.
Mark the following signals as deprecated:
event, event-after, button-press-event, button-release-event,
touch-event, scroll-event, motion-notify-event, enter-notify-event,
leave-notify-event, property-notify-event, selection-clear-event,
selection-request-event, selection-notify-event, selection-received,
selection-get, proximity-in-event, proximity-out-event. Most
of these have suitable replacements in event controllers and
gestures already. The selection-related signals will soon be
irrelevant when selection handling moves to GDK.
Set G_ENABLE_DIAGNOSTIC=1 to see deprecation
warnings for uses of these signals.
Instead, pass the actions as part of gdk_drag_begin() and insist DND is
always managed.
A new side effect is that gdk_drag_begin() can now return %NULL.
We were allocating the progress bar to the full size
of the entry. This made entry icons loose their cursors,
since they were 'covered' by the progress bar, even though
it doesn't draw anything there.
We need to notify ATK the description changed when the tooltip text associated
with the widget changes and gtk_widget_accessible_get_description() would use
it as the description.
https://bugzilla.gnome.org/show_bug.cgi?id=779009
This fixes stuttering in animations that rely on the regularity of
gdk_frame_clock_get_frame_time.
https://bugzilla.gnome.org/show_bug.cgi?id=787665
BEFORE
gdkgears:
58 FPS and visibly stuttering
gnome-maps on a 59.95Hz monitor:
"paint" g_get_monotonic_time +17278μs, gdk_frame_clock_get_frame_time +17278μs
"paint" g_get_monotonic_time +17449μs, gdk_frame_clock_get_frame_time +17426μs
"paint" g_get_monotonic_time +17620μs, gdk_frame_clock_get_frame_time +17600μs
AFTER
gdkgears:
60 FPS and smoother
gnome-maps on a 59.95Hz monitor:
"paint" g_get_monotonic_time +18228μs, gdk_frame_clock_get_frame_time +16680μs
"paint" g_get_monotonic_time +15010μs, gdk_frame_clock_get_frame_time +16680μs
"paint" g_get_monotonic_time +17134μs, gdk_frame_clock_get_frame_time +16680μs
This is the replacement for selection usage.
Backend implementations for X11 (missing support for backwards compat
formats like COMPOUND_TEXT) and Wayland are included.
GTK code should be adapted to use gdk_drop_read_*() functions instead
of gtk_drag_get_data().
The fix is twofold. First, when checking that a corner is resizable, we must
check the constraints on both edges. Second, when checking either edge we
must include both perpendicular sides in order to allow those to be
resizable when the constraint does not allow resizing the edge being
checked.
When looking for the cursor to apply, start from the innermost
widget and go up. This is the right behavior for cases like
entry icons. The top-down order we were using so far is the
right behavior for cases like global wait cursors. Since we
have entry icons in gtk, but not global wait cursors, lets
pick the other order for now.
Color matrix nodes as the child of other color matrix nodes can happen
quite frequently as a result of CSS. To ease the renderer
implementations, collapse chains of color matrix nodes into one.
In order to map a window with the correct initial parent-child
relationship when a modal dialog is set up to be a child of an imported
foreign window, the relationship must be set up before the window is
mapped.
In order to do this, if a window is not yet mapped, postpone the
relationship setup until when the window is eventually mapped.
https://bugzilla.gnome.org/show_bug.cgi?id=791062
The documentation about gtk_file_chooser_set_local_only() states
that "non-native files may still be available using the native
filesystem via a userspace filesystem (FUSE)."
The code that made this possible in GTK+2 was missing from GTK+3 and
that represented a regression for Linux users in numerous applications
(Firefox, Thunderbird, Chromium, ...)
https://bugzilla.gnome.org/show_bug.cgi?id=787128
Traditionally (and on most backends) there's a single master pointer driven
by all pointing devices. The notable exception is Wayland though, where
master pointing devices are created per capability in the case of
pointer/touch, and one for each drawing tablet.
This function call makes it easy to access all these.
https://bugzilla.gnome.org/show_bug.cgi?id=790920
Same reason as GtkViewport does it: We might allocate child widgets
outside of the paned's content allocation. For drawing, we add a clip
node.
This was causing the "Record" button in the inspector recorder to ignore
pointer events since the treeview column header label in the GtkPaned
was swallowing it.
As the summary says, this allows using g_autoptr(GtkTreePath). This is
useful for API that uses out parameters for GtkTreePath that need to be
freed.
https://bugzilla.gnome.org/show_bug.cgi?id=791234
We are using `path` unconditionally, but it can be conditionally filled.
To avoid inconsistent internal state, and a compiler warning, let's
assert that the variable is always set.
This API allows specifying a GType and va_args of a value of that type
to set the clipboard contents. This massively simplifies setting weird
object types into the clipboard.
2 example patches included in this patch are the GtkTextBuffer and the
file list in the file chooser.
Using gobject-introspection, this should work without specifying the
type, so that you can literlally say
clipboard.set ("Hello World")
or
clipboard.set (pixbuf)
which is why I've also marked all other setters as (skip). They just
exist in C as wrappers for type safety reasons.
This is in preparation of using input streams to show that these
coordinates aren't needed most of the time and can otherwise be saved
during GtkWidget::drag-drop.
I decided to put this in a custom subclass, because then I could keep
the whole gtk primary protocol self-contained.
The other option would have been reusing GdkWaylandClipboard, but that
didn't seem worth it, especially because that code needs to interact
with the DND machinery, while the primary doesn't.
When the reply to a TARGETS request comes in, the clipboard may already
be reclaimed by the local app. Deal with that case (in an ugly way,
strictly speaking we should use a cancellable here).
This happens for example at startup when the initial TARGETS requests
have not been answered until after the main widow popped up. And if such
a window immediately claims the primary clipboard (like when the initial
focus is inside an entry), this race will happen.
This object tracks the SelectionNotifyEvent that has to be sent in
response to a SelectionRequest.
Currently it just looks like code reshuffling, but it's a prerequisite
for handling MULTIPLE, which requires to only send the notify after
every stream has writtten at least once.
But anyway, code is cleaner now, so it's a win!
This is a GSList of GFile and we want it so we can operate with lists of
files and text/uri-list.
I chose GSList over GList because that's what the GtkFileChooser API
uses, too.
Instead of using GtkClipboard and handling everything ourselves, we now
put GtkTextBuffer into the GdkClipboard and register (de)serializers for
text/plain.
Also make clipboard_claim() a vfunc so backends can override it.
Because the whole operation a vfunc, backends have the option of adding
code before the actual claim is done and potentially even fail or do
something after the successful claim.
Instead of having just one function that has the gtype and mime type as
out arguments, have 3 functions: 1 that finds any match, 1 that finds a
GType match and one for a mime type match.
This makes the API way more convenient to use.
This requires implementing a "pipe" so we can have 2 streams running:
contentprovider => serializer => outputstream
inputstream => deserializer => reader
And the pipe shoves the data from the outputstream into the inputstream.
GdkContentProvider is the object that represents local data in the
clipboard.
This patch only introduces the object and adds the clipboard properties,
it does not yet provide a way for the actual implementations to access
it.
The only access that is implemented is the local shortcut GValue access.
(1) Try all passed in formats in order if one of them fails.
(2) Don't blindly accept all formats, make sure they are mime types
(3) Add a bunch of special non-mime types that plug converters to
get to mime types
This allows us not just to pass any mime type to the read function, but
it also makes it possible to pass multiple mime types and the clipboard
can then try them in order until it finds a supported one.
This is so far not implemented though.
Turns out, way too many async operations are implemented by running the
sync operation in a thread. The easiest solution is to support that is
to use a GAsyncQueue for the buffers and deadlock if called from the
main thread.
(1) Turn X11 clipboard event handling into a regular filter function
(2) Maintain a timestamp in the clipboard, so we can pass it when
querying selections.
No idea why it's here, the hash table can store any kind of data,
there's no reason why it wouldn't be able to store an old X string type.
Might be a holdout from the old days, when strings were handled in
a special way (stored directly in the clipboard?).
https://bugzilla.gnome.org/show_bug.cgi?id=786509
This prevents GTK from throwing a bunch of warnings when it tries
to get drag source window -> screen of that window -> ipc widget for that screen,
and then tries to attach a signal handler to that widget.
Specifically, this happens when we get a DnD move from another
application.
https://bugzilla.gnome.org/show_bug.cgi?id=786509
1) Ensure that any DELETE requests from the target are sent to GDK, even if
both the source and the target are in the same process and it
is therefore possible to use a shortcut and call the handler directly
in GTK layer
2) Ensure that target GDK doesn't do anything when GTK asks it to send
a DELETE request, just report back immediately (the code up the stack
does not check for successfullness when request is DELETE, so not giving
it any data is OK).
The source code already synthesizes a DELETE request, so that side is
also taken care of.
https://bugzilla.gnome.org/show_bug.cgi?id=786509
We need to know the target atom value to know when we need to
do something with side-effects (since side-effects are expressed via
special target values). Previously, the code side-stepped that by looking
at the data type (which was rather unique for the one side-effect
target that we supported, signalled by the TARGETS target),
but for the DELETE target that seems to be no longer an option, hence the new
field to carry this information past the convert_selection() routine.
This prevents GDK from throwing a warning when trying to convert
a DELETE target, which has no format or data objects set.
The side-effects for the DELETE target happen earlier, in GTK layer.
By the point it gets to change_property(), it's a no-op.
https://bugzilla.gnome.org/show_bug.cgi?id=786509
To do that, run the message loop for one second or until the side-effect
of running the selection request handler is achieved (as opposed to
running it until the event is no longer queued).
The disavantage of this method is that if the event handling is
somehow missed (due to a variety of reasons - after all, it's not
a straight path from an event being queued to property_change()
being called), this will loop for one second. Since we do process
events during that time, this will not hang the application, but
might still restrict some of the functionality.
https://bugzilla.gnome.org/show_bug.cgi?id=786509
Handle WM_CANCELMODE and do nothing in response to it when DnD is
active. Otherwise pass it to DefWindowProc, which will call ReleaseCapture()
on our behalf.
This prevents us from losing mouse capture when alt-tabbing during DnD
(this includes the feature of Windows Explorer where dragging stuff over
a window button in the taskbar causes that window to receive focus, i.e.
keyboardless alt-tabbing).
https://bugzilla.gnome.org/show_bug.cgi?id=786509
Without this patch layered windows are only updated when they are moved
by the user or then their contents changes. This patch adds opacity
changes to the list of things that make GDK update a window. Without this
windows that don't redraw and are not moved by the used (DnD drag indicator
windows, for example) don't change their opacity.
(This commit is cherry-picked from the gtk-3-22 branch)
https://bugzilla.gnome.org/show_bug.cgi?id=786509
Massive changes to OLE2 DnD protocol, which was completely broken before:
* Keep GdkDragContext and OLE2 objects separate (don't ref/unref them
together, don't necessarily create them together).
* Keep IDataObject formats in the object itself, not in a global variable.
* Fix getdata() to look up the request target in its format list, not in the
global hash table
* Create target GdkDragContext on each drag_enter, destroy it on drag_leave,
whereas IDropTarget is created when a window becomes a drag destination
and is re-used indefinitely.
* Query the source IDataObject for its supported types, cache them in the
target (!) context. This is how GTK+ works, honestly.
* Remember current_src_object when we initiate a drag, to be able
to detect later on that the data object is ours and use a
shortcut when querying targets
* Make sure GDK_DRAG_MOTION is only sent when something changes
* Support GTK drag cursors
* Ensure that exotic GTK clipboard formats are registered
(but try to avoid registering formats that can't be used between applications).
* Don't enumerate internal formats
* Ensure that DnD indicator window can't accept drags or receive any kind of input
(use WS_EX_TRANSPARENT).
* Remove unneeded indentation in _gdk_win32_dnd_do_dragdrop()
* Fix indentation in gdk_win32_drag_context_drop_finish()
* Remove obsolete comments in _gdk_win32_window_register_dnd()
* Check for DnD in progress when processing WM_KILLFOCUS, don't emit a grab
break event in such cases (this allows alt-tabbing while DnD is in progress,
though there may be lingering issues with focus after dropping...)
* Support Shell ID List -> text/uri-list conversion, now it's possible
to drop files (dragged from Explorer) on GTK+ applications
* Explicitly use RegisterClipboardFormatA() when we know that the string
is not in unicode. Otherwise explicitly use RegisterClipboardFormatW()
with a UTF8->UTF16 converted string
* Fix _gdk_win32_display_get_selection_owner() to correctly bail
when selection owner HWND is NULL (looking up GdkWindow for NULL
HWND always succeeds and returns the root window - not the intended
effect)
* More logging
* Send DROP_FINISHED event after DnD loop ends
* Send STATUS event on feedback
* Move GetKeyboardState() and related code into _gdk_win32_window_drag_begin(),
so that it's closer to the point where last_pt and start_pt are set
* Use & 0x80 to check for the key being pressed. Windows will set low-order bit
to 1 for all mouse buttons to indicate that they are toggled, so simply
checking for the value not being 0 is not enough anymore.
This is probably a new thing in modern W32 that didn't exist before
(OLE2 DnD code is old).
* Fixed (hopefully) and simplified HiDPI parts of the code.
Also adds managed DnD implementation for W32 GDK backend (for both
OLE2 and LOCAL protocols). Mostly a copy of the X11 backend code, but
there are some minor differences:
* doesn't use drag_window field in GdkDragContext,
uses the one in GdkWin32DragContext exclusively
* subtracts hotspot offset from the window coordinates when showing
the dragback animation
* tries to consistently support scaling and caches the scale
in the context
* Some keynav code is removed (places where grabbing/ungrabbing should
happen is marked with TODOs), and the rest is probably inert.
Also significantly changes the way selection (and clipboard) is handled
(as MSDN rightly notes, the handling for DnD and Clipboard
formats is virtually the same, so it makes sense to handle
both with the same code):
* Don't spam GDK_OWNER_CHANGE, send them only when owner
actually changes
* Open clipboard when our process becomes the clipboard owner
(we are doing it anyway, to empty the clipboard and *become* the owner),
and then don't close it until a scheduled selection request event
(with TARGETS target) is received. Process that event by announcing
all of our supported formats (by that time add_targets() should have
been called up the stack, thus the formats are known; just in case,
add_targets() will also schedule a selection request, if one isn't
scheduled already, so that late-coming formats can still be announced).
* Allow clipboard opening for selection_convert() to be delayed if it
fails initially.
* The last two points above should fix all the bugs about GTK+ rising
too much ruckus over OpenClipboard() failures, as owner change
*is allowed* to fail (though not all callers currently handle
that case), and selection_convert() is asynchronous to begin with.
Still, this is somewhat risky, as there's a possibility that the
code will work in unexpected ways and the clipboard will remain open.
There's now logging to track the clipboard being opened and closed,
and a number of failsafes that try to ensure that it isn't kept open
for no reason.
* Added copious notes on the way clipboard works on X11, Windows and GDK-W32,
also removed old comments in DnD implementation, replaced some of them
with the new ones
* A lot of crufty module-global variables are stuffed into a singleton
object, GdkWin32Selection. It's technically possible to make it a
sub-object of the Display object (the way Wayland backend does),
but since Display object on W32 is a singleton anyway... why bother?
* Fixed the send_change_events() a bit (was slightly broken in one of the
previous iterations)
* Ensure that there's no confusion between selection conversion (an artifact
term from X11) and selection transmutation (changing the data to be W32-compatible)
* Put all the transmutation code and format-target-matching code into gdkselection-win32.c,
now this code isn't spread across multiple files.
* Consequently, moved some code away from gdkproperty-win32.c and gdkdnd-win32.c
* Extensive format transmutation checks for OLE2 DnD and clipboard.
We now keep track of which format mappings are for transmutations,
and which aren't (for example, when formats are passed as-is, or when
a registered name is just an alias)
* Put transmutation code into separate functions
* Ensure that drop target keeps a format->target map for supported formats,
this is useful when selection_convert() is called, as it only receives a
single target and no hints on the format from which the data should
be transmuted into this target.
* Add clear_targets() on W32, to de called by GTK
* Use g_set_object() instead of g_ref_object() where it is allowed.
* Fix indentation (and convert tabs to spaces), remove unused variables
(This commit is cherry-picked from the gtk-3-22 branch)
https://bugzilla.gnome.org/show_bug.cgi?id=786509
application/x-rootwindow-drop is not useful anywhere else,
so put it under #ifdef GDK_WINDOWING_X11
On W32 this prevents toplevels from automatically becoming valid
drop targets with a useless drop type.
(This commit is cherry-picked from the gtk-3-22 branch)
https://bugzilla.gnome.org/show_bug.cgi?id=786509
Instead of using a boolean to indicate a modal operation being in progress,
use a set of flags, and allow these to be set and unset independently.
Specifically, this allows WM_CAPTURECHANGED handler to only act when a drag-move or
drag-resize modal operation is in progress, and ignore DND (which can also cause
WM_CAPTURECHANGED to be posted). This avoids a crash due to assertion failure when
OLE2 DND code tries to end a modal operation that was already ended by the WM_CAPTURECHANGED
handler.
(This commit is cherry-picked from the gtk-3-22 branch)
https://bugzilla.gnome.org/show_bug.cgi?id=786121
Commit 1d0fad3 revealed that there were some assumptions made that were
actually to compensate for the bug fixed by that commit, so we need to
remove those assumptions as they would result in AerSnap to not work
properly on HiDPI screens.
Also re-do how we set the x and y positions of our GdkWindow, so that we
are more consistent across the board when we go between a GDK window
coordinate and a Windows API window cooredinate.
This would also simplify the code a bit.
https://bugzilla.gnome.org/show_bug.cgi?id=785999
Some drivers don't do that (not sure whether that is the correct behaviour
or not). Remember each WT_PROXIMITY with LOWORD(lParam) != 0 that we get,
then look for a WT_CSRCHANGE. If WT_CSRCHANGE doesn't come, but a WT_PACKET
does, assume that this device is the one that sent WT_PROXIMITY.
Also include fallback code to ensure that WT_PACKETs for an enabled device
disable the system pointer, because WT_PROXIMITY handler might have
enabled it by mistake, since it's not possible to know which device left
the proximity (it might have been a disabled device).
https://bugzilla.gnome.org/show_bug.cgi?id=778328
Previously HiDPI scale was retrieved and applied too late in the initialization
process to affect monitor size and monitor workarea size, but the code that
initializes these sizes *did* try to use the scale, even though it was always
getting scale=1.
To fix this, move the too-late code into monitor enumeration routine.
This also fixes a probable semantic bug where width and height were divided
by scale, again.
Now monitor and workarea should be in application pixels (i.e. divided by scale),
as intended.
https://bugzilla.gnome.org/show_bug.cgi?id=778835
Previously GDK only made up monitors when it initially found none. Now it
also makes up monitors when it initially finds some, but later fails to get
their informatin in a normal way and finally prunes them out, being left with
zero monitors.
Having zero-length monitor array is unexpected and causes a number
of critical warnings and some critical functionality (such as displaying
drop-down menus) fails in such cases.
Ideally, there might be such a way to interrogate W32 API that produces the
information about non-real (but active) monitors out of it so that it isn't
necessary for us to make stuff up. However, this code is already complicated,
and i am not prepared to dig W32 API to find a way to do this.
This fixes the issues people had when they accessed a Windows desktop via RDP.
https://bugzilla.gnome.org/show_bug.cgi?id=777527
Windows WM handles AeroSnap for normal windows on keydown. We did this
on keyup only because we do not get a keydown message, even if Windows WM
does nothing with a combination. However, in some specific cases it DOES
do something - and we have no way to detect that. Specifically, winkey+downarrow
causes maximized window to be restored by WM, and GDK fails to detect that. Then
GDK gets a keyup message, figures that winkey+downarrow was pressed and released,
and handles the combination - by minimizing the window.
To overcome this, install a low-level keyboard hook (high-level ones have
the same problem as normal message loop - they don't get messages when
Windows WM handles combinations) and use it to detect interesting key combinations
before Windows WM has a chance to block them from being processed.
Once an interesting combination is detected, post a message to the window, which
will be handled in due order.
It should be noted that this code handles key repetitions in a very crude manner.
The downside is that AeroSnap will not work if hook installation function call fails.
Also, this is a global hook, and if the hook procedure does something wrong, bad things
can happen.
https://bugzilla.gnome.org/show_bug.cgi?id=776031
Instead of using some kind of flawed logic about modifying a keypress result
when CapsLock is toggled, just add a CapsLock shift level (and all derived
shift levels, i.e. Shift+CapsLock and CapsLock+AltGr and Shift+CapsLock+AltGr)
and query Windows keyboard layout API about the result of keypresses involving
CapsLock.
Keysym table is going to be (roughly) twice as large now, but CapsLock'ed
keypresses will give correct results for some keyboard layouts (such as
Czech keyboard layout, which without this change produces lowercase letters
for CapsLock->[0,2,3,4...] instead of uppercase ones).
Keymap update time also increases accordingly.
https://bugzilla.gnome.org/show_bug.cgi?id=165385
Instead of checking for window state and giving it extra styles that
fit, just give it all styles that it is missing. It turned out that
otherwise it is impossible to, for example, restore a maximized window
via sysmenu. Also, be more flexible towards GDK/WM window state mismatches
and consider the window minimized/maximized if *either* GDK or WM thinks so.
https://bugzilla.gnome.org/show_bug.cgi?id=776485
Just set check_for_dpi_awareness = TRUE and eventually it will be handled
correctly, even if setDpiAwareFunc() returns E_ACCESSDENIED or shcore functions
are NULL.
https://bugzilla.gnome.org/show_bug.cgi?id=777031
When primary monitor is smaller than the actual monitor on which the
window is being maximized, the WM will do widnow size adjustments
that will completely screw the window size if we try to make it
smaller than 100% fullscreen (to account for taskbar size, for example).
Fix this by overriding maximized window size during WM_WINDOWPOSCHANGING.
https://bugzilla.gnome.org/show_bug.cgi?id=775808
Change the name of the property from stock-size to icon-size,
and make it an enum property instead of uint. This makes it
impossible to specify invalid numeric values in ui files, and
at the same time makes it possible to refer to the existing
values by their nick.
Fix up the callers.
If the rounded clip node is rectilinear, we can simplify it to a normal
clip node. If not, we really need to use a rounded clip node. In both
cases, we can do the same check we do when collecting normal clips and
avoid the clip node altogether if the child node does not get clipped
anyway.
This saves between 3 and 10 nodes in the widget factory, depending on
what page gets rendered.
Clip nodes have a clip rect and we only need to actually create a clip
node if any child node gets clipped at all. If the clip rect conains the
child node bounds entirely, we don't need a clip node.
Every added widget having a separate random widget type makes it useless
to use the widgetbowl demo for any sort of performance comparison.
Instead. use only one widget type for all the moving children but make
that changable.
I got a lot of "clip in clip" cases, for example a CellClip with a
CellTextClip inside. It is really trivial to merge these when we
pop and makes it easier for all backends, so lets do that.
This affects a few apis, such as gtk_text_iter_get_pixbuf,
gtk_text_buffer_insert_pixbuf and GtkTextBuffer::insert-pixbuf,
which have all been replaced by texture equivalents.
Update all callers.
It's unused. Plain text is not using that framework, neither is
in-process same-display transmission.
So it was only useful for sharing text with custom tags across
applications, and nobody is doing that.
That is some old code that still uses IOChannels, and the only
pseudouser is at-spi-atk's commented out code that is still using
CORBA types.
So get rid of it now before I need to start adapting it to the new
clipboard.
The wayland backend currently never emits GDK_SELECTION_CLEAR events.
GtkClipboard uses this signal in order to clear the clipboard owner when
the selection is set to something outside the application.
This commit ensures the wayland backend emits GDK_SELECTION_CLEAR before
setting the clipboard owner to NULL, as this means we lost the
selection.
Signed-off-by: Christophe Fergeau <cfergeau@redhat.com>
https://bugzilla.gnome.org/show_bug.cgi?id=790031
This function returns global session state that may
not be available to applications (e.g. in sandboxed
environments), and is not needed by applications,
so just drop it, instead of keeping a function around
that can't be guaranteed to work.
There are too many stack elements in the main stack. So add a substack
for the pages that display common global state. The appropriate name I
found for it was "Global".
It's used to house the General, Visual, Resources and Statistics pages
for now.
When I rewrote that function to not use GdkDeviceManager,
I overlooked that the window filtering needs to apply
to the master pointer as well, as other code assumes
that _gtk_widget_get_device_window will return non-NULL
on the devices in the list. Fix this.
This signal will be emitted whenever the gesture received a
button release or touch end event without a pairing button
press or touch begin. This usually happens when grabs transfer
input from one widget to another mid-press.
https://bugzilla.gnome.org/show_bug.cgi?id=789163
We can just as well use GdkSeat to enumerate (attached)
devices. Note that this change excludes floating devices
from consideration.
This keeps the copy-pasted code in sync with gtkwindow.c
This means we can directly upload these as textures, rather than
create a new surface and draw it into that. We still have to upload,
but there isn't a lot we can do about this as for these nodes
we generally redraw everything each time.
Similar to GtkEntry, add an "Insert Emoji" context
menu item, and add the same keybindings. We don't
add the icon here, since it is not clear where it
would go.
https://bugzilla.gnome.org/show_bug.cgi?id=790029
Ensure that things build again, and instead use the Windows API to
acquire the screen dimensions (note: this may need to be scaled for
HiDPI, but since I do not own a WinTab-based device, I will need to
keep the dimensions as-is for now).
Also update the gdkdnd-win32.c code to use formats rather than targets.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
This fixes the build of GTK+ master on Visual Studio 2013 (and possibly
others) as snprintf() may not be supported even if the required C99
features are supported by the compiler.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
The state argument was removed in commit 1518fe0 (API: stylecontext:
Remove state argument from getters), but we missed updating this file
until commit 5b94fe6 (stylecontext: Make first property name explicit),
as the compiler did not issue any warnings on the (now-defunct) usage.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
This is not used by anything yet, but add it now, so people looking at
this new code can make sense of it.
Plus, the documentation mentions it, so better have the docs make sense.
It will be used once we add support for conversions to GDK and allow
doing cipboard/dnd by GValue.
Make sure the API reflects the idea that GdkContentFormats is a set
containing mime types. In particular, treat the object itself as a
plural - it's named content format`S' after all - and therefor use
the correct verb form.
Also make GdkContentFormats keep an array instead of a list, now that
it's immutable.
Whereever we handle long-press for touch, it makes sense to handle
right-click as a faster alternative for mouse-based interaction.
This commit makes right-click work to bring up the variation
selector for Emojis.
The generated file clienthtml.h is #included by broadway-server.c, which
is one of the sources of the broadway library — so clienthtml.h needs to
be one of the sources of that library too.
Signed-off-by: Philip Withnall <withnall@endlessm.com>
https://bugzilla.gnome.org/show_bug.cgi?id=790489
For a start, this makes gtk_main() work different from g_main_loop_run()
calls.
But most importantly, modern GDK does proper syncing itself and doesn't
need to rely on a catch-all to get it right.
Instead of allowing people to pass a uint user-data, insist on them
comparing mime types.
The user data was a uint instead of a pointer anyway, so uniqueness
could not be guaranteed and it caused more issues than it was worth.
And that's ignoring the fact that it basically wasn't used.
The HighContrast theme was not parsing anymore, due to
leftover widget style properties, and some missed cleanups,
like -gtk-icon-effect. Also update for the new focus handling,
and make checks and radios sharp again.
We no longer support registering custom icon sizes, so
we can make these properties just enums. This also lets
us specify them by nick in ui files. Nice!
Instead of looking at the icon size, look at the CSS value for
-gtk-icon-size. Set style classes depending on icon size instead.
Trivially change Adwaita and HighContrast to report the same values as
before.
Instead, add a function gtk_image_set_icon_size() for the cases where
overriding the icon size is necessary.
Treat icon sizes the same way as pixel sizes, too. So gtk_image_clear()
no longer unsets the icon size.
Instead of returning the icon size with them, make
gtk_image_get_icon_name() and gtk_image_get_gicon() only return the icon
itself.
As a benefit, we can turn them into regular getters that return values
instead of requiring out parameters.
Instead, provide gtk_image_get_icon_size() to query the icon size.
Instead, turn the functions into backend API:
gdk_broadway_display_add_selection_targets()
gdk_broadway_display_clear_selection_targets()
Remove the old per-backend functions, too.
This is a trivial commit that does a big change: We now ignore event masks.
Further commits will clean up code, but if bisection ends up here, you
know it's because code is getting delivered events that it weren't getting
before.
It wasn't taking into account whether the sidebar had support for them
or not, resulting in a file chooser with open in new tab/window menu
items when it's not supported.
To fix it, do as with the other menus and check for the availability of
new tab/window flags.
https://bugzilla.gnome.org/show_bug.cgi?id=786123
g_resources_enumerate_children expects the path to end
in a '/' (even though thats not stated in the docs), and
will copy it if that isn't the case. Avoid the copy
by putting a '/' there to begin with.
g_resources_enumerate_children expects the path to end
in a '/' (even though thats not stated in the docs), and
will copy it if that isn't the case. Avoid the copy
by putting a '/' there to begin with.
When the duration is set to 0, clamp it to 1us. This way we're almost
correct: We should really instantly finish, but we don't. But we do
respect the delay.
Doing this properly would require some refactoring of how the progress
tracker actually maintains progress, and this is just a quick fix.
Also, sanitize the RTL correction code that made sure resizing the width
of a treeview would keep the contents glued to the right border instead
of the left border.
GtkSourceView is not using it, so it's unneeded. And it's certainly
diving deep into event internals of GtkTextView which hinders a proper
gesturization.
The selection bubble is not part of the text windows, so hiding it
during scroll should not be done in the text window code.
Also remove an unused variable that was only set in that code but never
read.
Insist that a first non-NULL property is passed to
gtk_style_context_get().
This is in particular relevant because of dropping the state argument
since GTK3, and code like
gtk_style_context_get (context, state, "font", &font);
would keep compiling without warnings without this change.
Like the X11 and Wayland backends, re-work how the cursors are being
handled. So, we use a hash table to cache up the HCURSORS that we
create along the way.
We still need to cache up the icon/cursor themes since this is something
that is not part of Windows but was added on to support icon/cursor themes
such as Adwaita on Windows, but should be in-line with what is going on in
GdkCursor.
Also, remove the _gdk_grab_cursor global variable in gdkprivate-win32.h,
and replace it with another variable in the GdkWin32Display structure,
to make things cleaner in the process.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
Use the same approach we take for recoloring in GtkIconHelper now.
As part of this change, GtkCsSImageRecolor is changed to not derive
from GtkCssImageUrl anymore.
commit 475d916eb9 added various paths that
use theme-name for this, but the existing path already used THEME, with
a subsequent description referring to the latter. So use that everywhere
With the shader approach to symbolic recoloring, we must
not recolor the svgs anymore as we're loading them. Instead,
load them the same way that gtk-encode-symbolic-svg does.
This fixes the rendering of large symbolic icons e.g. the
'no search results found' page in the file chooser.
We must reset the image delay when stopping the timeout,
otherwise the code setting it up thinks it is still running.
This fixes cursor animation only working for the very first
enter of a widget with an animated cursor, as seen in the
cursors example in gtk4-demo.
This is necessary because picking is no longer automatically constrained
to a widget's box. So all clipping widgets need to constrain their
clipping, too.
This patch does that for GtkViewport only.
We cannot fast-track picking by using gtk_widget_contains(). Child
widgets may extend their parent using ie negative margins.
This is not just a theoretical concern, this is what's happening right
now with GtkScale's sliders relative to the trough.
The problem is that we now iterate through all widgets, even when they
aren't anywhere near the mouse pointer. So essentially every pick
operation is now guaranteed O(N_WIDGETS) which used to be the worst case
that pretty much never happened.
The trough widgets have the slider on top of the fill level and the
hilight widget. Make sure the widget stacking respects that.
This is particularly relevant because picking event targets should pick
the slider and not the hilight widget.
In particular, allow specifying a filename for a GDK_TYPE_TEXTURE
property. This makes it easy to transition properties from Pixbuf type
to Texture type without having to touch resource files.
We cannot unrealize a renderer in the dispose function, because that
would cause this chain to happen:
gsk_gl_renderer_dispose
gsk_renderer_dispose
gsk_renderer_unrealize
gsk_gl_renderer_unrealize
So we would call into thje GL renderers unrealize when it has already
(partially) disposed itself and ause accesses to dead variables.
and gdk_texture_new_from_resource().
While doing set, turn all GDK_AVAILABLE_IN_3_90 into
GDK_AVAILABLE_IN_3_94 because that's now true after the renaming.
A sideeffect is that we don't set the correct parent window on child
widgets anymore, but that is hopefully going to be fixed once we get rid
of child windows completely.
Since on Windows we need to use a good amount of temporary GL contexts,
we need to switch back to the original GL contexts we were using when
we are done with the temporary GL contexts, otherwise multi-GL windows
will cause confusions causing display artifacts and crashes.
Also, use the GdkWin32GLContext::gl_hdc consistently throughout
the code and remove the GdkWin32Display::gl_hdc as Lukas K pointed out
that GdkWin32Display::gl_hdc becomes out-of-date and so the HDC that the
GL context is bound to becomes incorrect in sceanarios using multiple
windows with GtkGLArea/GdkGLArea items (which would cause the artifacts in
programs that use multiple windows with GtkGLArea/GdkGLArea items, and it
turns out that GdkWin32Display::gl_hdc is actually not necessary to help
keep track of the HDCs we use for our GL contexts.
This will also fix on Windows with GDK_GL=always, or when GSK's gl
renderer is used.
Partly based on patch from Lukas K <lu@0x83.eu>
https://bugzilla.gnome.org/show_bug.cgi?id=789213
Move the leftovers from the removals to use the current APIs, to fix the
build. Also for gdk_device_virtual_set_window_cursor(), only do
something when a valid GdkCursor is passed in here.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
Widgets can now set their favorite cursor using public API.
This is very necessary because all cursor-setting APIs are still
setting it on their GdkWindow, which by now is the toplevel - oops.
Change constructors to reflect that.
While doing so, also add a fallback argument to the cursor constructors,
so it is now possible to create cursors with fallback.
Instead of creating a GdkX11Cursor, create GdkCursors. Cache the XCursor
in a hash table instead.
Also, make use of the new fallback mechanism for fallback code: Make
sure to provide cursors for the names that are guaranteed to exist, but
do not do bad attempts at displaying texture surfaces.
Black/White/transparent is not a replacement for those.
The check used to achieve discarding events not meant for the window
widget itself (because they are handled in the regular paths). Using
the target widget is the equivalent now.
Use g_value_set/get_boxed() in gtk_window_get/set_property(), case PROP_ICON.
icon_from_list() shall always add a reference to the returned icon.
gtk_window_set_icon() must accept icon != NULL.
https://bugzilla.gnome.org/show_bug.cgi?id=789870
Stop wrapping the xsettings manager window in a foreign
window. This means that we cannot use the gdk window filter
APIs anymore, so just do the filtering in a non-generic
way.
The preferred api to create cursors is by name, and the
GdkCursorType enumeration can directly trace its ancestry
to the horrible X cursor font. So lets stop using it.
gdk_display_get_default_screen is gone, but we still
have x11-specific screen apis that GTK+ is using, so
we need an alterative way to get the screen object.
• Remove the box-shadow at the top when the entry is in the foreground
• Bump precedence so that :disabled entries do not have .flat overridden
• Also add :backdrop to stop HCInverse getting a lighter BG in :backdrop
https://bugzilla.gnome.org/show_bug.cgi?id=789733
Move the default pos of the Paned handle to 400px from the left, i.e.
50% of the default width of the window. The previous position at 300px
from left meant the node treeview was too narrow & could easily result
in the (useful) State column not being visible in the case of many
apps. The properties pane doesn't need to be as big as it was anyway.
https://bugzilla.gnome.org/show_bug.cgi?id=788898
We disconnect from the GDK window, so the renderer can't keep any useful
state.
Plus, we might be using an entirely different window next time we
realize (after a call to gtk_window_set_display() for example) that should
use a completely different renderer anyway.
GTK+ now uses the gtk-xft-dpi setting directly.
Note: this commit only fixes the backends that
currently provide this setting. The win32 and
Quartz backends still need to be fixed.
This patch moves the "Copy to Clipboard" button into the same container
as the description label, to centre the button regardless of the number
of icons shown in the grid.
https://bugzilla.gnome.org/show_bug.cgi?id=789134
Drop the screen argument from gdk_dnd_find_window_for_screen
and rename the function to gdk_dnd_find_window. The screen
argument does not add anything here since the drag context
is already tied to the display. Update all backends, and
update all callers.
Consider the coordinates passed to gtk_widget_queue_draw_region to be
relative to @widget's origin, not its parent. That implies passing
priv->allocation or priv->clip to _queue_draw_region of a widget means
using its parent as @widget.
This fixes GtkScrolledWindow overshoot invalidation, which assumed the
coordinates to be widget-relative and not parent-relative.
And have a priv->display instead of a priv->screen.
Includes turning gtk_menu_set_screen() into gtk_menu_set_display(),
because that function just forwards to its window.
Implement GdkDisplay->get_setting() using the existing
_gdk_win32_screen_get_setting() and get rid of GdkScreen->get_setting()
as a result, to follow the changes in GDK.
Also, since we don't emit settings events in the Windows GDK backend,
but we acquire settings to print using GDK_SETTING, drop all references
related to GDK_SETTING since that is now removed. Update the debug
strings that are print out as a result
(gdk_screen_get_setting->gdk_display_get_setting).
https://bugzilla.gnome.org/show_bug.cgi?id=773299
This is just lots of renaming.
The interface remains private, so the public API does not change, apart
from removing the definition of the Interface object to avoid
subclassing.
We are not emitting these events anymore, so lets remove them
from the api. The GdkSettingAction enum is moved to xsettings-client.c
where its only use remains.
This commit adds gdk_display_get_setting and a ::setting-changed
signal, which will replace the settings event we use now. Note
that I've done away with the GdkSettingAction argument that the
event has, since we are not using it at all.
On Windows, when IME is used, each keystroke results in the
WM_IME_COMPOSITION event being sent first. This means that in our case
when one decides on to accept the input that is in the preedit buffer,
we first get from Windows the WM_IME_COMPOSITION event
(where we emit the commit signal), followed by the WM_IME_ENDCOMPOSITION
event (where we emit the pair of preedit-changed and preedit-end
signals).
Since commit f11f989 (GtkEntry: Remove recompute idle), we do the input
recomputation directly, this will cause a pair of "Pango-WARNING:
Assertion failed: (index >= 0 && index <= layout->length)" being shown,
as gtkentry.c's priv->preedit_length and priv->preedit_cursor was unable
to be reset to 0 in time as a result of the recomputation triggered by
the commit being done before the reset of priv->preedit_length and
priv->preedit_cursor (which are no longer valid as we essentially say
that we are done with the preedit buffer).
As we could only acquire the final string that was entered in this
preedit session when we handle the WM_IME_COMPOSITION event, fix this by
saving up the final string we acquire from Windows IME in UTF-8 when we
handle the WM_IME_COMPOSITION event from Windows, and emit the commit
signal with that string after we emit the preedit-changed and
preedit-end signals when we handle the WM_IME_ENDCOMPOSITION event from
Windows, which comes afterwards.
Also fix the formatting of the code around the parts of the files that
was changed.
https://bugzilla.gnome.org/show_bug.cgi?id=787142
When a widget unparents its child widget manually in finalize, this can
lead to the parent-set signal being emitted for those child widgets. The
parent already has a ref_count of 0 though, so it can't be used in a
meaningful way. Specifically, emitting the signal will already try to
ref the parent which prints a critical.
Since GtkWidget already has a "parent" property, one can use its notify
signal instead to get notified when the parent widget changes.
... instead of returning either itself with uncomputed images or a
copy of itself with only one computed image and lots of other
uncomputed images that we're never gonna look at again.
This fixes expressions like -gtk-scaled(-gtk-recolor(...),-gtk-recolor(...))
which Adwaita uses for checkmarks and bullets.
The Vulkan renderer creates a fallback surface for each shadow
node, even if we end up not rendering anything to it. Avoiding
this is a nice optimization.
This fixes blurry text and icons whenever we apply shadows
in a hidpi window. Shadow nodes are the last ones that we
still use fallback for, and this was causing us to render
the text blurry.
Pass a scale factor when caching glyphs or looking them
up in the cache. The glyphs in the cache are rendered
with subpixel precision determined by the scale. Update
all callers to pass a scale factor according to the window
scale. This lets us render crisp glyphs on hidpi systems.
The code that checks for the proper size of the our swapchain
was not taking window scale fully into account. With this change,
setting the window scale to 2 in the inspector causes the window
to grow and rendering to be scaled up as expected, with Vulkan,
in the same way it already is with cairo.
Epoxy 1.4 has new ad hoc API that we can use to check whether GLX is
available on the current system.
If we didn't use this API, we'd have to manually dlopen libGL (or its
equivalent on different OSes) and check if it had GLX symbols; since
Epoxy already does all of this internally, we can simply ask it instead.
https://bugzilla.gnome.org/show_bug.cgi?id=775279
The GtkWidgetClass::measure vfunc is not required to assign a value to
any of the (out) parameters, so we need to initialize the locals we pass
to it, otherwise we can end up with a garbage size request.
The path bar would crash if we disposed it before all pending I/O
operations had finished. Now we remember all the outstanding
operations directly in the GtkPathBarPrivate, and deal with them
consistently.
The copy of the PangoGlyphString we do here was showing up
in some profiles. To avoid it, allocate the PangoGlyphInfo array
as part of the node itself. Update all callers to deal with
the slight api change required for this.
According to the documentation, gdk_monitor_get_geometry() reports the
monitor geometry in ”application pixels”, not in ”device pixels”,
meaning that the actual device resolution needs to be scaled down by the
scale factor of the output.
x11 backend does that downscaling, whereas Wayland backend did not,
causing a discrepancy depending on the backend used.
https://bugzilla.gnome.org/show_bug.cgi?id=783995
If the compositor prefers server-side decorations and the client doesn't
customize the title bar, we disable client-side decorations and let the
compositor know. Otherwise, we continue to use client-side decorations.
Signed-off-by: Drew DeVault <sir@cmpwn.com>
https://bugzilla.gnome.org/show_bug.cgi?id=781909
Under Wayland, an xdg_surface.configure with size 0x0 means it's up to
the client to set its size.
When transitioning from maximized state to un-maximized, the Wayland
compositor will send such an 0x0 configure so that the client can
restore its original size.
However, the original size was already constrained, so re-applying
size constrains can lead to a smaller size when using size increments.
Avoid this caveat by not applying size constrains when we are restoring
the original size.
https://bugzilla.gnome.org/show_bug.cgi?id=777072
As documented, GtkAppChooser is "typically [used] for the purpose of
opening a file". However given that applications that support neither
opening files nor URLs are filtered out, the chooser is not actual
useful for any other (atypical) usage. Change that by only applying
the filtering if a content-type was set, and use the full unfiltered
list otherwise.
https://bugzilla.gnome.org/show_bug.cgi?id=789327
We were unnecessarily spewing warnings when blank cursors
were getting a new scale set. Standardize on "none" as the
name for blank cursors, and avoid the warning.
https://bugzilla.gnome.org/show_bug.cgi?id=775217
Since focus can now be represented by more than one state,
just looking at the focus_child is no longer sufficient - we
may fail to propagate :focus(visible) if we do so. For now,
just remove the shortcut and always do the work.
Some clients (e.g. gnome-online-accounts) quickly unmap and map
a window. With some backends the backend surface will be replaced
causing the application to crash because the GL context is still
using the old surface. Clearing the GL context when a window is
withdrawn fixes this.
https://bugzilla.gnome.org/show_bug.cgi?id=789141
__builtin_popcount is a GCCism that is used to count the number of bits
involved, which means any non GCC/CLang compilers won't like the code,
meaning that on MSVC builds we must implement it ourselves.
We first use __cpuid() to check whether the CPU supports the popcount
instruction, if it does, we use the __popcnt intrinsic, otherwise
(untested, since I don't have a system that does not have the
instruction), we use the suggested hacks at
http://graphics.stanford.edu/~seander/bithacks.html#CountBitsSetParallelhttps://bugzilla.gnome.org/show_bug.cgi?id=773299
We still need to access the GdkEvent structure here directly, as using
the GdkEvent getters is likely not worth the trouble involved.
Please see Emmanuele's comment (#97) of the following bug URL.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
The list of surfaces passed into the function may be NULL, so don't try
to initialize the surfaces if it is so, to avoid a crash.
Also, remove the cast to GdkPixbuf* for getting surfaces->data, as we
are already using a cairo_surface_t*.
https://bugzilla.gnome.org/show_bug?id=773299
When building with G_DISABLE_ASSERT, the g_assert_not_reached()
statement won't do anything, and we're going to fall through, and the
compiler will emit a warning that we're not returning anything from a
function with a return value.
These are no longer used, instead we always covert to surface as
early as possible and drop the pixbuf.
This means we never store both the pixbuf and the surface at
for any longer time, which is wasteful. Also, its one step further
to drop GdkPixbufs from generic use in our APIs.
Rather than store the pixbufs as themselves we immediately convert
them to surfaces. In the uncommon case that a pixbuf is read back
from the renderer we generate a new one from the surface data.
This drops the pixbuf property and the pixbuf getters. We keep
gtk_image_new/set_from_pixbuf, but these are small helpers that
immediately convert to a surface, and there is no way to later get
back the pixbuf you passed in.
The from file/resource codepaths are also changed to load a surface
instead of a pixbuf.
Rename the surface getter to peek, following other render
node getters, and make the surface-based constructor private,
since it is not something we want to encourage.
Update all callers.
The gtk_tools variable is an array of arrays; adding a new element
requires to maintain the same type, or we'll get a build failure when
we try to extract the newly added element.
Add the necessary machinery into the Meson definition files so that we
can build for Windows.
Since we don't have Wayland or X support for our use case here, disable
them once we know that we are building for Windows, as they are
(otherwise) enabled by default, and enable the items that need to be
built for Windows builds.
Exclude gtk4-launch from Windows builds as that is something that
is not supported on Windows.
As we won't have gio-unix on Windows, and PangoFT2 is optional, don't use
fallbacks for them when we are on Windows (but do use fallbacks for
gio-win32, as it will be used).
Also, clean up meson.build a bit as we can just force-include
msvc_recommended_pragmas.h from GLib since we depend on GLib, and so we
can handle these warnings from msvc_recommended_pragmas.h instead.
https://bugzilla.gnome.org/show_bug.cgi?id=785210
The color-matrix shader was creating pixels with r,g,b > a in
some cases, which leads to unexpected test failures. In particular
this as visible the opacity render node test for opacity 0.
When making mockups for GNOME apps in Inkscape, looking for symbolic
icons is a common task. Searching for icons in the file system is clumsy,
and icon-browser provides a much better interface for finding them.
However, currently there is no way to insert the symbolic icons as SVG
directly from icon-browser, so right now it is only useful for finding
the name.
This patch adds a sixth column to the modal window that appears when
clicking a symbolic icon. The icon in this column is labeled "scalable",
and dragging it onto another window results in the vector icon URI being
inserted.
This enables a much simpler workflow when designing with symbolic icons.
https://bugzilla.gnome.org/show_bug.cgi?id=778930
What is missing is the "allocation" part of x/y coordinates. Since
gtk_entry_realize doesn't call gtk_widget_set_window(priv->text_area),
the coordinates returned by gdk_window_get_origin don't include it.
This patch fixes this.
https://bugzilla.gnome.org/show_bug.cgi?id=784509
Bug 737175 aimed to ensure that scrolling up on a horizontal range would
result in its value increasing, as that’s what users intuitively expect.
However, its commit 416c370da1 meant that,
if the event gives scroll deltas, we inverted our delta unconditionally.
So it broke horizontal scrolling: scrolling left moved the slider right…
We must only invert if using dy as delta. dx already has the right sign,
so inverting it was wrong.
https://bugzilla.gnome.org/show_bug.cgi?id=788905
The gtk_widget_get_display call in this if statement is showing up in
profiles. It ends up walking up the hierarchy to the toplevel to get its
GdkScreen, etc. so it is relatively costly. Avoid that call in most
cases by first checking if the RESIZE debugging is enabled for any of
the displays and only then checking if it is enabled for the widget's
display.
If the call to set_parent() failed, we were still adding the child to
the internal list of children, despite that it was not really added.
That meant we could later try to do invalid stuff with that non-child.
Fix that by asserting and giving up if the child that the user is
attempting to add is already parented.
https://bugzilla.gnome.org/show_bug.cgi?id=701296
The language is useful for parsing tools, such as that of gtkmm, which
otherwise assumes these are C snippets and elides them from its
generated documentation.
The event coordinates are (so far) irrelevant to what we are testing here,
just make all events happen in the middle of the window in order to ensure
all widgets receive it. More importantly, avoid using fixed pixel distances,
since we don't get guarantees about window sizes.
Fixes the gestures testsuite on X11.
It was used to mark css properties that affect widgets with text, but it
caused unnecessary invalidations. E.g. 'color' was marked as
AFFECTS_TEXT but changing just the color of a label should not
automatically queue a resize, which is what the code in
gtk_widget_real_style_updated does.
Replace this flag with GTK_CSS_AFFECTS_TEXT_SIZE and
GTK_CSS_AFFECTS_TEXT_CLIP, which GtkWidget can use only if the widget
actually has text.
Now all widgets are mandated to handle the real thing, which means no
pointer events are emulated for the pointer emulating touch. The output
of these tests relied on this fact, so update to the tests handling real
touch events.
Legacy GtkWidget vmethods are now handled on an event controller, which
due to being the very first controller added to every widget, runs in
a different order than the previously hardcoded.
Probably testing legacy events is not really futurible, specially after
we stop installing this legacy controller by default. I'll leave the
choice to remove these specific tests for later though.
After a gesture first claims, and later rejects a touch sequence,
a press event will be propagated further along. However propagation
got messed up since we only emitted as far as the toplevel.
It does not hurt us to keep middle clicks doing the same
as shift-primary clicks. This makes the transition from gtk2
less painful in terms of muscle memory.
https://bugzilla.gnome.org/show_bug.cgi?id=787669
Clarifies the code and helps catch invalid enum values before they
propagate further. Also add a comment about why two seemingly legitimate
values are not handled there (they’re handled higher up in the file).
Coverity CID: 1457700
Signed-off-by: Philip Withnall <withnall@endlessm.com>
https://bugzilla.gnome.org/show_bug.cgi?id=788787
gtk_box_pack_end will put the label child at the right side of the label
(in LTR orientation), but we want it left, directly next to the icon.
Also remove the spacing from the box child as this is a theme thing.
~Company ╡ so TL;DR: we put the static style in the cache, but then
⤷ ╡ compute a child style from the animated style in the cache
⤷ ╡ and we put the child style also in the cache (because
⤷ ╡ it's not animated)
⤷ ╡ then we run the animation, but reuse the cache every time
⤷ ╡ for both child and parent
⤷ ╡ so after the animation is done, we end up with a cache that
⤷ ╡ has the correct static style for the parent but an
⤷ ╡ incorrect static style for the child
⤷ ╡ because that static style was computed from the
⤷ ╡ initial animated style
This fixes https://bugzilla.gnome.org/show_bug.cgi?id=763517
Check UUID for printers obtained via DNSSD whether
they are already installed on local CUPS server.
Don't show such printers.
Not all printers published via DNSSD have UUID entry though.
https://bugzilla.gnome.org/show_bug.cgi?id=786794
An empty container has the same effect as transparency
with the cairo renderer, but creates black with Vulkan.
To avoid this, explicitly use a transparent color node.
This fixes the css blendmode example in gtk4-demo with
the Vulkan renderer.
As Timm Baedert pointed out, the previous fix made the
menubar go on top of popovers, which is just wrong. Instead,
make gtk_window_snapshot handle all direct children of the
window, taking care to stack popovers correctly.
We were node handling coordinates correctly when dealing
with differently sized child nodes in a blendmode node.
This was showing up in the gtk4-demo css blendmode example,
for blendmodes other than normal.
GtkMenu’s own keynav code, which actually bothers to account for the
layout of items, only happens if columns > 1. So, adding items to 1
column using a reverse loop meant they were placed in the Menu’s list of
children in that order, and because we only have 1 column, Menu passes
keynav up to MenuShell, which doesn’t adjust for the items’ positions.
‘Fix’ that here by adding items in the same order they’ll have when laid
out in the Menu, so keynav does what you’d expect, not the opposite. For
that, it’s simpler just to use gtk_container_add().
Let’s presume users are using add(), attach() with a non-inverted loop,
or attach() with arguments that create 2+ columns and so GtkMenu keynav.
This is important since _push_state returns a pointer into a GArray
which could be invalidated and point to garbage after the subsequent
push_state call.
This is used for example in the source tab of gtk4-demo.
It broke because GtkScrollbar no longer is a GtkRange,
but rather has one. So we need to forward the signal.
It was selecting paned separator, which means any separator at any level
of descent within a paned, including the toplevel container in GEdit.
We need to be more specific and only select the relevant separator that
is the direct child of the paned. This is what Adwaita does.
https://bugzilla.gnome.org/show_bug.cgi?id=788573
Nulling priv->button in _unset_tree_view() is asymmetrical: we create
it via init(), not _set_tree_view(), so we shouldn’t null in the latter.
Worse, doing so manifests in criticals + a SEGV easily with basic use of
testtreecolumns, removing the TVC from a TV then trying to add it to one
Finally, the wrong null-out meant dispose() failed to unref the button,
so it leaked.
https://bugzilla.gnome.org/show_bug.cgi?id=728452https://bugzilla.gnome.org/show_bug.cgi?id=788614
It was in both [general] with a description and [other] with none.
Leave it in [other] with the other folder- icons, + the description.
bonus: this makes all of [general] fit in our default window size!
This patch makes that work using 1 of 2 options:
1. Add all missing enums to the switch statement
or
2. Cast the switch argument to a uint to avoid having to do that (mostly
for GdkEventType).
I even found a bug while doing that: clearing a GtkImage with a surface
did not notify thae surface property.
The reason for enabling this flag even though it is tedious at times is
that it is very useful when adding values to an enum, because it makes
GTK immediately warn about all the switch statements where this enum is
relevant.
And I expect changes to enums to be frequent during the GTK4 development
cycle.
-Wshadow these days does not overwarn anymore like it did in gcc 4.
There are no warnings inside gtk, so better enable it to keep it that
way.
-Wuninitialized also has no positives, so I'm gonna turn it on just
because.
GtkCellArea uses event coordinates (thus in treeview relative
coordinates), but calculations used to happen in bin window coords.
We can just offset the cell area by the bin window, fixes cell
renderer activation and edition.
If the column is not clickable, it may make some sense to stop
event propagation here for button events. However motion events
should be left alone.
Fixes treeview column resize pointer cursors, since that's
implemented up the bubbling phase in the treeview.
The operations rely there on bin window relative coordinates, but we
are receiving GtkTreeView relative coordinates there. Fixes clicking
on treeview expanders, which was offset by visible headers.
-Wint-conversion is important because it checks casts from ints to
pointers.
-Wdiscarded-qualifiers is important to catch cases where we don't
strings when we should.
The border and icon highlight are useful feedback that was defeated by
CSS precedence. It worked for .titlebuttons due to their implementation,
but the same was not true for custom .flat buttons. This makes it so.
https://bugzilla.gnome.org/show_bug.cgi?id=788580
When the window was backdropped, they suddenly regained their border.
This was clearly not intentional or of any practical use to anyone.
Shuffle around some selectors so that the backdrop ones do not override
the flat ones and make the borders magically reappear when backdropped.
Note that, whereas standard titlebuttons get the border on :hover, other
.flat buttons in the headerbar do not. That should probably be fixed too
https://bugzilla.gnome.org/show_bug.cgi?id=788580
They were hard-coded to a transparent black, but that is our bg colour
in HC Inverse, so windows stacked on top of each other or a dark
background blended together into a mush.
Fix this by making the $_wm_border* colours relative to the fg colour,
so that HighContrastInverse gets borders that are transparentised white.
https://bugzilla.gnome.org/show_bug.cgi?id=788575
A missing decoration selector meant that we got a solid black background
behind the rounded corners of the dialog.
Copy the equivalent code from Adwaita, including nicely rounding the
focus outline too (& sorry, but this needs more newlines to be readable)
There were various problems, like only selecting on .tooltip and not the
widget node tooltip, not being specific enough for tooltip.csd, etc. So,
specific theming was absent, and default popup window styles got applied
This commit copies in the better working tooltip CSS from Adwaita, but
applies a couple of changes to make it work better in the HC themes:
• Reduce the transparency of the tooltip, so we achieve higher contrast
• Drop the black text-shadow, as it is not useful on this more black bg
Note: we may then need to re-add some of this to the .tooltip class. But
it is unclear what needs done there. While Adwaita is not doing it, we
are better not to confuse by keeping it in HC only; we should try to be
as close as possible, to make it easier for HC to keep up with Adwaita.
https://bugzilla.gnome.org/show_bug.cgi?id=769879
We need
.window-classes decoration
but within the decoration parent selector, we were doing
&.window-classes, which gave us
decoration.window classes
We need to fix this by selecting on .window-classes &
https://bugzilla.gnome.org/show_bug.cgi?id=788496
n_attach_points is the result of g_strv_length(): the index at which the
string vector ends in NULL. So by definition, when i == n_attach_points,
string[i] == NULL, and there is no need to check for the latter. The
fact that we did appears to confuse static analysers, as the dereference
and index check were inverted from what would normally be safe. We could
reverse them, but we may as well just remove the unnecessary NULL check.
https://bugzilla.gnome.org/show_bug.cgi?id=788458
This gives consistent behavior with e.g. Qt, Mozilla's suites and
LibreOffice (with non-truly native backends like "gen" and "gtk",
but unlike "gtk2" and "gtk3" ones that probably use true GTK menus).
This behavior is expected by at least some accessibility users, and
it seems good to behave like other common applications and toolkits
in this area. There should be no issue in doing so either for current
users, as it only enters the submenu instead of not doing anything.
https://bugzilla.gnome.org/show_bug.cgi?id=778811
There is no guarantee that the gtk_surface won't be NULL,
and Wayland API does not safeguard against NULL, so we have
to do that ourselves here.
We were also mistakenly cheking for the surface version off
by one, fix that too by checking if the surface version is
equal or greater.
Ditch two items that were white and so weren’t visible on our standard
theme anyway, and use the new space to test extra grid-mode properties.
Note that if we do this then, as before, we set the ListStore on the
ComboBox before appending to it, that produced runtime warnings like:
Gtk-CRITICAL **: gtk_menu_attach: assertion 'left_attach < right_attach' failed
I didn’t look into that yet, but it may indicate that attaching items
vs. recognising their spans don’t occur in the correct order. For the
purposes of testing this, I just create the CB after filling its model.
ComboBox and TreeMenu warned in the doc for :row-span-column that the
value must not exceed :wrap-width, but :wrap-width does not interact
with the number of rows; it’s the :column-span-column that’s relevant.
Also: Warn that spans must be > 0 for rows too, and that column spans <=
:wrap-width are also not useful for items at menu column positions > 0.
Finally, refer to items having spans, not values, as we were already
talking about values in the model (and rows in the menu).
This reverts commit 6ee2bf6286.
There is a way to get different kinds of borders: it's CSS. It's better
to keep the 4 Frames and demo the different styles we can do using CSS.
Instead of creating one GPtrArray per GtkSnapshotState and saving nodes
in there, create one GPtrArray per snapshot and assign a
start_node_index to every GtkSnapshotState as well as a n_nodes variable
so every state knows which nodes belong to it.
In some cases, we were creating gigantic intermediate textures
only to clip out a small section afterwards (e.g. in the listbox
example in gtk4-demo). This is wasteful if we apply effects on
the texture, such as blur or color-matrix. So, clip the dimensions
of the intermediate texture with the current clip. To make this
feasible, we move the texture coordinate computation out of the
pipeline setup functions into the node_as_texture function where
this clipping happens.
One extra complication we encounter is that the node might get
clipped away completely. Since Vulkan does not allow to create
empty images, we bail out in this case and not draw anything.
With these changes, the listbox example in gtk4-demo goes from
32M pixels of intermediate texture to 320000.
It's not a GtkCssGadget anymore, it doesn't have any properties or
signals either and it's not public. Further, its lifetime is very clear
the way it's being used inside GTK+.
This showed up in profiles in certain scenarios, so export a
_get_n_shadows getter instead and let callers provide a sufficiently
large allocated array of GskShadows, which we can use with
g_alloc/g_newa.
Instead of having a function with lots of arguments in
GskVulkanRender that we call from GskVulkanRenderPass which
then just calls back into GskVulkanRenderPass, just create
the new render pass object locally, and an api to add it
to the list that GskVulkanRender keeps. This makes it
a lot easier to preserve all the relevant parameters from
the parent render pass.
Move away from the idea of intra-frame sampling, since we only
push samples once per frame, anyway. Instead, make the profiler
keep a rolling average of the last n frames.
Whenever we need a node as a texture, we now start a new render
pass that renders the node into a new intermediate texture, and
set up a semaphore to make the current render pass wait for it.
As part of this reorganization, much of the setup and drawing
code moved from gskvulkanrender.c to gskvulkanrenderpass.c.
Allow to pass in semaphores to wait for before executing
and to signal after executing the command buffer. This
just exposes the capabilities of the underlying Vulkan
api. Update all callers to pass no semaphores, for now.
We will use this in the future.
The GtkFlowBoxCreateWidgetFunc type lacked GObject Introspection
annotations for its arguments. This made gtk_flow_box_bind_model()
unusable from Python as the callback function would be passed useless
values.
The annotations that I've added match those of the similar callback
type GtkListBoxCreateWidgetFunc.
https://bugzilla.gnome.org/show_bug.cgi?id=780758
I've finally figured out the right combination of src and dest
stage and access flags to make all validation warnings go away.
This commit only fixes the direct upload code.
This is another example for a 2-texture shader.
So far, only separable blend modes are implemented.
The implementation is not optimized, with an
if-else cascade in the shader.
We were looking at uninitialized memory here, instead
of the type of the source clip, as we should.
This showed up as mispositioned clip in the first frame
of a crossfade stack transition, and also as overdraw in
sliding stack transitions.
We already move the descriptor set layout out of it,
so we can just as well keep the pipeline layouts in
the render object as well, and get rid of this extra
object. Update all callers.
Instead of doing multiple copy commands with a tiny buffer
for each glyph, we can just batch them all together. This
also avoids the issue of creating multiple barriers for the
same image.
By tracking the last transition we can build the appropriate barriers.
Also use the most appropriate initial layout/access at creation :
for linear image : predefined (we prepare the content ourself through memcpy)
for everything else : undefined (we don't care about the content, will most likely be erase)
Move the glyph caching api to something that can support using
multiple textures. We now split the text render ops into multiple
ops for different textures, and make each op render just a substring
of the text node's glyph string.
gdk_seat_default_grab() grabs POINTER_EVENTS if the capability is
GDK_SEAT_CAPABILITY_ALL_POINTING. But that enumerator is a union that
includes GDK_SEAT_CAPABILITY_TOUCH, but we never grabbed TOUCH_EVENTS,
an unused macro that was presumably created with this purpose in mind.
So, check which of the ALL_POINTING capabilities we have, and set the
right mask of POINTER_EVENTS and/or TOUCH_EVENTS as required.
As part of this, explicitly let TABLET_STYLUS take over pointer events,
as this is the intended behaviour and was the effective result before.
This should fix touch events being lost in migrating from Device.grab()
to Seat.grab(GDK_SEAT_CAPABILITY_ALL_POINTING), as found by Inkscape.
https://bugzilla.gnome.org/show_bug.cgi?id=781757
The behavior where a touchpoint takes over the pointer position is
really backend dependent. Since this went away from the generic code,
implement it here.
This was by all lights broken, and is basically an implementation detail
of the X11 backend since the pointer emulating touch just steals the pointer
cursor, so should be reimplemented there.
One used to point to the toplevel and the other to the client-side window
that the pointer pointed to. The latter was made to be like the former in
most places, so put those together, and fix the remaining cases where the
variable might not end up with a toplevel/native window.
This is not necessary now that there's no client-side windows to track.
The only removed piece that could make sense is emission of grab broken
events, but it's already an stretch since the semantics of those with
multi-touchpoint is unclear.
Anyhow, This should be fixed at the GTK level, while we let GDK deal with
seat/device level grabs.
GDK just needs to care about toplevels nowadays, which means these events
are already delivered from the windowing. We don't need to generate
intra-window crossing events ourselves.
Those should be interpreted by widget-local gestures, not guessed at a
high level with no notions of the specific context. Users will want
GtkGestureMultiPress to replace these events.
Those worked similarly to those in GtkFlowBox, but would additionally
handle "active" state for child rows. Simplify this to just enabling/
disabling active state on gesture press/release, we don't get the
nice state updates when hovering around with a mouse button pressed,
but the rationale from flowbox applies here, and makes a nice cleanup.
They just maintain priv->in_button and widget state up-to-date, this
basically matters during user interaction, and is already maintained
in the gesture ::update handler. This seems to be sufficient.
Those basically controlled priv->active_child_active, which would
1) trigger a redraw when the pointer enters/leaves it, and 2) ensure
that press/release happen on the same child for it to be activated.
The former is not necessary, and the latter can be simplified by
just checking again the child on the coordinates given by the
::release gesture handler. This makes all enter/leave/motion_notify
event handlers unneeded.
All kinetic scrolling initial velocity calculations are now
taken from the scroll controller. The handling of timeouts
to snap back when overshooting has been also made to just
apply on devices that can't emit ::scroll-begin/end.
This is a GtkEventController implementation to handle mouse
scrolling. It handles both smooth and discrete events and
offers a way for callers to tell their preference too, so
smooth events shall be accumulated and coalesced on request.
On capable devices, it can also emit ::scroll-begin and
::scroll-end enclosing all ::scroll events for a scroll
operation.
It also has builtin kinetic scrolling capabilities, reporting
the initial velocity for both axes after ::scroll-end if
requested.
This change is made for consistency, it doesn't make sense to expose
one-way propagation, as it can only break expectations from GTK+. This
function might be made entirely private in the future, but it still
makes sense to do this in one go for our internal usecases.
This will allow further cleanups and optimizations in capture/target/bubble
event delivery. For simplicity, ATM every widget will receive its own
GtkEventControllerLegacy, it could be desirable to add finer control over
this in the future, so widgets that fully use event controllers for input
management can do away without this legacy piece.
It may result in a protocol error on older mutters, as GTK+ will
invariably request a higher version than what's available. Make
GTK+ also accept v1 if it's all the compositor has got.
As Benjamin says, ident should only be used if any value
is valid, which is not the case here. So use enums instead,
which should also be more efficient. To handle the more
complicated cases like font-variant-ligatures, we have to
introduce flags-like values.
Otherwise, we can't negotiate the latest version with the
compositor, making the compositor use v1 of the protocol and
pretty much ignoring all the edge constraints work.
Clarify that ::destroy, not ::hide*, removes a window from its app, by
replacing the mention of open windows with the blurb on destruction from
:application, completing commit 7db4bee4b6
Also link to the equivalent gtk_application_(add|remove)_window() calls,
since Application.add_window() already links back to Window:application.
* unless you use gtkmm…
https://bugzilla.gnome.org/show_bug.cgi?id=639931
It was never unref()d, either when replacing the existing GObject in
set_property(), cleaning up in finalize(), or becoming a placeholder.
Fix by using g_set_object() and g_clear_object() to unref as needed.
This also drops the check that the newly set object is a valid cloud
provider account, as we don’t do the equivalent for any of the other
object-typed properties, and Carlos didn’t think this was important.
https://bugzilla.gnome.org/show_bug.cgi?id=787600
Drop the current css2-style font-variant property and
replace it with a shorthand as specified in the css3 fonts
module. Currently, we fully support the font-variant-ligatures,
font-variant-position, font-variant-caps, font-variant-numeric
and font-variant-east-asian subproperties. font-variant-alternatives
is only partially supported.
Otherwise it fails to build with:
FAILED: gtk/im-ipa@sha/imipa.c.o
...
In file included from ../gtk/gtkintl.h:4:0,
from ../modules/input/imipa.c:28:
/opt/include/glib-2.0/glib/gi18n-lib.h:27:2: error: #error You must
define GETTEXT_PACKAGE before including gi18n-lib.h. Did you
forget to include config.h?
#error You must define GETTEXT_PACKAGE before
including gi18n-lib.h. Did you forget to include config.h?
^~~~~
../modules/input/imipa.c:144:3: error: ‘GETTEXT_PACKAGE’ undeclared
here (not in a function)
GETTEXT_PACKAGE, /* Translation domain */
^~~~~~~~~~~~~~~
../modules/input/imipa.c:145:4: error: ‘GTK_LOCALEDIR’ undeclared
here (not in a function)
GTK_LOCALEDIR, /* Dir for bindtextdomain (not strictly
needed for "gtk+") */
^~~~~~~~~~~~~
Instead of relying on special values of edge constraints, this
patch adds an internal-only gdk_window_supports_edge_constraints()
function that by default returns FALSE, and is implemented by
GdkWindowWayland and GdkWindowX11.
This way, we can properly detect server-side support for this
feature and adapt accordingly.
https://bugzilla.gnome.org/show_bug.cgi?id=783669
The last touch on this patch series is making GtkWindow able to
selectively adjust various UI details based on the different
tiled edges. The main driver here is that we don't want to show
shadows on edges that are constrained.
This patch adds the necessary code to do that, while still
maintaining compatibility with the old ways.
https://bugzilla.gnome.org/show_bug.cgi?id=783669
Following the previous patch, where edge constraints support
was added to the Wayland backend, this patch introduces the
necessary code to handle the _GTK_EDGE_CONSTRAINTS atom from
X11 backend.
https://bugzilla.gnome.org/show_bug.cgi?id=783669
Now that GTK windows have the ability to properly handle
per-edge tiling constraints, this patch extends GTK's
internal Wayland protocol to have a proper enum with the
relevant edge data.
Once this approach is validated, we can think of upstreaming
this work as an official Wayland protocol extension.
https://bugzilla.gnome.org/show_bug.cgi?id=783669
GTK windows don't have their tiling states really
hooked into the client-side decoration code, and
the only effect it has is disabling the resizing
edges.
With the introduction of per-edge tiling information,
we are backed by much more precise data on how the
window manager wants the app to behave.
This patch, then, fixes GtkWindow to take into account
per-edge tiling information. For compatibility purposes,
the previous tiled field was kept, and thing will just
continue working if no edge information is supplied.
https://bugzilla.gnome.org/show_bug.cgi?id=783669
These states will be consumed by GtkWindow in order to
have better edge management on tiling situations. Their
values are supplied by the compositor, and will be send
through and X11 Atom or a Wayland protocol extension.
https://bugzilla.gnome.org/show_bug.cgi?id=783669
The outline-{top,bottom}-{left,right}-radius names have been
deprecated for a while, so lets remove them. Everybody should
just use the -gtk-prefixed names for these properties.
The focus outline disappeared as the colour of the swatch got close to
the normal focus outline colour, which is alpha(currentColor, 0.3).
Fix by making the outline an alpha’d version of the tick colour, but
more opaque than normal outlines. 0.6 seems good enough; feel free to
improve it, but at least this ensures the outline can’t vanish anymore.
HighContrast achieves this already because it applies the color property
to the main node, not the overlay. Doing that means the outline is fully
opaque, which is fine for HC obviously but was excessive for Adwaita.
https://bugzilla.gnome.org/show_bug.cgi?id=787757
It used $text_color unconditionally, but in :dark, text is white, so we
overlaid a white tick on any light colours, all the way to white itself.
Using these named colours doesn’t make practical or semantic sense.
Instead, use white/black over dark/light swatches, as in HC, so all
variant–swatch combos work. Light looks the same, & :dark works now.
For backdrop, use alpha 0.5, unlike 0.7 in HC, as that seemed excessive
& different from the current effect. 0.5 is almost identical to how
$backdrop_fg_colour is a 50% mix of $fg_color, & matches backdrop text.
https://bugzilla.gnome.org/show_bug.cgi?id=787531
For dependencies that do not generate pkg-config files for their Visual
Studio build systems, we need to look for them using cc.has_header() and
cc.find_library(), namely for Cairo and HarfBuzz, if one does not have
crafted pkg-config files for them (which, by themselves may be
error-prone).
As a result, we will still try to look for Cairo and HarfBuzz using
pkg-config, but will give another shot at them on Visual Studio using
cc.has_header() and cc.find_library() if they couldn't be found via
pkg-config.
https://bugzilla.gnome.org/show_bug.cgi?id=785210
The Vulkan .lib file that is supplied by the LunarG Vulkan SDK is
vulkan-1.lib, not vulkan.lib, so make sure we look for the right
libraries when building on Visual Studio (I am not sure whether the
LunarG SDK will work for MinGW/mingw-w64 builds, as only Visual Studio
.lib files are provided).
Note that this will require one to set LIB and INCLUDE appropriately to
find the Vulkan .lib and header files, and possibly PATH if one's video
drivers do not contain the Vulkan runtime DLL.
https://bugzilla.gnome.org/show_bug.cgi?id=785210
Visual Studio does not support things like -Wl,export-dynamic, so we
need to export those symbols by using __declspec(dllexport). So, we
decorate these with macros which we define accordingly for this purpose.
https://bugzilla.gnome.org/show_bug.cgi?id=785210
On Python-3.x, we need to set the encoding when opening files, when this
script is run, as it might contain items that are not supported by the
system's locale (for example, non-English Windows). So, we use a
wrapper to set the encoding on Python 3.x, but open the file as we did
when using Python 2.x, since file encodings are not supported there.
https://bugzilla.gnome.org/show_bug.cgi?id=785210
This class is not added by any widgets nor themed by Adwaita/HC.
However, it is presented here as if it does something. It doesn’t.
But we changed the 2 buttons with the .raised class to use symbolic
icons, unlike their ‘unraised’ counterparts, which is unnecessarily
confusing and might make people think .raised affects icons somehow.
So, make them use the same icons in all cases; that way, if .raised is
ever made to do anything, 6 years later, what it does will be clear.
https://bugzilla.gnome.org/show_bug.cgi?id=644248
Instead of showing the 4 types except for GTK_SHADOW_NONE, which are all
treated identically and provide no way for themes to differentiate, just
keep 2 Frames, and make one of them GTK_SHADOW_NONE to demo a flat Frame
along the orthogonal orientation. It seems a FlowBox on its own can only
handle being shrunk along its main orientation. The orthogonal requests
a huge min size – reserving what it would need if the main orientation
got its min size, which would flow all children in 1 line orthogonally.
Adding it to a ScrolledWindow (any policy) enables free shrinking, so
size_allocate() can reflow how users in this situation probably expect.
https://bugzilla.gnome.org/show_bug.cgi?id=787021
Without specifically connecting ::delete-event to something, the dialog
will be destroyed when it is closed, for example by pressing Esc. This
meant that when dismissing it this way, unlike by pressing Cancel, any
custom palette would be lost when the dialog was next opened, and so on.
Resolve this by making ::delete-event just do GTK_RESPONSE_CANCEL, so
closing the dialog has the same effect as clicking its Cancel button.
https://bugzilla.gnome.org/show_bug.cgi?id=787444
Make it slightly more obvious when things are about to slide sideways
because a NULL GtkSettings has been returned to a caller. This is a
valid return value, but is rarely handled correctly.
https://bugzilla.gnome.org/show_bug.cgi?id=778382
As reported in https://github.com/ibus/ibus/issues/1944,
typing u201e while holding Ctrl+Shift used to give a „
when letting go of Ctrl+Shift. This broke when we introduced
Ctrl+Shift+e to start Emoji sequences. Fix this by only
looking for Ctrl+Shift+e if we are not already in a hex
sequence.
This is just a proof of concept - we use a single 1024x1024 surface,
and just give up when we run out of space. The cache is populated
incrementally, and items are never removed.
This commit takes several steps towards rendering text
like we want to.
The creation of the cairo surface and texture is moved
to the backend (in GskVulkanRenderer). We add a mask
shader that is used in the next text pipeline to use
the texture as a mask, like cairo_mask_surface does.
There is a separate color text pipeline that uses the
already existing blend shaders to use the texture as
a source, like cairo_paint does.
The text node api is simplified to have just a single
offset, which determines the left end of the text baseline,
like all our other text drawing APIs.
This is meant to cut down build time in flatpak and similar
situations. Since it produces technically incomplete builds,
we list these options in the status output at the end of
the meson run.
This fixes the proper dependencies getting set up for generating
the shaders and only the necessary things getting rebuilt on
resources changing in gsk.
We were only selecting a section’s button if the adjustment y coord was
within its heading, so scrolling slightly into it unchecked all buttons.
This also fixes how we could end up with the first 2 selected, somehow.
https://bugzilla.gnome.org/show_bug.cgi?id=787172
Add integration of the libcloudproviders DBus API to the
GtkPlacesSidebar by showing name and sync status of the cloud providers.
The exported menu is rendered as a GtkPopover.
The sidebar will be updated if the list of cloudproviders changes e.g.
by adding or removing an account. If any cloud provider changes detailed
information like sync status only the individual sidebar row gets
updated.
Co-authored-by: Carlos Soriano <csoriano@gnome.org>
Co-authored-by: Daniel Boles <dboles@src.gnome.org>
https://bugzilla.gnome.org/show_bug.cgi?id=786123
Use opacity to differentiate unselected/hovered/selected buttons. It had
assumed bg < border < fg colours, which may be false, as in Adwaita:dark
This also means we do not need to special-case for the backdrop state.
https://bugzilla.gnome.org/show_bug.cgi?id=786956
I see 'out of memory' errors and crashes inside libvulkan when
creating nodes that have empty bounds and end up in the fallback
paths, like a shadow around an empty text node. Prevent this
by not creating text nodes in that case.
in a specific case, which was applying .slider as a class on the parent
switch, instead of correctly selecting on its child node named slider.
This makes the border on the outside of a switch in a selected listbox
row look better in the light variant. Since the code was never removed,
it was clearly meant to work, and making it work is a clear improvement.
Using this produced warnings about the Pango syntax of <Family> <size>
being deprecated, and the size being invalid due to no unit specified.
Also, that multi-word font family presumably wouldn’t work as expected.
This reverts commit 98e3018455.
As an English-speaker, I know nothing about complex grammar, and it’s
been brought to my attention that some languages might differ in the
translation of the same command depending on where it appears.
So, I’d better assume everyone else knows better than me. Apologies!
Currently, this information is not used since cairo_show_glyphs
deals with color glyphs for us. But when we get to uploading
glyphs to a texture atlas, we will need it to do the right thing.
We don't look at individual glyphs here, but just whether the
font has the has-color flag set. In practice, all glyphs in
such a font will be color glyphs, and we can avoid loading all
the glyphs this way.
The emoji chooser gets disposed already, because it is attached
to the toplevel as a popover. Doing it again when the object data
is cleared is leading to a crash.
https://bugzilla.gnome.org/show_bug.cgi?id=787103
Copy the PangoCairoRenderer into GTK+, rename it to GskPangoRenderer,
and strip it down far enough to build without private pango apis.
This means we currently don't support hexboxes or shapes.
Currently, this lives in gtk, but it might be nicer to put it
in gsk eventually.
• Use disconnect_by_data() to catch both _adjustment_changed() and now
_adjustment_value_changed(), as the latter had been missed until now.
• Also disconnect from indicator_value_changed(), which was not done in
destroy() due to indicator_reset() and remove_indicator() disagreeing.
https://bugzilla.gnome.org/show_bug.cgi?id=775074
Do not connect to get_settings_for_screen() if we have no screen…
Use g_signal_connect(), not connect_object(), to match how set_screen()
makes this same connection, and how finalize() already disconnects it.
https://bugzilla.gnome.org/show_bug.cgi?id=705640
Since the move from button-press to gesture events, Shift-clicking did
not work to start a selection (from none) or truncate an existing one.
This was due to the code being copy-pasted around and some logic being
broken in the process. This makes both of those work as they should, by
shuffling it again so the end result is the same as before. Highlights:
(1) ::button-press if extending due to a single press would call
set_positions(tmp_pos, tmp_pos), which is what made the Shift+click to
create a selection work. That was lost. Add it back to make that work.
(2) ::button-press in the “Truncate current selection” branch would not
execute all the stuff around “extend_to_left”, as that was the else
case. So, set extend_selection = FALSE so we skip over that later on.
(3) BUT! This Truncate case never fired because it was in the else
branch of if (in_selection())! Of course, it must be in the true branch.
(4) The IM context was not reset if the Shift-click occurred within an
existing selection, only if it did not. In ::button-press this was the
first thing done if extending a selection, regardless. Make it so again.
https://bugzilla.gnome.org/show_bug.cgi?id=780750
Themes should not enforce min sizes on blocks in continuous mode; in
this case, the filled block should be as large as it needs to be to
reflect the current value, and no larger or smaller than that. So, the
fact that the minimal size was selected on just levelbar block is wrong:
we should also require the levelbar.discrete class to apply min sizes.
The widget should enforce whatever correct minimum size results from the
above fix, by reapplying commit 78b4885fe8
Except: we should not allocate/draw the filled block if the value is 0,
as in this case, the LevelBar should be empty, not have a min-size fill.
This partially reverts commit 96062ffeae,
as it makes sense to set min sizes for discrete blocks, so keep that in.
https://bugzilla.gnome.org/show_bug.cgi?id=783649
This reverts commit 8c0e5adaab.
This is actually needed since GtkHeaderBar will allocate and snapshot
widget that coun_visible_children does not consider.
Under X, we were not setting the right drag cursor initially,
because at current_action == action == 0, initially. Fix this
by explicitly using the right cursor when grabbing.
Subsurfaces don't currently work with our new rendering,
and this makes popovers unusable. We can go back to using
subsurfaces for popovers when this is fixed.
.update_position() enforces that non-Wayland platforms must position a
Popover within its parent Window. We use the allocation of the Window
to translate the position and check for overshoot on each of its sides.
Calling Widget.get_allocation() of a CSD Window includes its shadows.
But shadows were not excluded from the area in which we can position.
Thus, Popovers could get positioned in the shadow of CSD windows, where,
at least on X11, no input is received. Therefore, positioning a Popover
over a shadow meant its child widgets within that area became unusable.
Fix by calling Window.get_shadow() and including it in the overshoot on
each side. This adjusts for how the allocation includes shadows, making
overshoots with and without shadows the same. Thus, we avoid considering
shadows as viable for positioning, favouring a side where input works.
https://bugzilla.gnome.org/show_bug.cgi?id=786209
This prevents the load_fonts() function from switching to the "no fonts"
page and back when the model is reloaded. Given
GtkSettings::gtk-fontconfig-timestamp is 0 on Wayland and style changes
happen often, the stack change messes up popovers and pointer focus
on the fonts treeview and test entry.
https://bugzilla.gnome.org/show_bug.cgi?id=784723 introduced support for
native file chooser dialogs on macOS, but due to the use of generics in
the patch, there will be compilation errors on pre-Xcode 7 platforms,
such as Mountain Lion and Mavericks.
I strongly recommend to revert this patch when the oldest supported
macOS release is bumped to Yosemite (10.10).
https://bugzilla.gnome.org/show_bug.cgi?id=785306
Instead of gtk_widget_draw'in the inspected widget inside the
magnifier's ::draw handler, just create a new GtkSnapshot and snapshot
in its snapshot handler, similar to what GtkStack is doing.
gtk_widget_draw_internal is now only used inside gtkwidget.c, so remove
the prototype from gtkwidgetprivate.h. And since all incovacations call
it with clip_to_size=TRUE, remove that parameter.
We cargo-culted this from Autotools, but GCC on Windows supports the
same __declspec syntax as MSVC. The only difference is the additional
flag needed for GCC-like compilers.
The linker on macOS does not support '=' in its command line; there's no
guarantee that we are using the correct compatibility versions compared
to the Autotools build, but for that we'll need to build GTK+ master on
macOS.
This property contains 5 integers, of which the last 2 respectively
contain the tool serial number and tool ID. We were only extracting the
first so far, but GdkDeviceTool also has API getters for the latter,
which remained 0.
https://bugzilla.gnome.org/show_bug.cgi?id=786400
They are not usually yellow anymore, the previous advice about how to
style them was for pre-3.20 versions, and the immediate replacement (CSS
class .tooltip) does not seem ready for primetime.
https://bugzilla.gnome.org/show_bug.cgi?id=784421
Spooky action at a distance is not really allowed in Meson, so the rules
to generate the SPV files should go in their own directory.
Tested by: Rico Tzschichholz <ricotz@ubuntu.com>
The ComboBoxes were initially empty, rather than reflecting the initial
values of the properties. The CheckButtons were only correct by chance.
Fix this by setting the initial values on the widgets and binding them
to the properties using SYNC_CREATE, so the two are always synced up.
https://bugzilla.gnome.org/show_bug.cgi?id=786209
No longer store variation sequences explicitly. Instead, put a 0
in the sequence where the modifiers will be inserted. This is more
compact, and it allows us to put variations directly into the
recent section.
Update the type of the recent-emoji setting to match these changes.
Add an "Insert Emoji" item to the context menu in entries.
We also add a show-emoji-icon property, which when set to
TRUE, will add an icon that can be clicked to bring up
the Emoji chooser.
When the popover is dismissed, we return the focus to
where it came from. However, by using gtk_widget_grab_focus,
we were messing up the selection if that widget happens to
be an entry. Special-case GtkEntry and use
gtk_entry_grab_focus_without_selecting to avoid this issue.
The json file is imported from the (MIT-licensed) emoji.json[0] node
module, which generates it from the emoji list published by the
Unicode Consortium.
This commit also adds a little tool to convert the data into
a compact GVariant, and the result of that conversion, which is
added to libgtk as a resource. The following commits will make use
of it.
[0] https://github.com/amio/emoji.json
We create various windows during the initial creation of display
objects, which causes some bootstrapping issues when we try to
find the default screen to get its root window. To work around this,
pass the display object into gdk_window_new.
This is not an API change, since gdk_window_new is no longer public API.
Interpret NULL as "root window" here - we only have one
screen nowadays, so there is no choice involved, and this
will let us avoid dealing with the root window in the
fontend code.
menu margins have been added has a hack to mitigate bug:
https://bugzilla.gnome.org/show_bug.cgi?id=591258 with gtk+4 this
doesn't work anymore on gtk+4, the margin should probably be moved
to the parent window node, but it's not selectable, commenting out
for now.
Since gtk+ draws more than the widget and allocates more size to it than
it knows about, this flag doesn't work anymore. Removing it (or setting
it to TRUE for widgets that used to set it to FALSE) fixes drawing
invalidation when these widgets get allocated a new size.
In gtk_container_real_set_focus_child(), we try to scroll to the
position of the new :focus-child if we have h or v adjustments.
gtk_widget_translate_coordinates() returns FALSE if neither widget is
realized or in other situations that cause output parameters x and y not
to be set. Thus, if the caller did not initialise x/y and uses them even
if the function returned FALSE, they are using uninitialised variables.
In gtk_container_real_set_focus_child(), we did not check the return
value but merrily went ahead and used x and y regardless. This is UB, as
revealed by Valgrind, as well as being pointless.
The trivial fix is to exit early if (!gtk_widget_translate_coordinates).
https://bugzilla.gnome.org/show_bug.cgi?id=776909
Commit 885bcd9fe4 trampled the bit here
that is meant to translate between the nominated focus child and the
actual innermost one that is used for updating the h/v adjustments.
So, we need to save the passed focus child before diving into its
children, then translate and get allocations between them both. This
makes GTK+ 4 behave like GTK+ 3 again: instead of priv->focus_child and
focus_child, we now have focus_child and child, serving the roles of the
nominated focus child and its innermost focus child respectively.
This also ditches the unnecessary call to Widget:get_focus_child(), as
Container::set_focus_child() gets that same new child as an argument.
process-stop-symbolic is unintuitive if represented as a stop sign as in
Adwaita, and completely ambiguous if represented as a cross like the
window close button in other icon themes.
Instead, use application-x-executable, which is already used elsewhere
as a fallback if no specific icon can be found for the application.
https://bugzilla.gnome.org/show_bug.cgi?id=784624
Don't beep when modifiers are released in entries.
This was an inadvertent change that snuck in with
the emoji support.
Also, don't beep while entering an emoji name.
There is entirely too much beeping here.
In GTK+ 2, the ch < 0x80 was ORd with klass->latin1_to_char, and that
was unconditionally set to TRUE in the class init function, so
effectively the ch < 0x80 never mattered before or served any purpose.
When klass->latin1_to_char was deleted from the class in commit
f760538f17, this check’s sense changed.
The resuls was that accel keyvals with gunichar value >= 0x80 stopped
being rendered as symbols, instead falling back to their keysym name.
Instead of recognisable symbols for these, we get raw, often obscure,
and untranslatable keysym names. This breaks accessibility as well as
client users who may be parsing such accels and migrating from GTK+ 2.
So, remove the < 0x80 to restore the behaviour from before said commit.
https://bugzilla.gnome.org/show_bug.cgi?id=783906
This commit adds some basic support for entering emoji by name
to GtkIMContextSimple. To begin an emoji sequence, use Ctrl-Shift-e
instead of Ctrl-Shift-u that is used for hex input. Otherwise, the
behavior is the same: you can can let go of the modifier keys and
end the sequence with space or enter, or hold on to the modifier
keys and end the sequence by releasing them.
Only a limited, fixed set of names is supported at this time, see
the GtkIMContextSimple docs for a full list.
• Add GtkLayout as a @See_also since it includes fixed-pos functionality
• Drop mention of the long-gone Linux framebuffer port
• Explain how to work around the problems with RTL text
Being addable to a ScrolledWindow is not interesting; now that SW
auto-adds a Viewport if needed, so can DrawingArea and any other widget.
Mention GtkFixed in case the reader just wants that bit of functionality
This adds support for the shortcut inhibitor protocol in gdk/wayland
backend.
A shortcut inhibitor request is issued from the gdk wayland backend for
both the older, deprecated API gdk_device_grab() and the new gdk seat
API gdk_seat_grab(), but only if the requested capability is for the
keyboard only.
https://bugzilla.gnome.org/show_bug.cgi?id=783343
We wait for a few 100ms for rendering to settle in various WMs. So far
we only did that for windows that were controlled by the WM (aka
toplevels).
With modern compositing gnome-shell however, this now also applies to
override-redirect windows, so we now wait there, too.
This makes the reftests a lot slower, but they now actually work when
running make check in gnome-shell.
If query.return_type is not one we want, binding_compose_params() is
not called, and so params remains a NULL pointer. However, the code was
then unconditionally iterating it regardless. Don't if it is still NULL.
CID 1452218 (#1 of 1): Explicit null dereferenced (FORWARD_NULL)
15. var_deref_op: Dereferencing null pointer params.
This would only happen if the last element was deprecated, but it should
be avoided anyway.
CID 1388852 (#1 of 1): Out-of-bounds read (OVERRUN)
12. overrun-local: Overrunning array pseudo_classes of 16 32-byte
elements at element index 16 (byte offset 512) using index i + 1U (which
evaluates to 16).
This function clearly assumes the parameter children cannot be NULL, and
the call sites seem to perform enough checks to confirm this.
CID 1388869 (#1 of 1): Dereference before null check (REVERSE_INULL)
check_after_deref: Null-checking children suggests that it may be null,
but it has already been dereferenced on all paths leading to the check.
CID 1432024 (#1 of 1): Uninitialized scalar variable (UNINIT)
2. uninit_use_in_call: Using uninitialized value rect.x when calling
calendar_arrow_rectangle.
Add a default case to the switch which will bail out with
g_assert_not_reached(), which should reassure Coverity that the method
is always called with a valid value that is handled in the switch.
If value->values[i] is NULL, then values[i] was left uninitialised.
The code then reads each element of values[].
CID 1432029 (#1 of 1): Uninitialized pointer read (UNINIT)
11. uninit_use: Using uninitialized value values[i].
Our ::query-tooltip handler first checks whether the pointer is over any
of the icons, returning their tooltip if so, and if not chains up to
Widget::query-tooltip in order to show the text for the widget overall.
But ensure_has_tooltip(), which exists to update :has-tooltip based on
whether ::query-tooltip is needed, only set :has-tooltip to TRUE if any
icon had a tooltip, without caring whether the widget as a whole does.
That is asymmetrical and meant that if the Entry had a tooltip, but
subsequently all icons had their tooltips unset, :has-tooltip would be
set to FALSE, and hence the tooltip for the widget would become lost.
The fix is to set :has-tooltip to TRUE if the widget has a tooltip of
its own, and we only need to check the icons if that is not the case.
https://bugzilla.gnome.org/show_bug.cgi?id=785672
glib-mkenums is now done in Python, but since the Visual Studio build
environment (cmd.exe) does not support shebang lines, we need to call
the interpretor explicitly to run the script.
This means that we need to update on how we generate
gsk/gskenumtypes.[c|h] in our projects, as at this point GTK+-3.91.x
does not require a GLib installation that ships with the Python-fied
glib-mkenums. As a result, we adapt to this by first using Python
to call glib-mkenums. If this fails (where the output file becomes 0
in size), then we use PERL to call the glib-mkenums script. Note that
during the build this will cause a warning message to be displayed,
stating that '&' cannot be found, but due to the way Windows .bat script
are done, we need to live with that until a solution can be found on
this.
This is likely a problem that does not exist in the Meson builds, as
Meson will take care of calling the interpretor for us by looking at
the shebang lines for our case.
Also, clean up the .batin Windows batch script that is used to call
glib-mkenums by using a for loop in there.
Just to test tooltips in all cases; what was already here
should have been sufficient, but this doesn't hurt.
While here, also add some instructive placeholder text.
https://bugzilla.gnome.org/show_bug.cgi?id=780938
Use conditionals to select the Python installation, so that we can stick
closer to the default Visual Studio versions used to compile each official
Python releases.
This means by default:
-2013 builds use Python 3.4.x, which is built with 2010
-2015 and 2017 builds use Python 3.6.x, which is built with 2015.
Also rename PythonPath/PythonPathX64 in the property sheets to
PythonDir/PythonDirX64 repsectively, as PythonPath is the envvar name
where additional Python modules is searched for, so we don't want to get
confused with it.
Last but not least, distinguish between the Python interpretors that are
used on x64 and x86/32-bit builds for generating the libgtk4.manifest
file and the gdbus-generated sources, for consistency reasons.
Refactor the code updating the active link under the current coordinates
into a separate function, and call it on GtkGestureMultiPress::pressed
so the link is updated on GDK_TOUCH_BEGIN. Based on a patch by
Jan-Michael Brummer <jan.brummer@tabos.org>.
https://bugzilla.gnome.org/show_bug.cgi?id=776903
It is not necessary to (re)set the cursor on every crossing
event, and can probably yield the wrong results if there are
multiple master devices involved. Just set it on init(), and
let the inner machinery update the cursor whenever necessary.
This patch is an adaption of commit 0daf79676 in gtk-3-22, the
side effects are not as bad here because the cursor was already
being set on the widget specifically instead of the parent
widget's, but there's still some nonetheless (plus, it's simpler)
https://bugzilla.gnome.org/show_bug.cgi?id=785375
This check must be done explicitly on Wayland as the master device for
tablet tools differ from the Core Pointer. This ensures that whenever a
tablet tool is inside a window and the cursor is programmatically changed,
it will be visually updated too.
https://bugzilla.gnome.org/show_bug.cgi?id=785375
Replace uses of VLAs (variable-length arrays) using g_newa(), since
Visual Studio builds will unlikely ever support VLAs (which became optional
in C11).
https://bugzilla.gnome.org/show_bug.cgi?id=773299
Adds support for creating scroll events from Wayland tablet wheel events.
Even though no Wacom tablet puck has a smooth-scrolling wheel, both event
types need to be generated to make the upper layers happy.
https://bugzilla.gnome.org/show_bug.cgi?id=783716
If a tablet device is used to perform actions like window moving or resizing,
GTK must provide the correct implicit grab serial number over Wayland to Mutter
in order for the action to succeed. This commit adds tablet support to the
implicit serial getters.
https://bugzilla.gnome.org/show_bug.cgi?id=777333
Since gtk_bin_add does a gtk_widget_set_parent call, we cannot use it in
a GtkBin implementation that has multiple child widgets and cares about
their order.
If a bad behaving application tries to make the window/display beep too
often, throttle the beep requests so that we don't end up filling the
Wayland socket queue.
The throttle is set to 50 beeps per second, which far more beeps than
will ever make any sense from a user experience point of view, but will
avoid terminating due to an excessive amount of requests.
https://bugzilla.gnome.org/show_bug.cgi?id=778188
So we can avoid creating a GtkCssPathNode in _init and then throwing it
away right after when using the _new_with_node constructor, which is the
one we use for all widgets.
Don't set the have_focused field of the window's toplevel to TRUE by
default and don't set the FOCUSED state in gdk_window_map. This a means
toplevel window's state is what the WM expects, and the FOCUSED state
will be set anyway when we map the window and receive a _NET_WM_STATE
message.
Since setting a clip is mandatory for almost all widgets, we can as well
change the size-allocate signature to include a out_clip parameter, just
like GtkCssGadget did. And since we now always propagate baselines, we
might as well pass that one on to size-allocate.
This way we can also make sure to transform the clip returned from
size-allocate to parent-coordinates, i.e. the same coordinate space
priv->allocation is in.
Remove the special case in gtkwidget.c where we didn't draw any css
background/border for popovers. Instead, rely on themes to not style the
popover node and add a contents gizmo that gets the actual css styling.
We then requeste enough space for the popover to draw both the contents
and the arrow on the side.
The reported minimum baseline is for the reported min height, but if the
css min-height is greater than that, we need to account for that fact
when saving the baseline.
Since the reported baseline is relative to the widget's origin, we also
need to add the top values for margin, border and padding to the
reported baseline.
We claimed the gesture previously to keep it from propagating to the
underlying entry, but now that the entry is in a box with the two
buttons, we can do this properly and restore the previous long-press
behavior.
It's getting harder and harder to find a dummy style property to use
here, so remove the test case since style properties should be going
away soon anyway.
Previously, we would request a size of 0×0 when the transition type was
NONE and the child un-revealed, making the revealer in this case a
gtk_widget_set_visible replacement. Instead, to the exact same thing we
do in the CROSSFADE case and request the child size instead. This also
keeps the revealer from under allocating the child when the transition
type is set to NONE.
Instead of hopping through 7 different functions to do that, just
remove all rows directly. This also mean we'll only remove rows and not
other children that've been added like placeholders.
Add :dir(ltr) where expected, i.e. everywhere we now have a widget but
had a gadget before.
Also, fix the expected output to expect mark subnodes in the order
specified in the GtkScale does, i.e.
├── mark
├── [label]
╰── indicator
for marks at the top of the scale and
├── mark
├── indicator
╰── [label]
For marks at the bottom of the scale.
Checking the given GtkAllocation against the current allocation insize
::size-allocate doesn't really work anymore. They are only different if
the content allocation (the one passed) and the widget allocation (the
current one) are different, so e.g. when the widget has padding >0
applied.
Since we get offset automatically to the widget allocation before
->snapshot is called, we still have to offset the difference to the
position of the content allocation.
We don't need to care in this case since the default values should
always be assumed to be 0, and setting a baseline of 0 is just wrong
when orientation == HORIZONTAL, it should be -1 (or unset).
This is optional for positive margins as they just increase the widget
allocation. However, with negative css margins, the allocation is
smaller than the clip.
This fixes scale sliders leaving a small trail behind.
This fixes the expansion not working. As a GtkBin, GtkExpander can only
have one child and if that's a GtkBox (and not the one added through
gtk_expander_add), things go wrong.
always initialize clips to the (content) allocation, don't walk up the
widget hierarchy in gtk_widget_set_clip, implement
gtk_widget_size_allocate in GtkSeparator. This way we don't end up using
uninitialized clip values.
The entire clip handling is up for major rework since we can't and don't
want to force every single widget to call _set_clip in size-allocate
implementations.
If widgets chain up in their size-allocate implementation, they pass the
content allocation and not the widget allocation which will cause the
wrong allocation to be set.
We need to adjust the passed for_size to fit into the content allocation
of the widget.
That also means that we can't call gtk_widget_measure(widget) inside
gtk_widget_measure(widget) since now the for_size will be adjusted
twice.
Events that get to gtk_main_do_event() have the toplevel GdkWindow
as event->any.window. Also, ensure that coordinates fall within
sensible places of the windows, since those might have shadows,
headerbars and whatnot on wayland.
That means the whole hierarchy is getting destroyed, leaving those
behind incurs not only in a leak, but also on weak refs (and unintended
repick) to happen in the wrong moment.
Showing all the different errors and warnings when renaming and creating
files/folders without potentially resizing popovers on every keystroke
requires us to know the size of the error messages beforehand, so pack
all of the possible error messages and warnings in labels and those into
a stack. This way we can also neatly crossfade transition between them.
https://bugzilla.gnome.org/show_bug.cgi?id=775636
GdkPixdata is deprecated. Warn when the application tries to load
pixdata embedded resources. The application developer will have to
remove the "to-pixdata" keyword from the GResource definition file.
https://bugzilla.gnome.org/show_bug.cgi?id=781583
The glib-genmarshal tool from GLib 2.54 added various command line
arguments that allow us to remove a bunch of as hoc manipulations of
the generated marshaller source files. The marshal generator tool can
now include an header in the source, and undef the G_ENABLE_DEBUG
pre-processor symbol for us. It can also generate the prototypes of the
marshallers in the C source, and avoid a 'missing-prototypes' compiler
warning.
Use the new predictable request object path and connect
to the Response signal before issuing the portal call.
This avoids a race that is pretty unlikely to hit in
the filechooser case.
Wacom tablets often have a "pad" device which houses multiple buttons. At
present, these devices are incorrectly marked as GDK_SOURCE_PEN which can
cause problems for some software.
https://bugzilla.gnome.org/show_bug.cgi?id=782040
Commit b52966a318 stopped the parser from
handling various deprecated pseudoclasses, which were aliases of others,
but it did not update the documentation to reflect that they were gone.
The label measuring code was only determining baselines
when the label was set to wrap, which does not seem right.
Non-wrapping labels have a meaningful baseline as well,
report it back.
When there is no externally allocated baseline, we should
do the same thing that GtkBox does, and determine one from
the children that want baseline alignment.
This commit adds a GtkCenterBox::baseline-position property
with setters and getters.
By relying on GtkSpinButton default activation behavior, the
collate icon doesn't get updated when a new number is typed
in the copies spin button.
https://bugzilla.gnome.org/show_bug.cgi?id=759308
Add beginning double asterisks and function names. Correct the parameter
names (next/previous_child -> next/previous_sibling). Make the documentation
of the two functions more similar.
https://bugzilla.gnome.org/show_bug.cgi?id=783445
The imcontext internals have been changed to use set_client_widget
instead of set_client_window in order to remove API dependency on
GdkWindow. Update the Windows IME support so that the code will
continue to build and work.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
The callback function that is used by VkDebugReportCallbackCreateInfoEXT
is decorated with VKAPI_CALL (which is __stdcall on Windows). This is
not detected on x64 Windows as __stdcall is not really meaningful on x64
Windows, and VKAPI_CALL expands to nothing on non-Windows.
As __stdcall functions are treated differently on 32-bit Windows, the
32-bit compiler does require that the function be declared as __stdcall
so that things will compile, link and run properly.
https://bugzilla.gnome.org/show_bug.cgi?id-773299
Under Wayland, when multiple keys are pressed and the user releases a
key, key repeat should continue unless the key released is the one
currently repeating.
In the case of:
- key1 press
- key1 repeat
- key2 press -> key1 repeat stopped
- key2 repeat
- key2 release
The behavior should be to cancel keyboard repeat, though key1 is still
held down. This is consistent with prior X11/XWayland behavior.
The following also must work:
- key1 press
- key2 press
- key2 release
- key2 press
- key1 release
- key2 should continue to repeat
The fix for bug #778019 should continue to work:
- key1 press
- key1 repeat
- key2 press -> key1 repeat stopped
- key1 release
- key2 should repeat
The choice to change the counter nkeys to the flag repeat_active
helps to solve the second test case.
https://bugzilla.gnome.org/show_bug.cgi?id=781285
begin_resize_drag() and begin_move_drag() check for xdg_surface being
not null, but those apply on xdg_toplevel so they should check for
xdg_toplevel being non-null instead.
https://bugzilla.gnome.org/show_bug.cgi?id=781945
When an event is received while a tooltip is showing, the GtkTooltip's
event handling code can end up calling gdk_window_set_transient_for()
from gtk_tooltip_set_last_window().
The Wayland GDK backend will try to automatically create a subsurface
in gdk_wayland_window_set_transient_for() but if the parent surface is
gone meanwhile, this will will cause a crash when trying to create a
subsurface from a parent with a null surface.
Checking for the parent is not sufficient, we ought to check for the
parent surface as well to avoid the crash.
https://bugzilla.gnome.org/show_bug.cgi?id=782283
Applications can specify the type hint as utility even on toplevel
windows.
When that toplevel is also marked as a transient for another window,
GDK Wayland backend would translate that as an xdg_popup which is not
appropriate.
While utility temp windows should remain mapped as subsurfaces (such as
the ones used by treeviews), regular windows should not translate as
neither a subsurface nor an xdg_popup.
https://bugzilla.gnome.org/show_bug.cgi?id=781945
The code used SIGDN_URL to get an URL for the selected item, but Windows URLs
are a mix of unicode and percent encoded characters in the locale encoding
and not something GFile can understand. The result is a garbage file
path.
Instead use SIGDN_FILESYSPATH to get a real file path if available.
Also checks the return value of g_utf16_to_utf8 because file paths on
Windows can contain lone surrogates which would make the conversion fail.
https://bugzilla.gnome.org/show_bug.cgi?id=783347
Another selector forces round corners for headerbars in a stack, and it
has higher priority than the selector covering the non-stack case from
commit 796f9b5bfb. Totem’s MainToolbar
happens to be in a stack, and we should maintain symmetry here anyway.
So, as window classes .maximized and .tiled are excluded from this other
selector, the newly handled .fullscreen case must be excluded here also.
https://bugzilla.gnome.org/show_bug.cgi?id=770513
Totem uses a fullscreen window with a headerbar at the top, and without
this change, that headerbar has rounded corners, which look different
from a maximised window and let video content show through beneath.
https://bugzilla.gnome.org/show_bug.cgi?id=770513
There is no need to have every application log a warning when the
Wayland display server goes away, and we are using _exit instead of
exit elsewhere.
This is also what the X11 backend does (see gdk_x_io_error).
https://bugzilla.gnome.org/show_bug.cgi?id=745289
As we now refrain from sending the crossing events if there's an
implicit grab, those events must be sent on button release when
the implicit grab is broken.
Check the grab widget (both explicit and implicit) and check for a cursor
from the target widget up to this grab widget. If the target widget is
outside the grab widget, only the grab wigdet's cursor will be checked.
This also means that we have to ensure the cursor is updated on button
releases, as an implicit grab being deactivated must trigger a cursor
lookup from the target widget.
In these situations we must perform the "is it claimed" check before removing
the (touch)point, as doing so when the gesture is empty will be too late if
the gesture actually claimed input.
This just applied to child windows, but now GDK should just take care of
toplevels, which shall get crossing events from the windowing when the right
conditions apply.
Removing this code fixes confused crossing state in widgets and messed up
window_under_pointer tracking (Which now is meant to be toplevels) when any
of the remaining child GdkWindows trigger these crossing events.
For some reason this wasn't done on windows with an impl, but it totally should.
Probably hidden by grabs in menus and somesuch being done on a child window.
We already issue the first _get_parent call before even entering that
loop, so make sure `parent` is not NULL. This happens when event_widget
is already a toplevel, and this change fixes row-dragging in treeviews.
Drop the in_widget flag since motion events the listbox receives are
always inside the listbox. Also drop the manual coordinate translation
code using GdkWindows.
We don't draw or size-allocate the titlebar when the window is
fullscreen or undecorated, so reflect this by setting it to
!child_visible. This can happen when changing the value of the decorated
property while the window is shown.
Instead of delegating on the parent shell of a menu item/shell on a variety
of situations, Simplify event handling so:
1) Menu item selection is handled entirely on GtkMenuItem through crossing
events.
2) The deepmost menu shell handles clicks inside and outside of it.
This avoids the rather hard to follow gtk_widget_event() calls going on all
throughout the handling of crossing and button events, and makes menus work
again.
As event->any.window is the toplevel, this is not useful anymore to
determine the window/widget that is the target for this event. Add
helper functions to attach user data to GdkEvents so the target
widget can be stored on the gtk/ side.
These calls should be made private with the rest of GdkEvent related
API.
It's not necessary anymore for clipping nor receiving events. So just
remove it. The event handling code was expecting events in bin_window
coordinates, and have been updated to relying on widget-relative coords.
We can just replace window comparisons with coordinate matching, the
cursor corresponding to edges is now set in a capture-phase motion
handler, as cursors aren't set on GdkWindows anymore.
It's not necessary anymore to receive input events. The pan gesture has
been set on the capture phase as the child widgets may capture during
bubbling.
There should be no circumstances where an implicit grab is requested but
no focus exists, there's however circumstances (like windowing grabs taking
input to a different window) where we might get implicit grabs being undone
when then new window didn't create a focus for the pointer itself.
Only if they fall outside the grab widget, in that case the widget holding
the implicit grab won't be receiving events anymore, so we can just undo
it.
We now rely on toplevels receiving and forwarding all the events
the windowing should be able to handle. Event masks are no longer a
way to determine whether an event is deliverable ot a widget.
Events will always be delivered in the three captured/target/bubbled
phases, widgets can now just attach GtkEventControllers and let those
handle the events.
Those are now needless and wrong, as we get guarantees that handled
events will contain widget-relative coordinates. A side effect is
that these events are very possibly not explicitly sent to the
GdkWindow that implementations expect, any extra checks performed
through gtk_gesture_set_window() will be wrong, so the function has
been dropped entirely.
And refurbish cursor management to be set on the GtkWidget. The
input window is not needed anymore to receive events either.
This is no longer set through the GdkWindow, so use the private
GtkWidget API.
The event shall no longer be "directed" to the event window, but the
widget. Getting a enter/leave event is enough now to know whether the
pointer is inside or outside the widget.
Unlike GTK+ grabs which are global to all/one device, the implicit grab
is per focus, which means each may have implicit grabs on different or
the same widget.
Now that gtk_main_do_event() is able to handle pointing events in toplevel
coordinates, forward all of these as is. Just minimal handling is still done
on the gdk side for GDK grab accounting, and toplevel tracking for each
pointer.
Implement target finding per-pointer/touchpoint through GtkPointerFocus and
_gtk_toplevel_pick(). Focus changes are handled through the emission of
crossing events between the old target and the new one.
Each toplevel will keep its own tracking of the current ongoing foci,
add the plumbing that will allow to create/update/remove those as they
come and go.
These objects (tied to a toplevel) track the focus of a pointer/touchpoint.
The info in these basically consists of current toplevel coordinates and the
current target widget.
This function will be useful in other places, such as determining the
widgets that must receive crossing events after pointer picking points
to another widget.
Aborting the application makes it look like an application bug, when
it is the expected thing to do when the Wayland display server goes
way. eg., when the user logs out. The log level is also demoted to
avoid a storm of warnings in the log from all applications whenever
this happens.
This is also what the X11 backend does (see gdk_x_io_error).
https://bugzilla.gnome.org/show_bug.cgi?id=783047
The :last-child selector supposed to reset the border was
overridden by the :hover selector. This is fixed by moving the
:last-child selector after the overriding one.
Thanks to Sebastian Keller for spotting.
Fixes https://bugzilla.gnome.org/show_bug.cgi?id=779078.
…erties clobbered by commit c92b7d4224.
That and its counterpart were for removing :expand and :fill child props
from GtkBox, but they ended up catching these for GtkToolItemGroup too.
While GtkToolItemGroup still has these, we may as well keep demoing them
Fix the sizing and spacing, blue tags for the bright variant,
similar to what gnome-documents was shipping, and inverted gray
tags for the dark variant, not vanishing on hover.
It was only testing the default configuration, where overlay scrolling
is on and both scrollbars use POLICY_AUTOMATIC. We should also test the
other 3 configurations that are available by including non-overlay
scrollbars and/or those that use POLICY_ALWAYS.
https://bugzilla.gnome.org/show_bug.cgi?id=778853
POLICY_AUTOMATIC means scrollbars are only shown when needed, i.e. when
the size of the window is not large enough to show the entire child. So
when measuring the preferred size, such scrollbars should be ignored.
But measure() added size for *any* non-overlay scrollbar of the opposite
orientation, e.g. for horizontal size, it added the width of vscrollbar.
So we requested for child + bar, & having enough for child meant that the
policy hid the bar, leaving extra space empty below/right of the child.
Fix this by only adding size for such bars if they use POLICY_ALWAYS.
https://bugzilla.gnome.org/show_bug.cgi?id=778853
• Only calculate the specified dimension – rather than measuring both &
discarding the other (which will often be recalculated right after)
• Only measure a given child scrollbar if it may be visible, not always
• Move variables into narrowest scopes & otherwise improve readability
https://bugzilla.gnome.org/show_bug.cgi?id=778853
Some files that this script will process might have UTF-8 items in
there, which can cause problems on Python 3.x as it is more strict and
careful on unicode issues. Fix this by:
-Doing what we did before on Python 2.x
-Opening the file with encoding='utf-8' on Python 3.x
The user data passed when exporting a Wayland window was supposed to be
freed using the destroy_func, as is commonly done. This was previously
broken, as the user data was just NULL:ed when exported, and only
actually destroyed when unexporting before having exported.
While e016d9a5db fixed this, it introduced
a regression, as GtkWindow was nice enough to free the memory anyway
after having received the exported handle, causing it now to double
free.
https://bugzilla.gnome.org/show_bug.cgi?id=782109
Otherwise in GC-ed environments the `g_source_remove` call during
disposal might be called on an already removed source, which results in
unnecessary console output.
https://bugzilla.gnome.org/show_bug.cgi?id=778301
Use the gravity enum values when converting to gravity. It doesn't fix
anything, since the enum values were identical, but it makes a coverity
warning go away.
https://bugzilla.gnome.org/show_bug.cgi?id=780301
5bb12474d9 removed the dnd window movement code to let
the gdk backends handle the window movement instead. While this
works for X11/wayland the win32 backend still uses the unmanaged
interface and expects the window movement to be handled on the gtk
side. This restores the functionality in case the dnd is unmanaged.
This fixes the drag window on Windows being stuck in the top left
corner instead of following the drag position.
https://bugzilla.gnome.org/show_bug.cgi?id=781737
Creating with `gtk_popover_new_from_model` should be exactly the same as
if via `gtk_popover_new` plus `gtk_popover_bind_model`.
Also remove the style if the model is unbound at any point.
Try text/plain;charset=utf-8 first, before falling back to
X11-isms like UTF8_TEXT. This makes things work on Wayland
compositors that don't carry a heavy X11 legacy around.
https://bugzilla.gnome.org/show_bug.cgi?id=781814
The `-export-dynamic` flag is a libtool-specific flag; since we're not
using libtool with Meson, we should instruct the C compiler to use the
appropriate linker flag instead.
Copy the location of the test data and binaries from the autotools
build, even though it's not really correct; currently we install the
test data under libexecdir, but it should live under datadir, and we
should use `G_TEST_DIST` to figure it out.
The `state` subdirectory is missing.
The common compiler and linker flags control, among other things, the
default visibility of symbols; without them, we leak symbols that ought
to be private.
GSK has various enumeration types that are currently not used; while
they may go away, currently they are built and introspected. If we want
the introspection machinery to work, and still use static libraries to
build GDK and GSK into the GTK shared library, then we need to reference
the get_type() function of these enumeration types somewhere, to avoid
the linker discarding it, and thus breaking the build.
As luck would have it, we have an autogenerated bit of C that refers to
all the get_type() functions in the library; if we add the GSK types to
it, then we get the reference we're looking for, and the build succeeds.
We need to reference the types file directly, because it won't be copied
into the builddir by Meson — except for GTK, which needs to generate its
own types file using configure_file().
We're mixing a lot of styles in the Meson build files. This is an
attempt at making everything slightly more consistent in terms of
whitespace and indentation.
If glslc is found, rebuild the shaders from GLSL to SPIR-V; otherwise,
we're just going to use the built files we have committed in the source
repository.
When building GTK+ straight from the repository without any assistance
from packaging tools, we need to trigger system-wide updates, like the
icon theme cache update, or the schema compilation.
We can build the name of the input and output files for the Wayland
protocols we use from the protocol name, stability, and version. This is
similar to how the autotools build does it, except much more clear and
without shelling out twice to sed just to resolve the Makefile rule.
We need to check if the linker flags we use are available, depending on
the platform, and we need to ensure that the shared library is
versioned appropriately.
GTK symbols are not visible by default, and only the ones annotated with
_GDK_EXTERN (and wrapper macros) are exported. We need to define
_GDK_EXTERN during the configuration, depending on the platform and
compiler we use.
The autotools build checks the version of GLib we are depending on in
order to generate the appropriate GLIB_VERSION values for the
min-required/max-allowed defines.
We have to work around some ordering problems here. We still
manage to keep most of the guts in modules/input/meson.build,
so it's not too ugly overall.
(The autotools build solves this with a 'make -C ../../input/modules'
inside gtk/Makefile, but that's not something we can or want to do.)
Remove workaround for gcc bug (Meson does that now), and
construct the right config.h defines for the headers on
the fly instead of listing them in the build file, which
is more error prone.
Add back dependencies on libgdk_dep and libsk_dep which are declared
dependencies. We removed this before because these declarations had
link_with: lines that dragged in the static libgdk.a and libgsk.a libs
which are linked into libgtk-4.so anyway and thus shouldn't be used
when linking internal exes/tools against libgtk-4. Remove the static
libs from the declared dependencies and have libgtk link those in
explicitly, so that the declared deps now just provide all the built
dependencies and include dirs and such for declared libgtk_dep users
such as the internal exes/tools, which want all the generated gsk/gdk/gtk
headers to exist before attempting to compile anything against the
gtk+ headers.
gdk and gsk are no longer separate libs but part of gtk now, so any
Gtk+ user should just link to gtk, there's no need to additionally
link against all those static helper libs that go into the gtk lib.
This means we need to specifically add confinc to include_directories
in more places to make sure the right config.h (i.e. ours) gets
included and not a subproject's like graphene's config.h.
Not dragging in static libs also fixes the issue of all executables
having to be relinked for any and all changes. With this change
it's super-fast now and can be skipped for most changes that don't
touch the external ABI.
gdkprivate-wayland.h includes generated wayland client protocol
headers and is included from gdkdisplaymanager.c, so we need to
generate those client protocol headers first also when building
main gdk itself.
This is how it's done in the autotools build. Also avoids problems
with multiple source files having the same name (gdkeventsource.c).
Also move broadway backend code into broadway subdir.
Almost all of these tests include gtk/gtk.h so we need
to dep on libgtk not just libgdk. Otherwise compilation
fails because graphene.h include can't be found.
Add libgdk_dep as dependency to the libgtk_dep declare_dependency(), so
that the generated gdk includes are generated before anything is built
that tries to include gtk headers (such as various tests that don't depend
on gdk directly).
This is needed for the Meson port, a file name .c that's included
and shouldn't be compiled into an object is difficult to manage
otherwise. And it's not actually a valid .c file anyway.
This was only every implemented under X11, and with CSD,
this is clearly in the application realm. We should not
pretend that we can support it on the toolkit level.
https://bugzilla.gnome.org/show_bug.cgi?id=775061
We used to inject the inclusion of the generated header file into the
generated body of the marshallers source code in order to avoid compiler
warnings about missing prototypes. The glib-genmarshal utility has been
fixed in GLib to include the prototype in the generated source, so now
we're going to trip -Werror=redundant-decls.
With Wayland, GDK_DEBUG=events would log key events but not explicitly
state whether the event is a key press or release, or if it's
originating from a key repeat.
Add some more verbosity to make sure these informations are logged on
key delivery when GDK_DEBUG is set.
https://bugzilla.gnome.org/show_bug.cgi?id=781767
The rect parameter in gtk_gesture_multi_press_set_area is annotated as
nullable and the code handles the rect==NULL case, but the
g_return_if_fail kept that case from ever happening.
Turns out that the destination is the last parameter, not the first one.
This fixes the flickering in the first page of the widget-factory when
using the expander on page 2.
It is generally a good idea to license individual files under the
same terms as the project license (in particular when the mismatch
boils down to having copied the wrong license header), so relicense
the code under the LGPL.
https://bugzilla.gnome.org/show_bug.cgi?id=781422
In the else branch of the if statement before this one, we're assigning
*smallest = *widest anyway, so this if statement is never true. Move it
to the if block before instead, where it can apply.
gtk_widget_set_parent (via gtk_widget_reposition_after) will queue a
resize on the parent widget automatically when adding a child widget, so
unparent should do the same
The center widget in GtkBox was only introduced to use it in
GtkActionBar. However, the implementation there is much more complex
than it needs to be, so move the center widget into GtkActionBar instead
and later remove it from GtkBox.
This replaces all internal gadgets with widgets.
Remaining problem: "block" nodes have a min-width of 32px in Adwaita,
but when allocated in continuous mode, the levelbar doesn't care and
underallocates them.
GtkGizmo is the easiest possible widget to implement. It does nothing
except give its creator a way to control measure/size-allocate/snapshot,
so it can be used in a variety of use cases.
Insert the css node before setting a parent widget on the column button,
so the gtk_widget_set_parent won't attempt to add the css node as child
of the parent widget css node.
Translating it seems pointless if we can use a non-translatable example
such as gnome.org instead of foo.example.com.
This will help to make changes in here without breaking string freeze.
https://bugzilla.gnome.org/show_bug.cgi?id=781622
We were send the "open-location" signal without mounting first the
location if necessary, making the open in tab/window context menu not
work for those.
This patch makes sure we mount the location before emitting the signal.
https://bugzilla.gnome.org/show_bug.cgi?id=771269
Because the network monitor can perfectly be NULL,
the tests were failing on that for GtkPlacesView
always tries to disconnect this handler.
Fix that by only disconnecting the handler when
the network monitor exists.
https://bugzilla.gnome.org/show_bug.cgi?id=781195
GtkPlacesView exposes local access points and network
shares transparently by using the 'network:///' URI,
which is handled by GIO.
Currently, however, it doesn't monitor the network
for new available points, such as computers that just
join the network. It may happen too that the backend
won't find all the networks before the network enumeration
finishes.
Fix that by keeping a file monitor inspecting the network
uri, and update the places list when that happens.
https://bugzilla.gnome.org/show_bug.cgi?id=781162
Instead of the deprecated g_object_newv().
This requires some internal surgery to create our own vector of names
and values, but it does not functionally change anything.
GLib 2.53 deprecated g_object_newv() and GParameter. If we want to stop
using those types without resorting to pretty convoluted pre-processor
dancing, we will need to bump up the dependency inside GTK+.
GLib has deprecated GParameter and g_object_newv(); until we switch to
the new g_object_new_with_properties() API, and bump GLib required
version, we should simply ignore the compiler warnings.
The addition of GdkMonitor broke the quartz backend. This patch restores
that support by adding a new class GdkQuartzMonitor, and by modifying
the existing classes GdkQuartzDisplay and GdkQuartzScreen where
necessary.
It should be noted that this patch is essentially a refactor as no new
functionality that will impact the user has been added or removed.
https://bugzilla.gnome.org/show_bug.cgi?id=779184
Instead of using Ruby/Sass to generate the CSS from SCSS files, we can
use the faster and more lightweight libsass/sassc binary.
We can keep the CSS files in Git to make it easier to dist GTK+, but we
can add rules to ensure they get rebuilt if the source SCSS changes.
https://bugzilla.gnome.org/show_bug.cgi?id=780041
Instead of creating a GtkWindow, connecting to ::draw and drawing the
surface in there, then adding that window to another GtkWindow... just
use a GtkImage. This also gets rid of a bunch of utility functions used
only in gtk_drag_set_icon_surface.
When the blend modes were ported to use gsk defines, some
dashes were accidentally turned into underscores. It also
turns out that we were expecting 'saturate' instead of
'saturation' as per the css spec. Fix that as well.
If the widget isn't drawable anyway, just return;
If the widget needs an allocate, print a warning, since it indicates a
problem in the widget workflow (e.g. forgot to size_allocate a child
widget).
This maches the previous checks in gtk_widget_draw (with the same
problems).
When the GtkWidget hierarchy does not match the GdkWindow hierarchy, the
GtkWidget code may find a common ancestor that cannot be found while
traversing the GdkWindow tree using gdk_window_get_parent().
This happens with for example on Wayland, a GtkPopover has another
GtkPopover as parent, in this case, the GdkWindow parent is the root
window, whereas the GtkWidget parent is the other GtkPopover.
That confuses the gtk_widget_translate_coordinates() logic which will
bail out in this case and won't return the translated coordinates.
Make gdk_window_get_effective_parent() aware of subsurfaces and use the
transient_for which represents the actual parent (whereas the parent
might be pointing to the root window).
Bugzilla: https://bugzilla.gnome.org/show_bug.cgi?id=774148
For some reason, we are seeing damage being NULL here.
While that should never be the case, crashing on it is
unkind and makes the Wayland experience unusable.
The TextIter is passed by pointer for efficiency. We neither need to
modify it, nor should we leave it possible to accidentally do so. So,
it should be passed as a pointer-to-const.
We do not need to go through the heavyweight process of constructing a
TextLineDisplay just to get the direction out of it, when we can simply
use TextIter API to get the text and then get its direction using Pango.
Adapted from a patch by Mehdi Sadeghi for GtkSourceView:
https://bugzilla.gnome.org/show_bug.cgi?id=779081#c20
Add a documentation annotation saying that set_page_ranges transfers
ownership of the GtkPageRange array.
Add a g_free() call to fix a memory leak when set_page_ranges is
used repeatedly.
https://bugzilla.gnome.org/show_bug.cgi?id=780234
Since the later gtk_style_context_add_class doesn't care about the order
of the style classes, we can as well just prepend style classes to the
list and avoid the squared behavior when appending to a linked list.
Explain where the adjustment comes from, clarify some of the wording
about how its fields influence the scrollbar, and also note that the
steppers may not be present, since they aren’t in our default themes.
If the child added is not a Scrollable, it gets wrapped in a ViewPort –
which is. So it is impossible to end up with a non-Scrollable child.
Just check we have /any/ child where needed, which is semantically nicer
https://bugzilla.gnome.org/show_bug.cgi?id=778853
• intro: Clarify that external policy and/or adjustments can be used.
• add(): Don’t waffle on about having to add a ViewPort since we handle
that transparently for the user, so they can add() any widget.
• Adjustment stuff: most of this was repeating the docs for Scrollbar,
so just refer the user to that. Also, mention how
policies NEVER and EXTERNAL interact with all this.
https://bugzilla.gnome.org/show_bug.cgi?id=778853
has_tooltip_widget was assigned twice in immediate succession.
return_value is not used anywhere else in this function since commit
14a864c8b5 and does not need a default
value anymore, so move it to the inner scope and don't init to NULL.
hide_tooltip gets overriden in any case 2 lines down, and return_value
isn't used later in that function. The second assignment was introduced
in ef1da5f6c2, directly below the first
assignment.
shade/alpha/mix() take colour(s) and a number that is the ratio by which
to transform them. It was written here that these shall be passed in the
order (number, colour). That was wrong: they must be passed in the order
(colour[s], number) to work, and for the Inspector not to flag an error.
gtk_shader_builder_add_define should check both define_name and
define_value for not-NULL and not-empty, but the second precondition
check checks define_name again for not-empty-ness.
If you set GTK_INSPECTOR_RENDERER to the same type of
values that GSK_RENDERER takes this can change the renderer
used for the inspector. This is useful if you're debugging
one renderer and don't want to affect the inspector.
We can e.g. get the entry dispose()d and a focus_out event after that
(because the toplevel unsets the focus which previously was the entry).
We then later use priv->current_pos in a call to pango API which makes
sure the given index is valid for the given layout. Since we lazily
create a GtkEntryBuffer in get_buffer() and a PangoLayout lazily in
gtk_entry_create_layout, these 2 are always valid but don't match
priv->current_pos in this situation.
Fix this by resetting priv->current-pos in dispose().
The :label-widget is drawn before the child, so put the controls that
set the alignment of the :label-widget before those that pad the child.
We set (horizontal|vertical) padding, not "[xy]thickness". Also change
to "label [xy]align" & use grid spacing, not spaces at end of Labels.
This was ruined, with only 1 of the 8 subwindows rendering any content.
This commit fixes the responsible errors in the embedded GtkBuilder UIs:
• Fix broken replace by commit fb3d9022ad
of HBox with a Box having a broken orientation <property>
• Replace VBox and [HV]Paned with GtkOrientable successors (properly!)
• Remove use of Button:use_action_appearance, as this no longer exists
This commit also adds error reporting, in case other errors creep into
the GtkBuilder UI definitions, plus cleanup for the Builders and Windows
Since margin-left and margin-right are gone, we don't have to care
about the difference between them and start/end anymore and we can just
save start as left and end as right.
Instead of mentioning the old _get_preferred_xxx functions, mention
measure() and print the for_size value as well. The orientation is given
by printing either "width" for GTK_ORIENTATION_HORIZONTAL or "height"
for GTK_ORIENTATION_VERTICAL.
There are GtkGestureSingle subclasses that can be made to handle multiple
fingers (GtkGestureSingle is a subclass of GtkGesture, and not the
opposite, after all). And GtkGestureSwipe already tries to handle
GDK_TOUCHPAD_SWIPE events, except this event handler silently ignores
those.
Falling back to the GtkGesture generic handler which already
handles touchpad gesture events fixes this.
Make sure to clear up the number of keys being pressed on enter/leave so
that we don't end up with leftovers if a new window is mapped by a
keyboard shortcut.
https://bugzilla.gnome.org/show_bug.cgi?id=779374
The key repeat is stopped as soon as a key is pressed, so if the user
quickly presses a key while another is already pressed and being
repeated, key repeat gets cancelled:
- key1 press
- key1 repeat
- key2 press -> key1 repeat stopped
- key1 release
- key 2 is not repeated even though it's kept depressed
This is a different behavior from X11, which confuses migrating users.
To mimic the X11 behavior, keep track of the number of keys pressed
simultaneously and cancel key repeat only when none is pressed.
This way, if a user pressed a key while another one is being repeated,
the new key press can possibly be repeated as well.
Bugzilla: https://bugzilla.gnome.org/show_bug.cgi?id=778019
When resizing an xdg_popup immediately after the initial mapping, there
is a race condition between the client and the compositor which is
processing the initial size given by the xdg_positioner, leading to the
xdg_popup to be eventually of the wrong size.
Only way to make sure the size is correct in that case is to hide and
show the window again. Considering this occurs before the initial
configure is processed, it should not be noticeable.
https://bugzilla.gnome.org/show_bug.cgi?id=772505
This reverts commit 901e5ff3a3.
This causes criticals in e.g. the Text View: Multiple Buffers demo.
More work is required to get a fix for Bug 778853 that does not cause
anything else to regress.
https://bugzilla.gnome.org/show_bug.cgi?id=778853
The fact that it doesn’t reuse the existing GtkLabel if present is not
immediately obvious to users (or is it just me?), so clarify that the
pre-existing :label-widget, if any, is always removed and replaced.
It was only testing the default configuration where overlay-scrolling is
TRUE and the policy is POLICY_AUTOMATIC. We should also test FALSE and
POLICY_ALWAYS. This commit adds those tests and makes the !overlay &&
POLICY_ALWAYS case pass by excluding the size of the relevant scrollbar,
as we are only interested in whether the content size is as requested.
https://bugzilla.gnome.org/show_bug.cgi?id=778853
POLICY_AUTOMATIC means scrollbars are only shown when needed, i.e. when
the size of the window is not large enough to show the entire child. So
when measuring the preferred size, such scrollbars should be ignored.
But measure() was adding size for bars for which policy_may_be_visible()
was TRUE, which it returns for POLICY_ALWAYS (good) & _AUTOMATIC (bad).
So we reserved space for child plus scrollbars, & because we have enough
space for the child, POLICY_AUTOMATIC hides the scrollbar, leaving the
extra reserved space empty at the right/bottom sides of the child. This
is very noticeable/inconvenient for non-overlay, automatic scrollbars.
Fix this by only requesting size for scrollbars that use POLICY_ALWAYS,
rather than basing the decision on policy_may_be_visible().
https://bugzilla.gnome.org/show_bug.cgi?id=778853
Using Ctrl + left/right to skip between words, or left/right to cancel a
selection, were causing movement on the screen in the opposite direction
of the glyph on the key. This was surprising and awful UX for RTL users.
This is based on a patch covering the former case by:
Author: Mehdi Sadeghi <mehdi@mehdix.org>
Date: Sat Feb 18 02:16:00 2017 +0000
https://bugzilla.gnome.org/show_bug.cgi?id=136059
Using Ctrl + left/right to skip between words, or left/right to cancel a
selection, were causing movement on the screen in the opposite direction
of the glyph on the key. This was surprising and awful UX for RTL users.
This is based on a patch covering the former case by:
Author: Ori Avtalion <ori@avtalion.name>
Date: Tue Apr 20 08:06:23 2010 +0000
https://bugzilla.gnome.org/show_bug.cgi?id=136059
It was "Missing name of pseudo-class", but the real problem is exactly
the opposite: we /have/ been given a name, but it is not a valid one.
Change it to "Invalid name of pseudo-class" to minimise confusion.
gboolean ret for whether gtk_text_iter_backward_line() moved the iter
was declared but not used anywhere. I presume it was meant to be
checked, and it passes now, so let’s do it.
the scrollbar passed in better be either priv->hscrollbar or
priv->vscrollbar. Ensure that by using a simple else instead of an
else-if and a g_assert.
When a widget is created, its default scale is the scale of the
primary screen (for instance 2). But once parented to another widget
its scale factor should be the one of its parent (if parented to a
widget on a screen at scale factor 1, it should be 1).
The problem is that we don't emit the notify::scale-factor signal when
reparenting happens.
https://bugzilla.gnome.org/show_bug.cgi?id=776821
Otherwise we wait for the next gdk_drag_motion() call, which will
happen on the next motion event, making the drag window briefly visible
on the 0,0 root coordinates.
https://bugzilla.gnome.org/show_bug.cgi?id=778203
gtk_text_iter_backward_line() checks the value of
real->line_char_offset without previously calling
ensure_char_offsets (real) to make sure the former
is up-to-date.
As a consequence of this, when gtk_text_iter_backward_line()
is called after a gtk_text_buffer_insert_range() in the
first line of buffer, the iter is not moved to the start of
the line, and the return value is wrong.
Fixed by adding the ensure_char_offsets() call.
A test case for this bug is added to the textiter gtk testsuite.
priv->trigger_event is never set, so it is always NULL. This means the
gtk_menu_popup*() methods use the current event. The only way to get any
other event to combobox_menu_popup() was from the button-press-event
handler I just removed, which would end up being the current one anyway.
So, bin priv->trigger_event & explicitly pass NULL to gtk_menu_popup*().
gtk_show_uri_on_window() will pass enough information for Portal helpers
to allow dialogue parenting in Flatpak, gtk_show_uri() won't, so
deprecate it.
https://bugzilla.gnome.org/show_bug.cgi?id=778678
Update the autotools scripts to support Visual Studio 2017 builds by
copying the Visual Studio 2013 projects and updateing the items as
necessary to obtain the Visual Studio 2017 projects.
Note that the format of the toolset string changed, so allow one to
pass in and thus use a custom toolset string, otherwise the default
toolset string will be generated as it was before.
Note also the Visual Studio 2017 aims to be compatible with Visual
Studio 2015 on the CRT level, so binaries built with 2017 should
work without problems with the binaries built with 2015.
We can't pass the same string to two different snapshot states since
removing one of them will free the passed string, so just create another
one for the second state.
Clamping the anchor values as introduced in commit 9a5ffcd to fix bug
777176 breaks menu positioning.
By keeping the anchors rectangle size greater than zero, we end up
deducting some positive value from the original position, so there is no
need to clamp() actually, keeping the values positive is enough and
avoids the issue with menu positioning on the menubar.
An additional benefit is to make the code a lot simpler.
https://bugzilla.gnome.org/show_bug.cgi?id=778009
Since GtkTreeMenu became a private class only used by GtkComboBox, all
this test actually did was to show a ComboBox constructed with a custom
CellArea. Now that the latter is no longer possible, the test just shows
a handful of settings that do nothing. Just test GtkComboBox directly.
Currently hiding destroys the wl_surface and all related interfaces,
(including the gtk_surface1) so the next time the GdkWindow is mapped,
we don't bother to set the DBus properties. Toggle the check off so
it's actually issued again after the GdkWindow gets a gtk_surface1.
https://bugzilla.gnome.org/show_bug.cgi?id=773686
The new CSS border-spacing does what Grid::(row|column)_spacing and
Box::spacing already did, i.e. controlling the space added between child
widgets, so it’s not a replacement for Container::border-width.
Now that priv->area is guaranteed to be constructed by us, and not
passed in by a user, we can move it to the .ui file and stop manually
managing its lifetime altogether. And once the area is there, we can
move the menu there too (and stop pointlessly destroying/rebuilding it).
We already have cell_layout_is_sensitive() to get whether at least one
cell in a Layout is sensitive, which we need because CellLayout/View
do not implement foreach(). So, since we wrote that, we can use it to
check our CellArea too, instead of doing foreach with a custom callback.
It was looping over all items, not breaking out when it found the first
selectable one, and then selecting the _last_ selectable one (if any)
found. So, it did exactly the opposite of its name. This made me quite
baffled when opening a submenu with right-arrow put me at its last item.
Originally, the loop set to_select and broke if the current item was
selectable and not tear-off, meaning that it would correctly select the
first suitable item. However, when tear-off functionality was removed
in commit 4ed9452e90, so was the break.
combo_box_popdown() currently skips popping down our menu if it is NULL.
But the required call to this at end-of-life was in destroy(), by which
point dispose() already NULLed the menu, so Menu::popdown() would never
run, even if it should. Fix this by trying popdown() earlier in unmap().
Also, add a converse assurance that we don’t popup() while not mapped.
Even once we remove all the now-pointless NULL checks, destroy() was the
wrong place to call combo_box_popdown(), and unmap() is the right place.
gtk_init() removed its support for supporting arguments, so we ought to do
likewise for Windows, which actually defines items that call gtk_init()
the old way (and also get rid of argument support in those functions,
since the direction is to not support them).
https://bugzilla.gnome.org/show_bug.cgi?id=773299
Commit fdc0c6426b removed the appears-as-
list style property, & hence the ability to put the ComboBox into list
mode – but it left behind a pile of hijinks that were only used in said
mode & so were now doing absolutely nothing. This commit deletes those.
While doing that, I got carried away…so this also stops pointlessly type
checking popup_widget, as that can never be anything but a GtkTreeMenu.
It still checks for NULL everywhere, which shouldn’t be needed, but (A)
this commit is already too big, & (B) simply removing such checks where
they _seem_ unnecessary causes bad times. I’ll puzzle through that later
Commit fdc0c6426b for removing (partly!)
appears-as-list also deleted the code that propagated wrap-width to the
TreeMenu and thus put us into “grid mode”. This restores that code.
And as Benjamin noted, calling check_appearance() here is wrong, so bye.
We want to simplify our initialization code and remove all commandline
argument handling from it. The first stop for this is to reduce the
number of gtk_init variants we have.
This is how windows are meant to be hidden as per the wayland
protocol, there's no need to destroy the xdg_surface and other
interfaces.
Also, rename gdk_wayland_window_hide_surface() to clear_surface(),
as that's what it does.
https://bugzilla.gnome.org/show_bug.cgi?id=773686
i.e. when wrap-width > 0. This was only being done for non-grid cases.
So, ComboBoxes in grid mode did not indicate their selection when popped
up and required users to keynav from ‘nothing’ (at the top-left) to the
item they wanted to select. By selecting the active item in advance, now
it’s highlighted & acts as the starting point for keynav around the grid
This previously only mentioned its effect on the displayed value, and
even after the previous commit, its rounding of the actual value upon
change still reads like too much of an afterthought. Worse, it wasn’t
mentioned at all in the doc for the @digits parameter. Change this to
emphasise rounding always occurs and the displayed value is secondary.
Whether it should is an open question, but for now, the documentation
should clearly indicate that currently rounding is only applied upon
changes to the value, not to the existing value when ::digits changes.
This is already clear in the doc for the underlying Range::round-digits.
https://bugzilla.gnome.org/show_bug.cgi?id=358970
The documents state that gtk_scale_set_digits() “causes the value of the
adjustment to be rounded off to this number of digits, so the retrieved
value matches the value the user saw.” Note the lack of any condition.
But in fact, if draw-value was false, rounding was disabled on the base
Range, so values that weren’t displayed weren’t rounded. This made the
docs wrong and made an apparently cosmetic detail alter functionality.
Fix by ensuring the number of digits set on Scale is always propagated
along to gtk_range_set_round_digits(), thus rounding to it in all cases
when the value changes, regardless of whether the value is displayed.
This doesn’t address the other idea from Bugzilla: that changing the
number of digits should clamp the _existing_ value if it’s more precise.
This contradicts digits docs in the base Range, but the above from Scale
can be read as implying it’ll happen. For now, that’s an open question.
https://bugzilla.gnome.org/show_bug.cgi?id=358970
GtkFileChooserButton installs a handler for the popped-up signal, which
refilters the menu, in order to hide the “(None)” item from the popup
if it was previously selected in the ComboBox. This oddity means that:
• Until recently, this item would be selected in the menu shell, which
would then be popped up and change the selection away from that item.
This was therefore redundant (more on which below!) but benign.
• After the patch for https://bugzilla.gnome.org/show_bug.cgi?id=771242
however, this causes a critical assertion fail, as now we stash the
originally selected item in a pointer so that it can be selected only
after realisation/popup – but by that stage, the model has just been
refiltered and the previous pointer no longer refers to a valid item.
This commit works around this problem by, after popping up the menu,
getting the active item again, in case a popped-up handler has gone and
invalidated the pointer to the active item that we saved before popup.
If a handler does this, everything done to find/use the original item is
pointless. But this avoids the ugly critical in FileChooserButton, while
not harming every other ComboBox that doesn’t mess with its model while
popping up (hopefully the vast majority), and it’s very difficult to
imagine a way to check if the active item is /going to/ be hidden later)
Previously, for compatibility with GTK 3.0, we allowed specifying
numbers without units and interpreted them as pixels, even when the CSS
specification didn't.
Remove that now that we can break API.
This reverts commit 4875c689a0.
This was a thinko. Writable is not actually settable from the
application side, but only for the user, from the backend side.
Elsewhere we already go through the keymap to get modifiers so we
should do the same here. In fact, this was relying on xkb modifier
mask values being bitwise compatible with GdkModifierType which isn't
necessarily true.
https://bugzilla.gnome.org/show_bug.cgi?id=770112
Gtk+ treats MOD1 as a synonym for Alt, and does not expect it to be
mapped around, so we should avoid adding GDK_META_MASK if MOD1 is
already included to avoid confusing gtk+ and applications that rely on
that behavior.
https://bugzilla.gnome.org/show_bug.cgi?id=770112
When a subsurface is used as a parent of a popup, GDK needs to traverse
up to the transient-for as the next parent, to properly find the parent
used by the popup positioner. This is because the parent of a popup
must always either be an xdg_popup or an xdg_surface, but traversing
the "parent" (in GDK terms) upwards from a subsurface will end up on
the fake root window before we hit the actual parent (in Wayland terms).
https://bugzilla.gnome.org/show_bug.cgi?id=776225
Instead of having 3 different shaders for the different clipping
versions, just have one shader and use a preprocessor define to use
different clip functions.
That preprocessor define is set in the Makefile.
Also use foo.frag and foo.vert as the file extensions instead of using
foo.frag.glsl and foo.vert.glsl, as that's what glslc suggests as
extension.
That way we don't need to move the clip rounded rect manually through
the vertex shader into the fragment shader but can just look at the push
constants.
Simplifies shaders a lot.
Passing a rectangle with zero width or height to xdg_shell-v6
set_anchor_rect() will cause a protocol error and terminate the client,
as with gedit when pressing the Win key.
Reason for this is because the rectangle used to set the anchor comes
from gtk_text_layout_get_iter_location() which uses the pango layout
width/height, which can be empty if there is not character at the given
location.
Make sure we don't use 0 as width or height as an anchor rectangle to
avoid the protocol error, and compensate the logical position of the
given rectangle if the size is changed, so that the actual position
remains as expected by the client.
https://bugzilla.gnome.org/show_bug.cgi?id=777176
Images with just an aspect ratio, but without a size, should be scaled
to be fully visible in the given area.
But we scaled them to completely cover the given area, which made them
partially invisible.
Reftest included.
gtk_snapshot_pop() => removed
gtk_snapshot_pop_and_append() => gtk_snapshot_pop()
So now there is no way to get a rendernode out of the snapshotting API
until you gtk_snapshot_finish().
... and use it.
The function is a bit awkward because it requires 2 calls to
gtk_snapshot_pop(), but once you accept that, it's very convenient to
use, as can be seen by the 2 implementations.
This is a free-form tab that can contain information about the
system environment. To see it, set GtkAboutDialog::system-information
to a non-NULL value.
https://bugzilla.gnome.org/show_bug.cgi?id=776604
The Mesa Vulkan drivers need XInitThreads() being called, because their
implementation has to use threads.
And I don't want to make the call depend on if Vulkan is compiled in
because that makes GTK's X11 behavior depend on compile-time flags, so
it's always called.
This way, we ensure that files that are built during make always get
properly listed. And we ensure that creating the resources actually
depends on them.
We already take ints when setting the translation, so it can't
currently take any other values. Additionally, I was seeing large
costs in int -> double -> int for the rects in
gtk_snapshot_clips_rect(), as all callers really are ints (widget
allocations) and the clip region is int-based.
This change completely cleared a 2% rectangle_init_from_graphene from
the profile and is likely to have nice performance effects elsewhere
too.
The width/height/aspect getters are called a lot, and almost all
callers already verify it from _gtk_css_image_get_concrete_size (),
so just skip these checks.
This means we allocate the collect data with the state, avoiding
an extra allocation. Also, a union means every state object
is the same size and we could reuse the state objects.
This was showing up quite high on the profiles, and there is
no real reason for copy to normalize, as the source is a
GskRoundedRect which should be normalized already unless
you did something very strange (and then you should have normalized
manually).
Simgle image cross-fade opacity was computed the wrong way, which caused
weird fade-in/out animations, for example in flat buttons.
I messed this up when porting cross-fades to snapshot().
Since the demise of theme engines, we can no longer hit
the case of id >= GTK_CSS_PROPERTY_N_PROPERTIES. So don't
check for this in a very frequently called function.
Using an image() fallback from svg to png doesn't make too
much sense, since the svg is always used (unless librsvg is
not present), while the png icon is faster and cheaper to
load and thus preferable.
Also, "ie" wasn't very clear, but fixing that to "i.e." would cause
truncation of the summary when processed by bindings using doxygen. So,
I replaced it with "in other words", which is no _less_ clear, at least.
It was suggested that the project files to be moved to win32/, so that we can
have one less layer of directories we need to go down into to reach the project files.
Tell people about what happens when generating projects when Visual
Studio 2013 or later is required, and mention that the .headers are
only needed when headers need to be copied.
So we can set the css name of a widget to something that's not related
to the class name. If the css-name property is set to NULL, we will
still fall back to the one set using gtk_widget_class_set_css_name which
is alwasys non-NULL since GtkWidget itself sets it to "widget".
Instead of relying on --generate-dependencies and the resource file,
actually list the resources in Make variables.
Fixes make not building new shaders because they're not inside the
resource file.
See the implementation of gtk_entry_create_layout():
pango_attr_list_splice() is used to add the PangoAttrList of the preedit
string. And that is done *after* applying the PangoAttrList of the
"attributes" property.
https://bugzilla.gnome.org/show_bug.cgi?id=776868
We now have GTK_CSS_AFFECTS_CONTENT for properties that have an effect
on content rendering.
Using GTK_CSS_AFFECTS_ICON is wrong for icon-transform and icon-filter
as they don't change the icon, just how the icon is rendered, so we use
GTK_CSS_AFFECTS_CONTENT for those.
We also introduce GTK_CSS_AFFECTS_POSTEFFECT for opacity and filter -
properties that affect the whole drawing of the widget by applying an
effect after everything is said and done.
...which treats the first '.' in doc comments as the end of the summary.
So, e.g., in gtkmm, get_kinetic_scrolling() is currently summarised as
"Changes the behaviour of @scrolled_window wrt." Not very informative!
No need for a period there & anyway, the phrase "wrt to" is superfluous,
and we have space to actually say "with regard to", so just do that now.
Instead of
-gtk-icon-effect: dim;
-gtk-icon-effect: hilight;
we now use
-gtk-icon-filter: opacity(0.5);
-gtk-icon-filter: brightness(1.2);
respectively.
This node essentially implements the feColorMatrix SVG filter. I got the
idea yesterday after looking at the opacity implementation.
It can be used for opacity (not sure if we want to) and to implement a
bunch of the CSS filters.
Since the status of the GDK broadway backend is more or less unsupported,
drop the projects that build gtk4-broadwayd and gdk-broadway, and update
the projects to not to refer to them.
However, keep the Broadway configs for now as we will later transform
them to become configs for Vulkan, so bascially besides "installation"
parts and output settings, they will do the same as their Release|Debug
counterparts with no support for Broadway.
...but disable them for now. Configs will be added for the projects to
support Vulkan-enabled builds which will then enable the builds of these
sources. Extra commands and items will be needed for the GSK resources
along with ensuring GSK_RENDERER_GSK being defined for the build of GDK,
GDK-Win32 and GSK so that the builds of Vulkan-enabled builds can be done
properly.
Filter out the Vulkan sources from the 'dist hook' rules in
gsk/Makefile.am as we don't want to in turn include them twice in the
projects when the 'make dist' is performed on a system with Vulkan
builds enabled.
If the signal handler ends up changing the label text,
the link is no longer around to update the css node.
Check for this possibility to avoid a crash here.
One cannot use #if...#endif within macro calls in Visual Studio and
possibly other compilers, and there are more uses of VLAs that need to be
replaced with g_newa().
There were also checks for the clip type in gskvulkanrenderpass.c which
were possibly not done right (using the address of the type value to check
for a type value), which triggered errors as one is attempting to compare
a pointer type to an enum/int type.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
This adds support to the GDK Win32 backend so that we can support Vulkan
context creation for use in the GSK Vulkan renderer, so that we can test
it on Windows platforms as well.
https://bugzilla.gnome.org/show_bug.cgi?id=776544
Use g_newa() instead of VLAs, as VLAs may never be supported by some
compilers as it became optional in C11 and there are concerns about their
implementations in compilers that do support it.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
Forces a full redraw every frame.
This is done generically, so it's supported on every renderer.
For widget-factory first page (with the spinner spinning and progressbar
pulsing), I get these numbers per frame:
action clipped full redraw
snapshot 0ms 7-10ms
cairo rendering 0ms 10-15ms
Vulkan rendering 3-5ms 18-20ms
Vulkan expected * 0ms 1-2ms
GL rendering unsupported 55-62ms
* expected means disabling rendering of unsupported render nodes,
instead of doing fallback drawing. So it overestimates the performance,
because borders and box-shadows are disabled.
It's faster to render once for every rectangle in the clip region than
rendering the outline of the clip region.
Especially because this reduces the time necessary to build up the frame
data.
In widget-factory (where we have 3 rectangles), this leads to a 5x
speedup in the rendering time rendering alone.
Snapshotting time goes from 10ms to ~1ms, which is another huge
improvement.
Note: We interpolate premultiplied colors as per the CSS spec. This i
different from Cairo, which interpolates unpremultiplied.
So in testcases with translucent gradients, it's actually Cairo that is
wrong.
Homogeneous branches repeated the calculation/assignment of the initial
space available to children. This avoids that by shuffling some code.
Perhaps more importantly, in doing that, I ended up with some ambiguous
names, and Company and I realised how vague the pre-existing naming was.
"size" becomes "extra_space", as this is what it represents. Conversely,
"extra" becomes "size_given_to_child" (albeit still given out in two
different ways depending on whether the Box is homogeneous). My hope is
that these sections of code are now somewhat less baffling than before!
This is now tracking the clips added by the clip nodes.
If any particular node can't deal with a clip, it falls back to Cairo
rendering. But if it can, it will render it directly.
Now that every call to GtkCellArea is a snapshot call and no more cairo
calls are left, move the actual differentiation between Cairo and
Snapshot down to the cell renderer.
Little tool that creates a bunch of test files to throw add the
rendernode binary.
They should really be part of a testsuite, but we have none, so OI just
put them here.
Only keep the version that calls gsk_render_node_draw() if people
specify the --fallback option.
The actual renderer selection works just as for regular GTK. The easiest
way to influence it is setting the GSK_RENDERER environment variable.
... and implement it for the Cairo renderer.
It's an API that instructs a renderer to render to a texture.
So far this is mostly meant to be used for testing, but I could imagine
it being useful for rendering DND icons.
That code doesn't do anything.
And what the code should be doing (clearing the abckground) isn't
necessary as cairo drawing is guaranteed to clear the surface.
This does a conversion to/from GBytes and is intended for writing tests.
It's really crude but it works.
And that probably means Alex will (ab)use it for broadway.
I had originally thought I'd use GskShadow for box-shadow, but didn't in
the end.
So now it's only used for text-shadow and icon-shadow, and those don't
have a spread.
VLAs are not supported by Visual Studio and possibly other compilers that
are supported by GTK+-3.90+, and probably never will be, although it is a
C99 specification, and it became optional for C11. It is also not a part
of the newer compiler requirements that are listed out for GTK+-3.90.x.
There exist concerns about the implementation of VLAs in compilers that
support them as well, so change it to a g_newa() approach.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
Instead of a separate allocation for any arrays in the render node
we allocate these as part of the render node itself, using C99
flexible arrays.
This leads to less allocations, which is nice, but the major reason
for this is that it allows us to change the allocation scheme further
in the future. For instance, we want to do stack-like allocation so
that all the render-nodes for an entire frame are allocated in one
(or a few) chunks.
Instead of constantly recalculating this (especially recursively for
parents!) we do it only on construction, because everything is
immutable anyway. Also, most nodes had a bounds already and can
use the new parent member instead.
We also do direct access to the node bounds rather than calling
gsk_render_node_get_bounds in various places, which means
we do less copying.
This causes the snapshotting algorithm to dump all widget nodes into
their own container node. We then name that group accordingly (ie
"GtkSwitch<0xdeadbeef>") so you can easily see which node belongs where.
The feature is toggleable in the inspector's visual tab.
There's a few problems with it, becuse GtkSnapshot optimized container
nodes away if they are not needed, so we are losing some widgets...
When the first/last color stop is not at 0%/100%, we need to start the
repeating at their offsets and not at 0%/100%.
Attached reftest demonstrates the problem.
Instead of making people intiialize a rectangle and then applying border
radius manually, provide a constructor that does it for them.
While doing that, also allow people to instead request the padding box
or the content box.
Refactor all relevant code to use this new constructor.
... and make the icon rendering code use it.
This requires moving even more shadow renering code into GSK, but so be
it. At least the "shadows not implemented" warning is now gone!
The node draws a solid CSS border, which can be used to cover everything
but dashed and dotted borders (double, groove, inset, ...).
For different border styles, we overlay multiple nodes and set their
colors to transparent for sides with non-matching styles.
This way we can pass the command pool around.
And that allows us to allocate and submitcustom buffers.
And that is necessary to make staging images work.
It turns out, some simple getters - such as
gdk_drawing_context_get_clip() - love copying things before returning
them.
I guess somebody has to burn cycles...
When the background-clip of the background is smaller than the
background-clip of blended images, not pushing a group is wrong.
Test testing exactly that included.
This uses the new push()/pop() mechanism to its fullest extent when
implementing transitions. It's fun to inspect the results in the
inspector.
Crossfades don't work yet, they continue using a Cairo fallback.
A side effect of the stack conversion is that widget-factory now uses
snapshots for a lot more things.
It is now possible to call push() subfunctions for simple container
nodes with just a single child. So you can for example
gtk_snapshot_push_clip() a clip region that all the nodes that get
appended later will then obey.
gtk_snapshot_pop() will then not return a container node, but a clip
node containing the container node (and similar for the transform
example).
This is implemented internally by providing a "collect function" when
pushing that is called when popping to collects all the accumulated
nodes and combine them into the single node that gets returned.
To simplify things even more, gtk_snapshot_pop_and_append() has been
added, which pops the currently pushed node and appends it to the
parent.
The icon rendering code has been converted to this approach.
This code makes renderers fall back to Cairo rendering if they don't
know how to handle a render node's type.
This allows adding new render nodes with impunity.
Instead of appending a container node and adding the nodes to it as they
come in, we now collect the nodes until gtk_snapshot_pop() is called and
then hand them out in a container node.
The caller of gtk_snapshot_push() is then responsible for doing whatever
he wants with the created node.
Another addigion is the keep_coordinates flag to gtk_snapshot_push()
which allows callers to keep the current offset and clip region or
discard it. Discarding is useful when doing transforms, keeping it is
useful when inserting effect nodes (like the ones I'm about to add).
Instead of having a setter for the transform, have a GskTransformNode.
Most of the oprations that GTK does do not require a transform, so it
doesn't make sense to have it as a primary attribute.
Also, changing the transform requires updating the uniforms of the GL
renderer, so we're happy if we can avoid that.
I'm about to move children handling to the container node, which means
the generic code can no longer assume children APIs existing.
So rewrite the treemodel to work without it.
gsk_render_node_get_bounds() still exists and is computed via vfunc
call:
- containers dynamically compute the bounds from their children
- surface and texture nodes get bounds passed on construction
In the brave new world of refactored render nodes, this function doesn't
really make any sense anymore. We could turn it into a vfunc, but I
don't think it's useful.
Especially because even in the brave old world, this function was
causing a vastl overallocation of nodes when the GL renderer needed render
targets.
If we ever feel, we need this function again, we can readd it later.
But nobody is using it other than for overriding opactiy. And you can
just override opacity directly if you care.
Creating render nodes is fire-and-forget, so all one should do is create
a container, append, append, append and then send it off to the
renderer. So there's no need to replace, insert between or anything
else.
We want to split nodes into containers and nodes that do actual drawing.
So pushing nodes that do drawing is exactly the wrong thing.
Also fix up GtkPopover. There's no need for it to push anything.
When we generate the Visual Studio 2013 projects, we need to remove the
*.vs12.sourcefiles and *.vs12.sourcefile.filters that are generated during
the process, so that 'make distcheck' won't complain about leftover files.
When running uninstalled tests with GtkApplication on an autobuilder with
a fake session bus, warnings will cause the tests to abort. The GNOME
session manager, the Xfce session manager, and the Inhibit portal are all
not needed for normal operation of GTK, so we should not log warnings if
they are not found.
As well as not being present on a fake session bus, it's also not
expected that they'll be present on all platforms.
https://bugzilla.gnome.org/show_bug.cgi?id=774784
... with gtk_list_box_get_row_at_y. It would be nice to avoid the
'find' versus 'get' discrepancy since we are planning to expose it as
public API.
https://bugzilla.gnome.org/show_bug.cgi?id=776187
Fix the build after the branch wip/alexl/simplify-gdkwindow was merged, as
there are some changes that broke things in the Windows backend, namely:
-gdk_win32_input_shape_combine_region() should not be removed at this
point (though it is a stub--otherwise GDK/Win32 will crash)
-Some more code need to be removed due to the removal of items in the
above-mentioned merged branch
Also, like the X11 backend, do not allow the creation of native child
windows, and stop checking for subsequent child windows
(GDK_WINDOW_CHILD), so that we can clean things up a bit.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
It does weird clipping that
(a) nobody likes
(b) is hard to support in the new rendering world.
So we take the easy way out.
The actual frame is now drawn by the frame node around the label.
GtkCellView has a gadget, so peopl can do all their shenanigans with
CSS.
And the original use case (overriding the background so that the
cellview's GdkWindow shares the background color of the combobox) is
outdated since we have transparent backgrounds.
We're not currently using this, and dropping it allows us to loose
a bunch of code which leads us towards the goal of having GdkWindow
only for toplevels (and reparenting makes not sense for toplevels).
We can't really support these on e.g. wayland anyway, and we're trying
to get rid of subwindow at totally in the long term, so lets drop this.
It allows us to drop a lot of complexity.
For subsurfaces, the new state which includes the input shape is not
applied by the compositor if the subsurface is in effective synchronous
mode.
So we need to apply the input shape once parent surface is in effective
desynchronized mode, which is when it's committed, otherwise the input
shape may never be applied if the widget is not using being_paint() /
end_paint() to draw on its subsurface, like clutter does.
We do that only for empty input shape as those won't need update when
the subsurface is resized, for all other non-empty input shape, the
client still has to use begin_paint()/end_paint() for the input shape to
be applied.
https://bugzilla.gnome.org/show_bug.cgi?id=774534
It's using a GtkCssPositionValue, even though that name is wrong. But
the functionality of managing 2 lengths is exactly what we want.
Nobody is using this yet.
- Recognize "gl" as well as "opengl" for the GL renderer
- GSK_RENDERER=help now works
- g_warning() for an unrecognized renderer (typo detection!)
- g_print() the actual renderer that is used (and error messages when
selecting) when a GSK_RENDERER is given, so you'll notice if your
renderer isn't taken.
Previously, code would work fine with --disable-vulkan if the Vulkan
headers were installed - code would happily just use them as they're
installed in /usr/include.
By creating unlimited render objects, we would never wait on the GPU.
This would mean that if the GPU was the bottleneck, we would fill its
queue with render commands faster than it could process them.
And because the nvidia binary driver and my code work surprisingly well
and bugfree, this lead to exhaustion of RAM. I had 50GB of swap
configured and my hard disk was quicker as swap storage than my GPU was
at processing the commands, so stuff still filled up.
At that point my computer became rather unresponsive and I decided to
reboot it, so I that could write this patch.
Add SURFACE and TEXTURE operations. This way, we actually render more
than one node every frame because not everything is a fallback node
anymore that gets composited with its children into a cairo surface.
Instead of pushing the root matrix, push the world matrix for the
current node. That way, the bounds we emit as vertices are actually
properly transformed.
First, we collect all the info about descriptor sets into a hash table,
then we use its size to determine the amount of sets and allocate those
before we finally go ahead and use the hash table's contents to
initialize the descriptor sets.
And then we're ready to render.
We can let the GPU do its stuff without waiting. The GPU knows what it's
doing.
Which means we now get a lot of time to spend on doing CPU things (read:
we're way better in benchmarks).
The old behavior is safer, so we want to keep it around for debugging.
It can be reenabled with GSK_RENDERING_MODE=sync.
And move the actual rendering code there.
A RenderPass is a collection of operations on the same target that
get executed one after another. It roughly targets VkRenderPass or
rather the subpasses of a VkRenderPass.
For now, only the infrastructure is there. No real stuff is happening.
This is refactoring work.
GskVulkanRender is supposed to be the global object for a render
operation, ie GskVulkanRenderer.render() will create this object for
what it does.
The object will be split into stages that perform the operations
necessary to create a drawing.
Instead of using a staging iamge, we require the final image to be
linearly allocated and have host-visible memory.
This improves performance quite a bit.
The old code is still there and can be enabled with a simple change
to a #define in gskvulkanimage.h
Instead, complain if somebody calls gdk_x11_window_get_xid() on a
non-native window.
We cannot make random windows native anymore because there's no GSK
renderer associated with them, so we cannot draw them.
We can now upload vertices.
And we use this to draw a yellow background. Which is clearly superior
to not drawing anything.
Also, we have shaders now. If you modify them, you need glslc installed
so they can be recompiled into Spir-V bytecode.
1. Output Vulkan status in summary
2. Add missing "test" call
3. Check for glslc
The glslc check will be necessary later for the code that automatically
compiles the Vulkan glsl source to Spir-V.
Nothing happens if glslc is not available - unless you modify the glsl.
gdk_window_create_vulkan_context() now exists and will return a Vulkan
context for the given window. It even initializes the surface. But it
doesn't do anything useful yet.
Adds the gdk_display_ref_vulkan() and gdk_display_unref_vulkan()
functions which setup/tear down VUlkan support for the display.
Nothing is using those functions yet.
These only exist for the window dragging which does not exist anymore
currently. It will be reintroduced later in a form that does not require
these handlers.
I read the code as if (use_gl) instead of if (!use_gl) and commented it
out in bddfd7bb41. That broke drawing on
Wayland without OpenGL completely.
Whoops.
Now it's back.
There were some parts that need some updates after the refactoring in
GDKGL, so that the code will continue to build and run.
For gdkwindow-win32.c, comment out the parts where we check for use_gl
(which was removed), since we are going to move all drawing to OpenGL,
but don't remove/disable the whole portion as that transition is not
complete at this point.
There a is new GDKGL function that checks for the damaged area of the back
buffer, but since the notion of "damage" is for *NIX (GLX/EGL for
Wayland/mir), meaning that there is no such extension for Windows in this
regard, so we can't support this on Windows as-is, at least for now.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
This is a way to query the damaged area of the backbuffer.
The GL renderer uses this to compute the extents of that damage region
(computed via buffer age) and use them to minimize the area to redraw.
This changes the semantics of GL rendering to "When calling
gdk_window_begin_frame() with a GL context, the area by
gdk_gl_context_get_damage() needs to be redrawn and every other pixel of
the backbuffer is guaranteed to be correct.
After gdk_window_end_frame() on a GL-drawn window, the whole backbuffer
must be correct.
We can always glXBufferSwap() now because of this.
... instead of a gl context.
This requires some refactoring in the way we mark the shared context as
drawing: We now call begin_frame/end_frame() on it and ignore the call
on the main context.
Unfortunately we need to do this check in all vfuncs, which sucks. But I
haven't found a better way.
That way we can capture both the actual changes (clip region) and the
area that was redrawn (render region), which in OpenGL might not be
identical.
Nothing shows the render region yet though...
Reenable GL drawing, but do it without Cairo.
Now, the context passed to gdk_window_begin_draw_frame() decides how
drawing is going to happen. If it is NULL, Cairo is used like before.
If a context is passed, Cairo may not be used for drawing and
gdk_drawing_context_get_cairo_context() is going to return NULL.
Instead, the GL renderer must draw to the GL backbuffer and
end_draw_frame() is then swapping that to the front.
The GskGLRenderer has lost the texture it used to render to and adapted
to render directly to the backbuffer instead.
The only thing missing is for GtkGLArea to gain back a performant way to
render. But it didn't have one since the introduction of GSK, this
patchset doesn't change anything about it.
The new rendering avoids two indirections (the GSK renderer's texture
and the GDK double buffering surface).
It improves icon count in the fishbowl demo by 30%.
This way, we can query the GL context's state via
gdk_gl_context_is_drawing().
Use this function to make GL contexts as attached and grant them access
to the front/backbuffer for rendering.
All of this is still unused because GL drawing is still disabled.
No visible changes as GL rendering is disabled at the moment.
What was done:
1. Move window->invalidate_for_new_frame to glcontext->begin_frame
This moves the code to where it is used (the GLContext) and prepares it
for being called where it is used when actually beginning to draw the
frame.
2. Get rid of buffer-age usage
We want to let the application render directly to the backbuffer.
Because of that, we cannot make any assumptions about the contents the
application renders outside the clip area.
In particular GskGLRenderer renders random stuff there but not actual
contents.
3. Pass the actual GL context
Previously, we passed the shared context to end_frame, now we pass the
actual GL context that the application uses for rendering. This is so
that the vfuncs could prepare the actual contexts for rendering (they
don't currently).
4. Simplify the code
The previous code set up the final drawing method in begin_frame.
Instead, we now just ensure the clip area is something we can render
and decide on the actual method in end_frame.
This is both more robust (we can change the clip area in between if we
want to) and less code.
This is a temporary switch-off of the GL dawing code that will make
things keep running. All GL related code (like the GSK renderer or
GtkGLArea will now fall back to software.
Wayland subsurfaces can have other native window parents, but those need
to be destroyed along with the rest of the window hierarchy otherwise
an assert() is reached.
https://bugzilla.gnome.org/show_bug.cgi?id=774915
GtkListBox is not a windowed widget anymore so we can't use
gtk_widget_get_window. Just directly access priv->view_window instead to
get the right window.
For a menu mode CB with wrap_width == 0 and an active item, that item is
selected in gtk_combo_box_menu_popup. Selection causes the MenuShell to
activate and hence take a grab. This was done before the menu was popped
up. A patch distributed in Debian sid - after being proposed on our BZ -
revealed that on the 1st popup of any such ComboBox, within grab_add,
the MenuShell's toplevel's GdkWindow is NULL. This causes a Gdk-CRITICAL
assertion fail on the 1st time opening any such CB, on Debian and if
that patch were merged to GTK+. By selecting after popup, we ensure the
MenuShell is realised before its grab_add and so avoid the critical.
https://bugzilla.gnome.org/show_bug.cgi?id=771242
This way, we don't spam criticals when GL is not available. Instead, we
print a useful debug message to stderr and continue with the Cairo renderer.
Signed-off-by: Emmanuele Bassi <ebassi@gnome.org>
and remove gsk_renderer_get_for_display().
This new function returns a realized renderer. Because of that, GSK can
catch failures to realize, destroy the renderer and try another one.
Or in short: I can finally use GTK on Weston with the nvidia binary
drivers again.
Signed-off-by: Emmanuele Bassi <ebassi@gnome.org>
Instead of having a gsk_renderer_set_window() call, pass the window to
realize(). This way, the realization can fail with the wrong window.
Signed-off-by: Emmanuele Bassi <ebassi@gnome.org>
gdk_window_get_toplevel() walks up the windows tree looking for the
corresponding toplevel window, but needs to account for subsurfaces as
well on Wayland.
https://bugzilla.gnome.org/show_bug.cgi?id=775319
We were producing org.symbolic.png from org.gnome.Recipes-symbolic.svg,
which is not useful. Look for the last dot in the original name, to
produce the expected org.gnome.Recipes-symbolic.symbolic.png instead.
gskrenderer.c includes gdk/wayland/gdkwayland.h and as a consequence
we need to be able to locate wayland's headers in case they are not
in standard location.
https://bugzilla.gnome.org/show_bug.cgi?id=775038
Now that subsurfaces can be created as child of another GdkWindow (and
not just the root window), they must be placed according to the location
of their parent, i.e. the abs_x/abs_y must be updated and taken int
account when placing and moving subsurfaces under Wayland.
https://bugzilla.gnome.org/show_bug.cgi?id=774917
Since at-spi-atk commit 96621a5e95 fixed PropertyChange notifications
for AccessibleParent, setting the parent will result in a call to
ref_state_set() which assumes that the object is fully initialized.
https://bugzilla.gnome.org/show_bug.cgi?id=774939
While GtkEventController implementations today are all GtkGesture, it is
possible to create a GtkEventController manually. This is an extrac check
to ensure we only add gestures to the list.
https://bugzilla.gnome.org/show_bug.cgi?id=774760
We need to unrealize the children manually for that to happen, but so it
goes.
The order is necessary because we want the renderer to still be alive
while children are unrealizing.
Only attempt to initialize Wintab after the display manager announces
that the first default display has been set. Fixes a segfault during
initialization of specific tablet drivers' wintab32.dlls. Add assertions
and verbose comments explaining this nonsense because this stuff is a
pain to have to keep fixing.
https://bugzilla.gnome.org/show_bug.cgi?id=774379
Move the orientation sanity-checks into the packet decode func.
Rationale: the packet handling func may otherwise read beyond the end of
device->last_axis_data.
Also expand them to cope with my test Huion's weird reporting.
Also correct the azimuth angle to align with GDK's presentation.
Most importantly, fix annoying comment typo.
https://bugzilla.gnome.org/show_bug.cgi?id=774265
We no longer have GtkPlug nor GtkWin32EmbedManifest for GTK+-4.x, and it
is not entirely clear at this point what would be the "best" replacement
for them, but this issue here prevents GTK+-3.89.x building on Windows.
As a result, this is a fast port to avoid using APIs that have been
removed for 4.x, and things seem to work properly (the print.c page
printed).
https://bugzilla.gnome.org/show_bug.cgi?id=773299
Fix a regression introduced in 4ce6d10601
which causes devices with an odd-numbered zero-based index in the list
to be passed over incorrectly. This might present as yet another "device
does not send pressure" bug for ~50% of devices out there.
This commit also closes off another potential segfault for wintab_devices
lists which have an odd length.
https://bugzilla.gnome.org/show_bug.cgi?id=774699
When checking if a rectangle is contained by the rounded box, the code
will refuse a rectangle which is the exact size as the one backing the
rounded box, since it checks for greater or equal width and height.
Check for greater only instead.
https://bugzilla.gnome.org/show_bug.cgi?id=774114
The functions gdk_pixbuf_get_from_window() and
gdk_cairo_set_source_window() are unreliable and depend on the windowing
system (they work great on X11 and Win32, less so on Quartz and Wayland).
With the switch to new drawing API and OpenGL, we can definitely no
longer support a generic way to snapshot windows.
People should either write windowsystem-specific code or draw their
widgets directly - like with gtk_widget_draw() - if they need to get a
rendering.
The tests read a nonexisting colorprofile, try to convert stuff read
from the window into it, do things that gdk-pixbuf should test and
then aren't even integrated into the testuite.
Sheesh.
- Make the rows larger
- Display the elapsed time between renderings
- Display if it was a full or a partial redraw
- Add a toggle button to display profiler info
Empty doc comments make gtk-doc complain about undocumented
functions, even though these functions are not supposed to
be documented in the first place.
Under Wayland, a subsurface can have another surface as parent, but
gdk would not allow native windows if the parent is not the root window.
Allow native subsurface for all parent under Wayland, not just for the
root window.
https://bugzilla.gnome.org/show_bug.cgi?id=774475
This can be triggered on workspace switches, and on hidpi results in
the scale factor being reset to 1 while the window is not in the
current workspace.
https://bugzilla.gnome.org/show_bug.cgi?id=774476
Just to avoid having to do NULL checks when calling
widget_class->snapshot. We were crashing with drawing areas who don't
have a draw or a snapshot vfunc (woot!).
We need so subtract the allocation from the clip to get the clip offset,
not the other way around.
This was screwing in particular with marks on GtkScale, because GtkScale
mark clip computation is broken and always returns (0,0) which makes
scales have a waaaaay too large clip.
But that's another bug.
And use this to cull widgets and gadgets that are completely outside the
clip region.
A potential optimization is to apply this clip region to cairo contexts
created with gtk_snapshot_append_cairo_node(), but for that we'd need to
apply the inverse matrix to the clip region, and that causes rounding
errors.
Plus, I hope that cairo drawing becomes exceedingly rare so it won't be
used for the whole widget factory like today (which might also explain
why no culling happens in the widget factory outside the header bar.
The equivalent to cairo_matrix_multiply (a, b, c) is
graphene_matrix_multiply (c, b, a).
graphene_matrix_multiply (a, b, c) may not be called with b and c being
the same matrix.
Grips have long been unused in GTK, so remove all support for them.
This removes the GTK_STYLE_CLASS_GRIP and the special
gtk_render_handle() code for drawing those grips.
This allows renderers (or anyone really) to attach "render data" to
textures. Only the first render data sticks.
You can gsk_texture_set_render_data() with the key you will use to
look the data up again, and if no data has been set yet, yours will be
set.
You can retrieve this data via gsk_texture_get_render_data() later on.
If your data has been cleared, NULL will be returned.
When gsk_texture_clear_render_data() is called (which the texture will
call when it is finalized), your destory notify will be called and you
have to release your render data.
The GL driver uses this to attach texture ids to GskTextures.
We do no longer bind textures to a renderer, instead they are a way for
applications to provide texture data.
For now, that's it. We've reverted to uploading it from scratch every
frame.
The snapshot vfuncs must only append at most a single node,
otherwise things are going to break if the widget is the root node.
Unfortunately there is no code that can check this in a generic fashion,
so we'll have to debug this on a case-by-case basis.
This happens in regular code paths for example when trying to render the
empty text string. We don't want to store a surface on the render
node in such a case (so actual rendering isn't slowed down), but we do
want to return a working cairo context that is not in an error state
(so the cairo rendering can continue without error messages).
We want to unrealize the renderer only after all widgets have been
unrealized. Otherwise, the widgets cannot release rendering resources
like textures.
Note that this implementation does not respect GDK windows at all. If
your widget requires respecting them, you should write your own
snapshot implementation and not chain up.
We now look at which of get_render_mode, draw or snapshot vfuncs is the
latest to have been overwritten in the class tree and then use that one.
This allows GtkContainerClass and GtkBinClass to override all of them
for without screwing things up.
and gtk_snapshot_render_frame() to be direct replacements for the
old gtk_render_*() functions.
Use them to replace Cairo usage completely in gtk_window_snapshot().
We now try to emulate cairo_t:
We keep a stack of nodes via push/pop and a transform matrix.
So whenever a new node is added to the snapshot, we transform it
by the current transform matrix and append it to the current node.
Unlike other container widgets, GtkStack would allocate its children
prior to moving its windows, which might prevent further valid size
allocation signals to be emitted.
Re-order the size allocation of child widgets to be performed after
moving the GtkStack windows.
Thanks to Owen for spotting the real issue here.
https://bugzilla.gnome.org/show_bug.cgi?id=767713
As in the last commit on gdkdisplay-win32.c, we need to define that to be
0x0600 (Vista) or later so that the items needed in the Windows headers be
activated.
See: https://bugzilla.gnome.org/show_bug.cgi?id=768081#c62
... to be for Vista (0x0600) or later. This is so that the necessary
items in the Windows headers be activated so that the code will build
properly on mingw-w64, and we already require Vista or later for GTK+.
Thanks Ting-Wei Lan for pointing this out.
See: https://bugzilla.gnome.org/show_bug.cgi?id=768081#c62
This fixes a DOS where any app can cause all running gtk apps
to use arbitrary amounts of memory.
Originally reported against mate-panel, where running a big slideshow
in eye-of-mate caused increasing RAM usage in mate-panel.
v2: Hardcode the value
Signed-off-by: Lauri Kasanen <curaga@operamail.com>
https://bugzilla.gnome.org/show_bug.cgi?id=773587
Making sure the surfaces are using the same scale factor makes it more
likely a fast path will be used when pixman gets involved, as pointed
out by Benjamin Otte.
https://bugzilla.gnome.org/show_bug.cgi?id=772075
We are currently truncating job names to 255 bytes, because that's the
maximum allowed length of job-name attribute in CUPS. This is a CUPS
limitation that GtkPrintOperation shouldn't need to know, and it
shouldn't affect other backends, that might have other limitations or
even no limitation at all. This has another side effect, that what you
set as GtkPrintOperation:job-name could be different to what you get if
the property is truncated, this is not documented in
gtk_print_operation_set_job_name(). So, I think the job name should be
truncated by the CUPS backend, right before setting the job-name
attribute.
https://bugzilla.gnome.org/show_bug.cgi?id=774097
The GApplication platform data may contain a startup ID that on X11
is used to set the startup notification ID when activated. Do the
same on the wayland backend to make startup notifications work for
DBus-activated applications where the DESKTOP_STARTUP_ID environment
variable is not set.
https://bugzilla.gnome.org/show_bug.cgi?id=768531
For wayland clients, the startup notification ID is currently only set
from the DESKTOP_STARTUP_ID environment variable. As that variable is
only set for clients launched via exec(), startup completion is not
indicated correctly for DBus-activated applications unless an explicit
ID is specified - usually that is not the case, as the default handling
uses gdk_notify_startup_complete().
To address this, we need API to set the startup notification ID from GTK
as we have on X11.
https://bugzilla.gnome.org/show_bug.cgi?id=768531
The spec says:
"If <shape> is omitted, the ending shape defaults to a circle if the <size>
is a single <length>, and to an ellipse otherwise."
Make it so.
It wasn’t clear that gtk_style_context_get[_valist]() behave like
g_object_get() — i.e. pointer-based types are returned newly-allocated.
Clarify that.
https://bugzilla.gnome.org/show_bug.cgi?id=773954
GtkLevelBar supports adding custom offsets as style classes, and they
are applied whenever the :value property matches. The current code,
however, only updates any CSS nodes when an offset is found, causing
it to not update when a discrete value changes but no custom offset
is added.
Fix that by always updating the CSS nodes.
https://bugzilla.gnome.org/show_bug.cgi?id=773799
This way we can recommend that applications use the
fullscreen_on_monitor() API on both X and Wayland otherwise they'd
have to keep a path for each backend to achieve this functionality.
https://bugzilla.gnome.org/show_bug.cgi?id=773857
This enables HiDPI support for GTK+ on Windows, so that the
fonts and window look better on HiDPI displays. Notes for the current
work:
-The DPI awareness enabling can be disabled if and only if an application
manifest is not embedded in the app to enable DPI awareness AND a user
compatibility setting is not set to limit DPI awareness for the app, via
the envvar GDK_WIN32_DISABLE_HIDPI. The app manifest/user setting for
DPI awareness will always win against the envvar, and so the HiDPI items
will be always setup in such scenarios, unless DPI awareness is disabled.
-Both automatic detection for the scaling factor and setting the scale
factor using the GDK_SCALE envvar are supported, where the envvar takes
precedence, which will therefore disable automatic scaling when
resolution changes.
-We now default to a per-system DPI awareness model, which means that we
do not handle WM_DPICHANGED, unless one sets the
GDK_WIN32_PER_MONITOR_HIDPI envvar, where notes for it are in the
following point.
-Automatic scaling during WM_DISPLAYCHANGE is handled (DPI setting change of
current monitor) is now supported. WM_DPICHANGED is handled as well,
except that the window positioning during the change of scaling still
needs to be refined, a change in GDK itself may be required for this.
-I am unable to test the wintab items because I don't have such devices
around.
https://bugzilla.gnome.org/show_bug.cgi?id=768081
Now that GTK+ is built as a single DLL, and the .lib that is built is
gtk-4.lib, we need to update the autotools sections in generating the
NMake Makefile snippets so that we can have the correct commands and flags
for building the .gir files, which will all now link to gtk-4-vsXX.dll (or
so).
As with the autotools builds, use gtk-4 as the name of the .lib file that
is produced from the build.
Actually this is already done with GTK-3.x with the autotools builds,
but this update is not done there as gtk-3.0.lib/gdk-3.0.lib/gailutil-3.0.lib
was used for such a long time that changing it there might have caused
trouble for people there.
Update the project configs to build GDK/GSK as a static lib and include
them into the GTK+ DLL as a monolithic DLL, which is in line with what is
done in the autotools builds, since the code changes needed for Windows
builds for a monolithic build are now in place.
Now that the autotools build folded the GDK/GSK bits into the main GTK+
DLL, there are some updates that need to be done for this. We need to:
-Fold the DllMain() of GDK-Win32 into the main GTK+ DllMain(), as we need
the HINSTANCE to register the window. We can't have two DllMain()'s in a
single DLL.
-Remove the GDK rc(.in) files, as that is not used anymore. Make the GTK+
.rc(.in) file load the gtk.ico GTK+ logo file instead so that we still
get the GTK+ logo for the application icon by default. Update the
autotools build files as well.
-Revert commit b9f9980 as LRN pointed out in comment 25 in bug 773299, as
GTK+ is now a monolithic DLL, and we ought not to export this private
function.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
gdk_wayland_window_attach_image() is normally called from
gdk_window_end_paint() to notify the compositor of newly staged drawing.
If any of the drawing code inadvertently dispatches the wayland event
loop (for instance with a gdk_flush() call), then it's possible that by
the time gdk_window_end_paint() is called, the staged drawing is already
destroyed.
This commit bypasses the attach_image call in scenarios where the staged
drawing is prematurely dropped.
https://bugzilla.gnome.org/show_bug.cgi?id=773274
Commit d249e77 (API: screen: Remove gdk_screen_is_composited()) attempted
to update the GDK-Win32 for the removal of the API, but some parts were
missed. This updates the code so that things continue to build and run.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
gtk/inspector/rendernodeview.c calls this private function from GSK, so we
need to ensure that this function is exported so that GTK+ can link
properly on compilers that do not support automatic exporting.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
This is a problematic struct, and giving direct access to it
has kept us from making improvements to GtkTextView. Drop it
from the public API, together with the auxiliary APIs. If
it turns out that this functionality is needed, we should add
individual getters.
We now record all render operations and display them.
Warning: This is very brute force, you can't clear the recordings or
turn recording off. And this thing easily records 25MB per recorded
frame, so be careful to not run out of memory and get your browser
killed. ;)
This one introduces the Recording object which is essentially a single
instance of something that happened.
The RenderRecording is an instance of an actual rendering operation.
The GLSL versions are:
OpenGL 2.1: #version 110
OpenGL 3.0: #version 130
OpenGL 3.2: #version 150
OpenGLES 2.0: #version 100
OpenGLES 3.0: #version 300 es
So we need to check the version of the GdkGLContext if we want use the
appropriate version, especially for legacy OpenGL contexts, which can be
both 3.x and 2.x.
This ensures that the drawing does not extend the actually drawn area.
It also ensures that our math is sane, because the math assumes the clip
area cannot extend the window. After all, before GTK4 it always was like
that.
Fixes a bunch of drawing bugs when the clip area does indeed extend too
far.
We want to have the coordinate system of the created cairo surface to be
identical to the coordinate system of the node's bounds. For that, we
need to translate the cairo surface by the bounds' origin.
In that case, we can't just rely on the stack allocation being big
enough. Especially, the child can actually be bigger than the current
stack allocation, so take that into account when positioning it.
The typical UI file has a lot more <property> tags than it has
<requested> or <interface> tags, etc. so order the string comparisons
according to this expected case.
We need an overridable entry point for GskRenderer to create Cairo
surfaces.
Implementations of GskRenderer can override create_cairo_surface() to
create efficient surfaces, possibly with zero copies involved, depending
on the GDK backend.
Switch code to use gdk_display_is_composited() instead.
The new code also doesn't use a vfunc to query the property but rather
requires the backend to call set_composited()/set_rgba() to change the
value.
Also add properties for those two properties.
The first property is equivalent to checking if an RGBA visual exists,
the 2nd is equivalent to gdk_screen_is_composited().
This reverts commit 8e29222d95.
This needs more work - spin buttons need to be converted at
the same time, and we should make sure that text still appears.
Update the GDKGL implementation:
-Allow legacy contexts to be created.
-Use finer-grained attributes to ask for a pixel format when possible,
which also adds support for anti-aliasing
In fact the changes here are required for GTKGL to work properly on
Windows for 4.x.
Note that creation of gles contexts is not done here, as the system does
not support such contexts directly on Windows, but only through means such
as ANGLE, which is a totally different issue here.
https://bugzilla.gnome.org/show_bug.cgi?id=773528
Add the needed custom build steps to generate the GResource and
enumeration sources that is needed for the build, and make sure that the
build is able to find the Graphene headers and lib.
Also add the necessary CFLAGS needed for building GSK.
This follows what happens in the autotools builds.
Also fix some project namespaces and the install project where there is a typo
in a project dep, which may cause a problem during the build.
Includes the ability to turn on updates in the inspector. Animations are
now run via a tick function which allows us to neatly overlay a
semi-transparent red rectangle and fade it out over time.
It also probably enables way more, but somebody with more UI neatness
than me needs to figure out what it eanbles first...
Before, we would immediately invalidate the GdkWindow of the widget, now
we call the parent's GtkWidgetClass.queue_draw_child() function.
This allows the parent to track redraw queueing of children.
By default GtkWidgetClass.queue_draw_child() will again chain up to its
parent while respecting the GdkWindow hierarchy for clipping.
GtkWindow is then the only widget actually invalidating the GdkWindow.
This essentially moves redraw queueing from GDK to GTK.
For some reason we end up allocating the colorplane widget
before it is realized, and then never initialize the surface.
Fix this by explicitly doing it on realize.
https://bugzilla.gnome.org/show_bug.cgi?id=773474
This merged gtk, gdk and gsk into one library, making it possible to
have internal private APIs between gtk them, as well as producing more
efficient code.
https://bugzilla.gnome.org/show_bug.cgi?id=773100
This adds the initial MSVC build items needed to build GSK under Visual Studio,
this is part of it that is required, we need to add items to the property sheets
to generate the code that is generated via glib-mkenums and glib-compile-resources.
This set includes, with the autotools scripts for the complete:
-GSK project files, which is integrated into the gtk+-4.sln.
-The NMake snippets to build the introspection files for GSK.
-The .bat files to call glib-mkenums to generate the enumeration sources.
We now need C99 features from the compiler which are only supported by
Visual Studio 2013 and later, so drop the MSVC 2008~2012 projects, and make
the baseline supported Visual Studio version be 2013. Update the build files
as a result.
GTK+-3.89.0 and later will require C99 features that is only supported on
Visual Studio 2013 and later, so prepare build/Makefile.msvcproj for this.
We still keep the 2008~2010 stuff here as this strives to be a shared module,
and there are projects using this that still supports building on pre-2013
Visual Studio.
These complicate a lot of GdkWindow internals to implement features
that not a lot of apps use, and will be better achieved using gsk.
So, we just drop it all.
Add a new ::measure vfunc similar to GtkCssGadget's that widget
implementations have to override instead of the old get_preferred_width,
get_preferred_height, get_preferred_width_for_height,
get_preferred_height_for_width and
get_preferred_height_and_baseline_for_width.
Now that the use_es field is an int with a possible negative value, we
cannot use it its truth value directly; we need to check if it's a
positive value, instead.
GDK defaults to asking for an OpenGL 3.2 Core Profile, but if we get a
legacy profile from the underlying windowing system, the OpenGL version
will be fixed to 3.0. If that happens, we need to set the legacy bit on
the GdkGLContext, since that bit will be used to determine the version
and type of GLSL shaders that will be used by application and toolkit
code alike.
With best-effort, try to use gdk_window_move_to_rect() more often, when
all pieces fit together. For the non-legacy paths to be triggered for
when gtk_menu_popup_for_device() or gtk_menu_popup() were used, the
following conditions must be met:
1) There is no custom positioning function specified
2) The menu is attached to a widget (using gtk_menu_attach_to_widget())
3) There is a associated grab device
https://bugzilla.gnome.org/show_bug.cgi?id=772922
xdg_shell v6 allows grabless popups, whose behavior is not that
different from override redirect windows with no grab to take
keyboard input (and pointer events outside).
This means we can relax the requirement to have a grab before
creating an xdg_popup. The warning is still useful to have so
people stop relying on gdk_window_show();gdk_device_grab() being
an ok pattern to popup a window, it's been moved to wayland
implementation of gdk_device_grab() instead, so we warn if trying
to grab a GDK_WINDOW_TEMP window that's already visible.
https://bugzilla.gnome.org/show_bug.cgi?id=771694
Newer versions of Automake warn about forward compatibility when the
build uses sources in sub-directories without the subdir-objects option.
Both GTK+ and GDK have an almost-but-not-quite non-recursive Automake
layout, with sources in sub-directories contributing to the build of a
top-level object.
In theory, just adding subdir-objects to AM_INIT_AUTOMAKE would be
enough, but the test suite references sources in a different top-level
in order to build tests that verify the implementation of private data
structures. This is not really allowed when using subdir-objects and out
of srcdir builds. In order to fix this case, we require some ad hoc
rules to create symbolic links in the appropriate build directory.
the darker bottom border used on buttons looks bad on circular ones
so now a gradient clipped on the border-box and a transparent
border is used in that partcular case.
See https://bugzilla.gnome.org/show_bug.cgi?id=771205 for details.
$button_fill contains the background-image property value of
buttons, having it readable outside the drawing mixin allows, for
example, stacking background images in an easier way.
This updates all the projects files to be be named appropriately as we move from GTK-3.x to 4.x,
and updates the autotools files so that things are distributed and generated properly.
Also remove deprecated/gtkstatusicon-quartz.c from gtk/Makefile.am, as that was causing 'make dist'
to fail as that file has been removed.
This fixes 'make dist' with the updated existing project files in proper order.
Note that this does not include the new GSK, which will be added later, so the project files do
not yet build the whole stack on Visual Studio at this point.
Make the Makefile.am targets for generating the Visual Studio projects re-generate the
project files and the header listings whenever the Makefile.am's that include
build/Makefile.msvcproj changes, so that whenever a source/header is added, they will
be reflected in the projects and in the property sheets that are used to copy the
headers.
Also ensure that these are applied to the vs11, vs12 and vs14 projects when this
happens, as they are copied and processed from the Visual Studio 2010 projects.
While porting GTK to GskRenderer we noticed that the current fallback
code for widgets using Cairo to draw is not enough to cover all the
possible cases.
For instance, if a container widget still uses GtkWidget::draw to render
its children, and at least one of them has been ported to using render
nodes instead, the container won't know how to draw it.
For this reason we want to provide to layers above GSK the ability to
create a "fallback" renderer instance, created using a "parent"
GskRenderer instance, but using a Cairo context as the rendering target
instead of a GdkDrawingContext.
GTK will use this inside the gtk_widget_draw() implementation, if a
widget implements GtkWidgetClass.get_render_node().
We're going to need to allow rendering on a specific cairo_t in order to
implement fallback code paths inside GTK; this means that there will be
times when we have a transient GskRenderer instance that does not have a
GdkDrawingContext to draw on.
Instead of adding a new render() implementation for those cases and then
decide which one to use, we can remove the drawing context argument from
the virtual function itself, and allow using a NULL GdkDrawingContext
when calling gsk_renderer_render(). A later commit will add a generic
function to create a transient GskRenderer with a cairo_t attached to
it.
Renderers inside GSK will have to check whether we have access to a
GdkDrawingContext, in which case we're going to use it; or if we have
access to a cairo_t and a window.
GskRenderNode is, at its core, a write-only API; you're supposed to set
up the render nodes instead of querying them for state.
Querying render nodes is left to the GskRenderer implementation.
Add a should_propagate function for render nodes. Eventually,
this is meant to avoid creating render nodes for child widgets
that are outside the parents clip area. Since we don't have
that information available right now, just filter out nondrawable
children for now.
Change get_render_node to return nodes that are sized to the clip
area and expect to be placed at the clip position; change
gtk_container_propagate_render_node to place child render nodes
accordingly, and change gtk_css_gadget_get_render_node to return
nodes that are sized accordingly as well.
We store the vertices in (unscaled) window coords (but the item size
is still scaled to match the texture size). Also, the
projection/model-view multiplication order is switched so that the scale
is applied at the right place.
The renderer will always use nearest-neighbor filters because it renders
at 1:1 pixel to texel ratio.
On the other hand, render nodes may be scaled, so we need to offer a way
to control the minification and magnification filters.
If we already have a GL texture we definitely don't want to use
gdk_cairo_draw_from_gl() to draw on a Cairo context if we're going
to take the Cairo surface to which we draw and put it into an OpenGL
texture.
The details of the modelview and projection matrices are only useful for
the GL renderer; there's really no point in having those details
available in the generic API — especially as the Cairo fallback renderer
cannot really set up a complex modelview or a projection matrix.
Instead of using the background as the gadget's node, we add a
non-drawing node that can be used to apply offsets; all other nodes are
children of the "box" node.
Just like we reuse texture ids with the same size we can, at the expense
of a little memory, reuse vertex buffers if they reference the same
attributes and contain the same data.
Each VAO is marked as free at the end of the frame, and if it's not
reused in the following frame, it gets dropped.
The child-transform is useful only if we also provide clipping to the
parent nodes, otherwise children will just be drawn outside of the
parent's bounds.
We'll introduce child transforms either at a higher layer, or once we
add clipping support to GskRenderNode.
This is the first example of indirect rendering involving
a box gadget. For now, we iterate the child gadgets manually,
and rely on gtk_container_propagate_render_node for the
child widgets. Eventually, we may want a better solution
here.
...and implement it for GtkCssGadget and GtkCssCustomGadget.
This allows us to decide on a per-object basis if a custom
gadget needs a render node for content or not.
The custom gadget draw function has the side effect of informing
the gadget machinery wether to draw focus or not. Bring the
draw function back, just for its boolean return value. We may
want to find a better solution for this.
I don't think this should stay in the code long-term, but it
is useful for debugging. It helped me track down some suspicious
placements of render nodes.
Give all nodes the same detail about the owner widget.
This reveals that every GtkCssCustomGadget gets a
DrawGadgetContents node, even if their draw_func is NULL.
We may want to come up with a better solution for that.
When creating the GskRenderNodes for the gadgets we should not translate
the coordinates inside the Cairo context, but we should tweak the
coordinates of the anchor point.
This is still not enough to get an appropriate rendering, as the result
is still slightly offset to the left.
Instead of passing the size of the buffer, we should pass the number of
quads; we know what the size of a single quad structure is, so we can do
the multiplication internally when creating the VAO.
This allows us to print the quads for debugging purposes.
GtkWidget.create_render_node() sets up a GskRenderNode appropriate for
rendering the contents of a widget, including its bounds,
transformation, and anchor point.
The naming is consistent with other scene graph libraries, as it
represents an additional translation transformation applied on top of
the provided transformation matrices.
We can also simplify the implementation by applying the translation when
we compute the world matrix.
We keep the textures used inside a frame around until the end of the
following frame; whenever we need a texture with the same size, and
it's not marked in use, then we just reuse the existing texture.
We were allocating a surface thats big enough for the clip, and
we were setting the transform for that, but then GtkContainer
was overriding the transform with the one for the allocation.
Also, we were drawing at the clip position, not the allocation
position.
This was overwhelming other useful debug output, so make it
opt-in. We print the render items for both opengl and transforms,
since the matrices bleed into each other, otherwise.
Since we use an FBO to render the contents of the render node tree, the
coordinate space is going to be flipped in GL. We can undo the flip by
using an appropriate projection matrix, instead of changing the sampling
coordinates in the shaders and updating all our coordinates at render
time.
We need to apply a scaling factor whenever we deal with user-supplied
coordinates, like:
- when creating textures
- when setting up the viewport
- when submitting the scene
Render nodes need access to rendering information like scaling factors.
If we keep render nodes separate from renderers until we submit a nodes
tree for rendering we're going to have to duplicate all that information
in a way that makes the API more complicated and fuzzier on its
semantics.
By having GskRenderer create GskRenderNode instances we can tie nodes
and renderers together; since higher layers will also have access to
the renderer instance, this does not add any burden to callers.
Additionally, if memory measurements indicate that we are spending too
much time in the allocation of new render nodes, we can now easily
implement a free-list or a renderer-specific allocator without breaking
the API.
We cannot implement GtkWidgetClass.get_render_node() in GtkContainer
without breaking the fallback path that renders a widget to a single
render node rasterization. For GtkContainer subclasses we should provide
a simple API, similar to gtk_container_propagate_draw(), that gathers
all the render nodes for each child.
If a node is non-opaque and has a non-zero opacity we need to paint its
contents and children first to an off screen buffer, and then render the
resulting texture at the desired opacity — otherwise the opacities will
combine and result in the wrong rendering.
We're not going to use separate rendering lists soon, and the way we
render items is less similar to a gaming engine and more similar to a
simpler compositor. This means we don't need to perform a two pass
rendering — opaque items first, transparent items later.
Use appropriate names, and annotate the names with the types — 'u' for
uniforms, 'a' for attributes. The common preambles for shaders are split
from the bodies, so we need some way to distinguish the uniforms and the
attributes just from their name.
We want the GL driver to cache as many resources as possible, so we can
always ensure that we're in a consistent state, and we can handle state
transitions more appropriately.
Drop the texture parameters handling from the texture creation, and
associate them with the contents upload. Also, rename the function to
something more in line with what it does.
We want to add the list of FBOs tied to a texture; this means we cannot
trivally copy the Texture structure when adding it to a GArray. We're
also going to have more textures than VAOs, so it makes more sense to
use a O(1) access data structure for them.
We can use the GL_ARB_timer_query extension (available since OpenGL
3.2, and part of the OpenGL specification since version 3.3) to query
the time elapsed when drawing each frame. This allows us to gather
timing information on our use of the GPU.
The clip rectangle may have non-zero offsets, so we need to ensure that
the GskRenderNode associated to the rendered area is translated by those
same offsets.
For the root node we do not need to use blending, as it does not have
any backdrop to blend into. We can use a simpler 'blit' program that
only takes the content of the source and fills the texture quad with
it.
We should use ShaderBuilder to create and store programs for the GL
renderer. This allows us to simplify the creation of programs (by moving
the compilation phase into the ShaderBuilder::create_program() method),
and move towards the ability to create multiple programs and just keep a
reference to the program id.
We should keep the ShaderBuilder around and use it to query the various
uniform and attribute locations when needed, instead of storing those
offsets into the Renderer instance, and copying them. This allows a bit
more flexibility, once we have more than one program built into the
renderer.
The GL renderer should build the GLSL shaders using GskShaderBuilder.
This allows us to separate the common parts into separate files, and
assemble them as necessary, instead of shipping one big shader per type
of GL API (GL3, GL legacy, and GLES).
GskShaderBuilder is an ancillary, private type that deals with the
internals of taking GLSL shaders from resources and building them,
with the additional feature of being able to compose shaders from a
common preamble, as well as adding conditional defines (useful for
enabling debugging code in the shaders themselves).
Using GObject as the base type for a transient tree may prove to be too
intensive, especially when creating a lot of node instances. Since we
don't need properties or signals, and we don't need complex destruction
semantics, we can use GTypeInstance directly as the base type for
GskRenderNode.
We need a virtual function to retrieve the GskRenderNode for each
widget, which is supposed to attach its own children's GskRenderNodes.
Additionally, we want to maintain the existing GtkWidget::draw mechanism
for widgets that do not implement get_render_node() — as well as widgets
that have handlers connected to the ::draw signal.
This commit changes the way GskRenderer and GskRenderNode interact and
are meant to be used.
GskRenderNode should represent a transient tree of rendering nodes,
which are submitted to the GskRenderer at render time; this allows the
renderer to take ownership of the render tree. Once the toolkit and
application code have finished assembling it, the render tree ownership
is transferred to the renderer.
Whenever the render tree changes we want to drop the RenderItem arrays,
as each item contains a pointer to the GskRenderNode which becomes
dangling once the root node changed.
The surface-to-GL upload logic has become more complicated with the
addition of the GLES code paths; it's more logical to have a public
utility function that can be called from GDK users, instead of copy
pasting the whole thing multiple times.
GSK is conceptually split into two scene graphs:
* a simple rendering tree of operations
* a complex set of logical layers
The latter is built on the former, and adds convenience and high level
API for application developers.
The lower layer, though, is what gets transformed into the rendering
pipeline, as it's simple and thus can be transformed into appropriate
rendering commands with minimal state changes.
The lower layer is also suitable for reuse from more complex higher
layers, like the CSS machinery in GTK, without necessarily port those
layers to the GSK high level API.
This lower layer is based on GskRenderNode instances, which represent
the tree of rendering operations; and a GskRenderer instance, which
takes the render nodes and submits them (after potentially reordering
and transforming them to a more appropriate representation) to the
underlying graphic system.
This is an attempt to get rid of gdk_window_new() for more specific use
cases. These 2 are for client-side windows - regular ones and input-only
ones resepectively.
So far all those functions just call into gdk_window_new().
On windows you might not have a theme installed by default which
means that when trying to create the context quark it will fail.
If then we try to replace a NULL key in the hash table it will crash.
https://bugzilla.gnome.org/show_bug.cgi?id=769859
Just like GLib, GTK+ would benefit from getting warnings and errors from
the compilers.
We check various, common warnings, especially for a future use of C99;
additionally, we promote some warnings to errors, in order to ensure
that simple mistakes are caught during the development phase, before
they are submitted to the code repository.
Use a static array for the known icon sizes, now that we don't allow
registering custom icon sizes any more. This allows us to cut a one-off
allocation that makes Valgrind sad.
The update tracking code was ugly and using deprecated drawing APIs. It
was also in the wrong place.
So instead of trying to keep it working, I'll remove it. We need to find
a better way to put it and make it work there.
This API was only used in GtkModifierStyle and GtkStyleProperties and
they are both on their way out.
CSS properties must now be set using strings via the regular parser API.
And replaces its usages in GtkTextView/GtkStyleContext with a hard-coded
0.04 which was the default value for cursor-aspect-ratio. Also remove
the public gtk_draw_insertion_cursor which used draw_insertion_cursor
which in turn looked up cursor-aspect-ratio
And with it, gtk_widget_get_visual() and gtk_widget_set_visual() are
gone.
We now always use the RGBA visual (if available) and otherwise fall back
to the system visual.
The cursor was set using gdk_window_set_cursor() even in
gdk_window_new().
So instead of having yet another flag, just make the users of that flag
call gdk_window_set_cursor() directly after the window was created.
X11 was the only backend to support it and people can just override it
using XSetClassHint() directly.
The docs already advertised the function as "Do not use".
Keep the existing call to XSetClassHint() in place, so that we keep
setting the same values as in GTK3.
... and gdk_screen_get_window_stack().
Those functions were originally added in
5afb4f0f11 but do not seem to be used as
they are not implemented anywhere but in X.
As GDK is not meant to fulfill window management functionality I'm going
to remove these functions without replacements.
... and gdk_screen_get_width_mm() and gdk_screen_get_height_mm() and
the shortcut counterparts that call these functions on the default
screen.
Modern display servers don't provide an ability to query the size of a
screen or display so we shouldn't allow that either.
Nobody ever does a NULL check there so all that causes is crashes. So
we better return a non-primary monitor than NULL.
Fixes gdk-wayland always returning NULL.
gtk_widget_destroy() removes widgets from their container. However
_internal_ widgets must be unref'ed using gtk_widget_unparent() instead.
This is symmetric with the fact that these widgets were ref'ed by direct
call to gtk_widget_set_parent(). It's also the method that was used in
gtk_headerbar_destroy().
https://bugzilla.gnome.org/show_bug.cgi?id=772859
> Due to Gtk+ keeping a reference to the window internally,
> gtk_window_new() does not return a reference to the caller.
> To delete a GtkWindow, call gtk_widget_destroy().
Caller(s) aren't expecting a need to delete help_overlay themselves
once they've installed it. (E.g. see gtk_application_window_added()).
I didn't notice any direct precedents, but there's a parallel in the
current implementation of gtk_container_destroy() which uses
gtk_widget_destroy() on any added widget.
This avoids leaking 100s of kB per window, when I tested nautilus.
https://bugzilla.gnome.org/show_bug.cgi?id=772859
This allows the use of a "text-direction" hint set to one of "none", "rtl",
or "ltr" to enforce the text direction of a "horizontal-buttons"
display-hint.
This is useful when a menu has buttons that map to physical space in the
UI and therefore must match the application widgetry.
https://bugzilla.gnome.org/show_bug.cgi?id=772775
ClutterEmbed on Wayland uses a subsurface and relocates it on configure
events, but when placed within a scrolled window, no configure event is
emitted and the ClutterEmbed subsurface remains static.
Emit a configure event for native windows in GdkWindow's internal
move_native_children() so that custom widgets relying on configure
events such as ClutterEmbed can relocate their stuff.
Similarly, when switching to/from normal/maximized/fullscreen states
which change the shadows' size and possibly shows/hides a header bar,
we need to emit a configure event even if the abs_x/abs_y haven't
changed to make sure the subsurface is size appropriately.
https://bugzilla.gnome.org/show_bug.cgi?id=771320https://bugzilla.gnome.org/show_bug.cgi?id=767713
to hilight drop target there is a wildcard selector which turns
the border and shadow to green, this clearly shouldn't happen when
the whole window is a drop target.
...by putting it in a stack. The busy_spinner and eject_button are
mutually exclusive, but only the latter was coded to ensure that its
visibility did not cause the rest of the row to reflow. By putting both
widgets in a stack and setting child_visible on that, the row allocates
enough space to show one - or none - at once, avoiding any misalignment.
https://bugzilla.gnome.org/show_bug.cgi?id=772345https://bugzilla.gnome.org/show_bug.cgi?id=772348
Calling eglGetDisplay forces libEGL to guess what kind of pointer you
passed it. Different EGL libraries will do different things here, and in
particular glvnd will do something different than Mesa. Since we do have
an API that allows us to explicitly type the display, use it.
The explicit call to eglGetProcAddress is working around a bug in
libepoxy 1.3, which does not understand the EGL concept of client
extensions. Since it does not, the normal epoxy resolver for
eglGetPlatformDisplayEXT would not find any provider for that entry
point, and crash when you attempted to call it.
Signed-off-by: Adam Jackson <ajax@redhat.com>
https://bugzilla.gnome.org/show_bug.cgi?id=772415
We currently beep when a character is appended at the end in
overwrite mode. That is obviously not right. Patch based on
a patch by Ian MacDonald.
https://bugzilla.gnome.org/show_bug.cgi?id=772389
- You will need to run the following upon completing install, from the build
directory in the Visual Studio 2008/SDK 6.0 command prompt (third line is not
needed unless -Dbuiltin_immodules=no is specified) so that the built binaries
can run:
for /r %f in (*.dll.manifest) do if exist <gtk_install_prefix>\bin\%~nf mt /manifest %f /outputresource:<gtk_install_prefix>\bin\%~nf;2
for /r %f in (*.exe.manifest) do if exist <gtk_install_prefix>\bin\%~nf mt /manifest %f /outputresource:<gtk_install_prefix>\bin\%~nf;1
for /r %f in (*.dll.manifest) do if exist <gtk_install_prefix>\lib\gtk-3.0\3.0.0\immodules\%~nf mt /manifest %f /outputresource:<gtk_install_prefix>\lib\gtk-3.0\3.0.0\immodules\%~nf;2
- The more modern visual style for the print dialog is not applied for Visual
Studio 2008 builds. Any solutions to this is really appreciated.
Using GTK+ on Win32
===================
To use GTK+ on Win32, you also need either one of the above mentioned
compilers. Other compilers might work, but don't count on it. Look for
prebuilt developer packages (DLLs, import libraries, headers) on the
above website.
Multi-threaded use of GTK+ on Win32
===================================
Multi-threaded GTK+ programs might work on Windows in special simple
cases, but not in general. Sorry. If you have all GTK+ and GDK calls
in the same thread, it might work. Otherwise, probably not at
all. Possible ways to fix this are being investigated.
Wintab
======
The tablet support uses the Wintab API. The Wintab development kit is
no longer required. The wintab.h header file is bundled with GTK+
sources. Unfortunately it seems that only Wacom tablets come with
# YourProject_HEADERS_EXCLUDES = ... # <list of headers to exclude from installation, separated by '|', wildcards allowed; use random unsed value if none>
#
# dist-hook: \ # (or add to it if it is already there, note the vs9 items will also call the vs10 items in the process)
@for %%f in (..\..\gtk\cursor\*.png)do @echo ^<file^>cursor/%%~nxf^</file^>>> $@
@for %%f in (..\..\gtk\gesture\*.symbolic.png)do @echo ^<file alias='icons/64x64/actions/%%~nxf'^>gesture/%%~nxf^</file^>>> $@
@for %%f in (..\..\gtk\ui\*.ui)do @echo ^<file preprocess='xml-stripblanks'^>ui/%%~nxf^</file^>>> $@
@for %%s in (16222432 48)do @(for %%c in (actions status categories)do @(for %%f in (..\..\gtk\icons\%%sx%%s\%%c\*.png)do @echo ^<file^>icons/%%sx%%s/%%c/%%~nxf^</file^>>> $@))
@for %%s in (scalable)do @(for %%c in (status)do @(for %%f in (..\..\gtk\icons\%%s\%%c\*.svg)do @echo ^<file^>icons/%%s/%%c/%%~nxf^</file^>>> $@))
@for %%f in (..\..\gtk\inspector\*.ui)do @echo ^<file compressed='true'preprocess='xml-stripblanks'^>inspector/%%~nxf^</file^>>> $@
for %%s in (16 22 24 32 48 256) do ((mkdir $(CopyDir)\share\icons\hicolor\%%sx%%s\apps) & (copy /b ..\..\..\demos\gtk-demo\data\%%sx%%s\gtk3-demo.png $(CopyDir)\share\icons\hicolor\%%sx%%s\apps))
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.
