Respect the matrix in use at time of encountering a repeat node so that
the offscreen uses roughly the same device pixel density as the target.
Fixes the handling of the clipped-repeat test.
Make it use an alpha value that is well defined, ie 0.4 instead of 0.5.
0.4 * 255 = 102
0.5 * 255 = 127.5
This avoids rounding issues where some math may cause the resulting
alpha value to be 127, and some other math ends up with 128.
We want to always reserve space for the clear icon,
but let the text widget use that space when the icon
isn't shown. A plain box layout can't do that, so
do our own size allocation.
Sometimes the GPU is still busy when the next frame starts (like when
no-vsync benchmarking), so we need to keep all those resources alone and
create new ones.
That's what the render object is for, so we just create another one.
However, when we create too many, we'll starve the CPU. So we'll limit
it. Currently, that limit is at 4, but I've never reached it (I've also
not starved the GPU yet), so that number may want to be set lower/higher
in the future.
Note that this is different from the number of outstanding buffers, as
those are not busy on the GPU but on the compositor, and as such a
buffer may have not finished rendering but have been returend from the
compositor (very busy GPU) or have finished rendering but not been
returned from the compositor (very idle GPU).
The idea here is that we can do more complex combinations and use that
to support texture-scale nodes or use fancy texture formats (suc as
YUV).
I'm not sure this is actually necessary, but for now it gives more
flexibility.
For blend and crossfade nodes, one of the children may exist and
influence the rendering, while the other does not.
Previously, we would skip the node, which would cause the required
rendering to not happen. We now send a valid texture id for the
invalid offscreen, thereby actually rendering the required parts.
Fixes the blend-invisible-child compare test
Current state for compare tests:
Ok: 397
Expected Fail: 0
Fail: 26
Unexpected Pass: 0
Skipped: 2
Timeout: 0
Instead of having a descriptor set per operation, we just have one
descriptor set and bind all our images into it.
Then the shaders get to use an index into the large texture array
instead.
Getting this to work - because it's a Vulkan extension that needs to be
manually enabled, even though it's officially part of Vulkan 1.2 - is
insane.
If we have a rectangular clip without transforms, we can use
scissoring. This works particularly well because it allows intersecting
rounded rectangles with regular rectangles in all cases:
Use the scissor rect for the rectangle and the normal clipping code for
the rounded rectangle.
The idea is to use it for clip nodes when they are integer-aligned.
To do that, we need to track the scissor rect in the parse state, so we
do that, too.
Also move the viewport offset out of the projection matrix, as it is
part of the transform between clip and scissor, so it needs to live in
the offset.
We align the data to a multiple of vertex stride, that way we use more
memory, but we could compute an offset into the vertex buffer without
changing the offset.
We can set the vertex offset while counting the data, this gets rid of
the need of passing all the counting machinery into the actual data
collection code.
When attempting a complex transform, check if the clip can be ignored
and do that if possible.
That way we don't cause fallbacks when transforming the clip is too
complex.
The idea is that for a rectangle intersection, each corner of the
result is either entirely part of one original rectangle or it is
an intersection point.
By detecting those 2 cases and treating them differently, we can
simplify the code to compare rounded rectangles.
Instead of emitting the render commands once per rectangle of the clip
region, just emit them once with the region's extents.
This is generally faster because it emits fewer commands to the GPU,
even though it may touch significantly more pixels.
For a proper method, we'd need to record the commands per clip rectangle
instead of emitting all of them all the time.
The border and color shaders - the ones that do AA - now multiply their
coordinates by the scale factor, which gives them better rounding
capabilities.
This in particular improves the case where they are used in fractional
scaling situations, where the scale is defined at the root element.
Previously, we just used the defaultscale factor, but now that we're
having it available in push constants, we can read it back for creating
offscreens and rendering fallbacks.
So do that.
It's a 1:1 replacement for GskVulkanPushConstants, just without the
indirection through a different file.
GskVulkanPushConstants as a struct is gone now.
The file still exists to handle the push_constants operation.
1. Use a graphene_vec2_t
2. Ensure it's always positive
3. Don't break with fallback
The scale value is nothing more than an indication of how many pixels to
assume per unit of a node.
We don't want to render the offscreen trnsformed, we want to render it
as-is.
We lose the correct scale factor, but that requires some separate work,
so for now it gets a bit blurry on hidpi.
This introduces the rect object and adds a rect_distance() and
rect_coverage() function.
_distance() returns the signed distance tp the rectangle.
_coverage() returns the coverage of a pixel centered at that position.
Note that the pixel size is computed using dFdx/dFdy.
When the node bounds were a non-integer size, the texture would get
ceil()ed pixels, but various viewport or scissor computations might
floor() instead, leaving the right/bottom row of pixels untouched.
Make sure those functions ceil(), too.
Instead of trapping errors for the whole loop trying to create GL
contexts, trap them once per GL context.
Apparently GLX does throw an error when a too high version is requested
and doesn't just return NULL and then that error lingers when we try
lower versions.
Fixes#5857
We may try to update the XRR outputs and Crtcs when they're changing in
the server, and so we may get BadRROutput that we're currently not
handling properly.
As per this, use traps and check whether we got errors, and if we did
let's ignore the current output.
It's not required to call init_randr13() again because if we got errors
it's very likely that there's a change coming that will be notified at
next iteration during which we'll repeat the init actions.
When registering an observer, we send a notification and for that we need
to query the action's state and param type. When setting up a muxer parent,
same thing happens, except the action is queried on the parent instead.
This means that the muxer will notify observers about the parent's actions,
but not about its own.
Add a test to verify it works.
Fixes https://gitlab.gnome.org/GNOME/gtk/-/issues/5861
This is not the optimal way of doing it: we're
reuploading the texture with client-side conversion.
But it fits nicely into our current handling of mipmaps.
We can do better once we use shaders for colorspace
conversions.
Add some odd-sized texture sizes to the
download tests, to trigger alignment issues
in the various upload code paths. And add
a size that is bigger than the max-texture-size
we force in one of our test setups.
To compensate, reduce the number of
runs per size from 20 to 10.
For non-gles, make it handle unpremultiplied formats,
and everything else, by downloading the texture in its
preferred format and, in most cases, doing a
gdk_memory_convert afterwards.
For gles, keep using glReadPixels, but handle cases
where the gl read format doesn't match the texture
format by doing the necessary swizzling before calling
gdk_memory_convert.
Make the callers of this function check for
straight alpha themselves, and only do the
version compatibility check here. This makes
the function usable in contexts where straight
alpha is acceptable.
Use &__ImageBase for the GTK DLL and GetModuleHandle (NULL)
for the application module. Then remove DllMain as it's not
necessary anymore.
References:
[1] Accessing the current module's HINSTANCE from a static library:
https://devblogs.microsoft.com/oldnewthing/20041025-00/?p=37483
The display xevent signal connection takes the ownership of the stream
until we get a valid event, so it should manage the stream lifetime.
So make this clearer, by automatically removing the stream reference
when we disconnect from the xevent signal handler.
We create a new stream during gdk_x11_selection_input_stream_new_async()
then such stream is referenced when passed to the task via
g_task_return_pointer(), so there's no need to reference it again before
returning it, or we'd end up leaking.
Fixes: https://gitlab.gnome.org/GNOME/gtk/-/issues/4892
The GL renderers like to premultiply content that isn't, and due to the
data loss with alpha == 0 (transparent white, transparent black and
transparent anything are all represented by (0, 0, 0, 0) when
premultiplied) these values cannot be converted back.
There is no longer a need to use gdk_texture_download() and force
conversion to ARGB8 format. We can download the pixels in the original
format again.
That way we avoid testing the conversion code and avoid having to deal
with differences in representable colors.
However, some formats do do conversions, so we allow pixel comparisons
to be accurate (requires 16bit comparison accuracy) or inaccurate (we
only care about 8bit).
Note that for the default RGBA formats, this is identical and means they
need to be bit-exact the same, no matter what.
But the higher bit depth formats may be more different - floating point
can even have different values with high accuracy (the float mantissa is
23 bit, we only care about 16).
The groups hash table is initialized lazily when inserting
the first GActionGroup (gtk_action_muxer_insert ()). Do as
all surrounding code does and check for NULL before using
groups.
This avoids triggering a warning
When we emit items-changed due to a section
sorter change, don't also emit sections-changed.
Instead make the items-changed signal cover the
whole range.
Tests included.
When the section sorter changes, we need to update
the keys, otherwise the sorter will continue to report
the old sections.
This code is currently a bit suboptimal, since the
creation of sort keys and section sort keys are
muddled together.
Fixes: #5854
When the section sorter changes, we need to update
the keys, otherwise the sorter will continue to report
the old sections.
This code is currently a bit suboptimal, since the
creation of sort keys and section sort keys are
muddled together.
Fixes: #5854
And recreate header and footer tiles as needed.
This commit was tested using a sortlistmodel, changing
the section sorter from sorting only by first char
to sorting by the first two chars, which changes
the number of sections, but leaves the alphabetic
order of items unchanged.
Without this, there are still GdkMonitors present for displays that are
present but disconnected (such as when a laptop disables the internal
display to connect to an external monitor).
XWayland (at least on gnome-shell) does not support SGI_swap_control,
which we were using to unset the swap interval.
It does support EXT_swap_control though, which is the more modern
version of the same thing, so this commit adds support for that.
And now GDK_DEBUG=no-vsync gives me >1000fps instead of just 60fps,
With XWayland and direct scanout it is possible that some apps get into
a situation where more than 2 buffers are in flight and in that case we
want to be able to still track the change regions for those buffers.
Usually 3 buffers are in use, so we go one higher, just to be safe.
Some mice send a value slightly lower than 120 for some detents. The
current approach waits until a value of 120 is reached before sending a
low-resolution scroll event.
For example, the MX Master 3 sends a value of 112 in some detents:
detent detent
| | |
^ ^ ^
112 REL_WHEEL 224
As illustrated, only one event was sent but two were expected. However,
sending the low-resolution scroll event in the middle plus the existing
heuristics to reset the accumulator solve this issue:
detent detent
| | |
^ ^ ^ ^
REL_WHEEL 112 REL_WHEEL 224
Send low-resolution scroll events in the middle of the detent to solve
this problem.
Related to https://gitlab.gnome.org/GNOME/mutter/-/issues/2469
We are using placeholders in the 'check' column
that are put in a size group, so that they all
take the same space once a check or radio is shown.
Unfortunately, for the inline-buttons option, we
were using a GtkBuiltinIcon as placeholder, and those
respect the -gtk-icon-size CSS property and take
a minimum size of 16px. Use a GtkGizmo instead to
get the expected result of no extra padding unless
there's a check or radio.
Fixes: #5839
The non-portal fallback method for launching a file manager to show the
file in its parent directory was incorrectly using the `ShowFolders`
method (open a folder) instead of `ShowItems` (open the parent directory
and show the file).
The `show_item` function (previously `show_folder`) had an unused
`callback` parameter; it has been removed and the type of the parameter
containing the GTask has been renamed and now uses the correct type
instead of gpointer to reduce the amount of casting required.
FixesGNOME/gtk#5842
The want to use the footer tile at the end
to fill leftover space at the bottome right.
So lets assert that we actually dealing with
a footer tile, just in case something changes
in the future that might have us end up with
some other kind of tile.
This reverts commit e121a5ca6f.
The tile that was causing the critical in #5836
(what that commit was about) was a FILLER, and we
are getting rid of FILLER tiles here. Which will
avoid the issue in a more elegant way.
In height-for-width and hscrollbar-policy = never, we can provide
the child with a proper for_size when measuring it. The same is true for
width-for-height and vscrollbar-policy = never.
This allows for accurately measuring the size of eg. wrapping labels.
The cancellation path already clears the GCancellable, if we let it
continue, it causes a later assertion, so just exit early in this case
and hope a new path has been set.
Fixes: #5792
When the command queue is out of batches, there is
no point in doing further work like allocating uniforms.
This helps us avoid assertions in the uniform code
that we would hit when we run out of uniform space
too.
Commit 3090795351 accidentally caused all
CI builds (or at least the ones with -Werror) to no longer build tests,
examples and demos, so none of them had made sure that they compile.
When we start ignoring batches, we must do it everywhere,
or we may run into assertions. This was triggered by an
enormous text node tree produced by tests/rendernode-create.
The documentation says that the model returned by
gtk_notebook_get_pages() implements the GtkSelectionModel interface, but
checking the history confirms this is a lie.
Instead of fixing the documentation, we can easily make it true, and
reduce the differences between GtkNotebook and GtkStack.
Fixes: #5837
If there is a value passed to GSK_RENDERER, display it in the window
title.
This is mostly so that when I show off screenshots, people know what
renderer I'm using.
Vulkan has a different initial coordinate system to GL.
GL:
(-1, 1, -1) +------+.
|`. | `.
| `·--|---·
| : | :
+------+. :
`. : `.:
`·------· (1, -1, 1)
Vulkan:
(-1, -1, 0) +------+.
|`. | `.
| `·--|---·
| : | :
+------+. :
`. : `.:
`·------· (1, 1, 1)
so adjust the near and far plane we pass to
graphene_matrix_init_ortho() to make it end up with the same
projection as the GL renderer.
This one tests a crossfade between two non-overlapping nodes with a clip
region that covers neither of the two nodes.
This tests that renderers can deal with clip regions that doesn't
overlap nodes in a situation where they will most likely want to create
an offscreen.
As offscreens are typically clipped to the clip region, this would cause
an empty offscreen and that can cause failures.
This was an experiment where an offscreen was translated inside an
existing clip.
Because renderers try to limit offscreens to the clip rect, this is
interesting, because they might get the translation wrong.
Using gdk_texture_new_from_resource() is not valid here because we are
not sure if the given resource is valid.
Plus, the previous optimization is no longer relevant, because we are
not using gdk_pixbuf_new_from_resource() anymore - which was what this
optimization was about before it was ported to GdkTexture.
Test attached.
The filesystemmodel tracks changes and additions to child files
through G_FILE_MONITOR_EVENT_ATTRIBUTE_CHANGED. This event will also
occur if the parent directory is changed. Since the parent directory
doesn't exist in the model, it creates a non-existent item.
Fixes: https://gitlab.gnome.org/GNOME/gtk/-/issues/4233
Make unexport_handle take the handle, which will
let us deal with multiple exports in the future.
Update all callers to store their handle, and
pass it to unexport_handle.
The actions document ended with a : where there used to be a period,
seemingly a bug in the conversion from XML. Replace that : with a
period, and remove another spurious : inserted during the conversion.
Add some tests for handling of failures.
The test data here is taking from gdk-pixbufs
tests/test-images/fail directory, excluding anything
but png, tiff and jpg images.
Strangely, the document's last sentence ended with a colon,
giving the impression there's missing text that should follow.
There is not.
Even more strange, it's _always_ ended with a colon,
ever since the file was converted from XML
(in commit 10cd539104).
BUT, the XML file it was converted _from_ ended that same sentence
with a period! I have no idea where the colon came from.
gtk_css_provider_load_from_data has turned out
to be problematic for language bindings. Add two
new variants, from_string and from_bytes, to
replace it.
This partially reverts commit 93a875bf20.
Removing this annotation broke bindings, which now treat the
length of the array as a standalone parameter.
This broke Spiel, and probably other non-C apps that rely on
GtkCssProvider through bindings.
gtk and gdk have their own marshallers, for historic
reasons. But there's no reason to duplicate the same
code here.
Eventually, we should just move all marshallers to
the same place.
Make all length values from 0px to 8px available
as static values. This will help with cutting
down on the number of corner values (we have a
lot of 5px corners).
The fedora-x86-64 build does not only build with debug,
it also does the hello build, and it runs the testsuite
multiple times.
Move the hello build to the fedora release build. The idea
is that this lets us do more work in parallel, and spend
less time waiting for the longest-running ci job.
We are ignoring failures here, and nobody is working
on fixing them. And the failures end up at the end
of the log, adding annoyance to finding the actual
failures.
We only have one reference to the surface,
and it is dropped by gdk_surface_destroy.
All the users of surfaces in gtk had the same
bug: they were all stealing an extra reference
to drop - the one that the renderer was leaking
until recently.
This was the intention, but the object data by itself
does not achieve that: We do run dispose on the display
when it is closed, but object data is only cleared in
finalize. So listen to the ::closed signal and remove
the driver ourselves.
Fix up the drivers dispose implementation enough for
that to actually work.
Public headers should mainly include gdktypes.h, which already include
the symbol visibility and versioning macros; we can also modify
gdktypes.h to include the enumerations.
Since the corresponding source files do not include "config.h", and are
not compiled with `GTK_COMPILATION`, they will generate the wrong symbol
exporting under Windows.
Instead of injecting `-fvisibility=hidden` depending on a compiler check
ourselves, let Meson do it for us.
This also avoids us having to filter `-fvisibility=hidden` when reusing
the common compiler flags.
Fix the circular dependency by moving the generated
headers to gdk/version/, and build that directory
first.
Misc other fixes, such as putting the custom targets
as sources, not depedencies, and using the correct
major version in the generator script.
Let's poach the same script used by GLib to avoid having to add all the
version macros by hand every time we increment the GTK version.
This is a work in progress:
- need to rename the GLIB_STATIC_COMPILATION check
- circular dependency: libgtkcss depends on gdkversionmacros.h, but libgdk
depends on libgtkcss
The inertness concept introduced in 62e9d1e470 assumed a listview was
inert when no factory was set. This has 2 problems:
1. columnview uses a listview without factories.
2. header factory being set but factory not being set technically makes
the listiew inert - but should it?
So for now, make inertness only depend on visibility and root.
A side benefit is that this matches columnview semantics.
When ensuring widgets, ensure that their section is known. This will
be relevant when we use section widgets.
Also ensure that sections that don't cover any widget get destroyed.
The get_section() implementation is a slow and steady implementation
that has to be careful to not screw up when an incremental sort is only
partially sorted.
When GDK_DEBUG=no-vsync is on, we might have more than one outstanding
frame. Don't assert when that hapens. Just request a frame callback for
the first and skip the others.
Not all frames get timing info with GDK_DEBUG=no-vsync, so make sure
that even when we render tons of frames, the one frame that does get
timing info is still there when the timing info arrives.
I set it to 128 from 16 now.
This is roughly good enough to go to 5000fps from on a 60Hz monitor.
...when we are using wglChoosePixelFormatARHB(). This ensures that we
hvae a HDC with a pixel format that will really support alpha bits, as
we did for the traditional ChoosePixelFormat().
Thanks to Patrick Zacharias for testing and pointing things out.
We were sending random junk to ChoosePixelFormat().
Also assert that we don't overflow the array. That might be usefu to
know if we carelessly add attributes later.
... for creating the actual WGL contexts, so that we can cut down on the
number of times where we need to create the base, legacy WGL contexts in
order to create the WGL contexts with attributes. We could just use the
dummy context that we have to make it current to create the needed
WGL contexts.
If we are querying the best supported pixel format for our HDC via
wglChoosePixelFormatARB() (i.e. we have the WGL_ARB_pixel_format extension),
it may return a pixel format that is different from the pixel format that we
used for the dummy context that we have setup, in order to, well, run
wglChoosePixelFormatARB(), which sadly requires a WGL context (HGLRC) to be
current in order to use it, which means the dummy HDC already has a pixel
format that has been set (notice that each HDC is only allowed to have its
pixel format to be set *once*). This is notably the case on Intel display
drivers.
Since we are emulating surfaceless GL contexts, we are using the dummy GL
context (and thus dummy HDC that is derived from the notification HWND used in
GdkWin32Display) for doing that, we would get into trouble if th actual HDC
from the GdkWin32Surface has a different pixel format set.
So, as a result, in order to fix this situation, we do the following:
* Create yet another dummy HWND in order to grab the HDC to query for the
capabilities the GL drivers support, and to call wglChoosePixelFormatARB() as
appropriate (or ChoosePixelFormat()) for the final pixel format that we use.
* Ditch the dummy GL context, HDC and HWND after obtaining the pixel format.
* Then set the final pixel format that we obtained onto the HDC that is derived
from the HWND used in GdkWin32Display for notifications, which will become our
new dummy HDC.
* Create a new dummy HGLRC for use with the new dummy HDC to emulate surfaceless
GL support.
We are currently using g_clear_pointer() on the intermediate WGL contexts
(HGLRC)'s that we need to create in the way, which means that we need to ensure
that the correct calling convention for wglDeleteContext() is being applied.
To be absolutely safe about it, use the gdk_win32_private_wglDeleteContext()
calls, which will in turn call wglDeleteContext() directly from opengl32.dll
(using the OpenGL headers from the Windows SDK) instead of going via libepoxy,
which will assure us that the correct calling convention is applied.
Fixes issue #5808.
Our notify tests would fall over if there was
a duplicate enum value (within the first 10 values).
Make it handle that, by skipping the duplicate value.
It turns out that the old behavior of GTK_ALIGN_BASELINE
was actually used in libadwaita, so bring it back, and
introduce a new GtkAlign value for the new behavior.
Allow control-clicks on some fields to bring up
a more specific UI. This functionality is also
available via Ctrl-E and the context menu.
At this point, it can edit colors, fonts and
files in some places, as well as a few enums.
This is failing because I can't figure out
how to make wireplumber and pipewire work
in ci enough to let me add a new monitor :(
As usual, the test works fine locally.
Add some input tests that are using headless
mutter, and python with our in-tree gir files.
So far, test that we can roundtrip key events,
and move the pointer around.
Add some monitor tests that are using headless
mutter, and python with our in-tree gir files.
So far, we test that we get expected signals
when monitors are added and removed.
Transition to the color that is in use instead.
Fixes crashes because currentColor is not an RGBA color and
therefor could not be queried later.
Fixes#5798
Most of the time we want to compute them based on the child node we
render to the offscreen, but not always.
For blend and cross-fade nodes, they need to be computed based on the
node's bounds.
Fixes widget-factory page fade animation weirdly resizing the fading
pages.
We weren't looking in the build dir for generated files.
Actually make sure that we look in the build dir *first*, otherwise
glib-compile-resources will still use the wrong files.
... and use it in rendernodes.
Setting up textures for diffing is done via gdk_texture_set_diff() which
should only be used during texture construction.
Note that the pointers to next/previous are allowed to dangle if one of
the textures is finalized, but that's fine because we always check both
textures' links to each other before we consider the pointer valid.
When slicing the texture, the GL renderer was
forgetting to apply the viewport origin. This
shows up when rendering things with negative
scales, leading to negative origins.
The rounded-clip-in-clip-3d test fails in GL when
flipped. Given that it was already excluded from cairo,
and also fails cairo when flipped, give up on it for now.
Our coverage computation only works for well-behaved
rects and rounded rects. But our modelview transform
might flip x or y around, causing things to fail.
Add functions to normalize rects and rounded rects,
and use it whenever we transform a rounded rect in GLSL.
The repeated tests were not careful enough to produce
the correct reference image to match what the repeat
node does.
With these changes, all cairo tests pass.
Add separate suites for running the gsk compare-render
tests with the --flip, --rotate or --repeat options.
A bunch of these fail currently, and need diagnosis.
Add options to the gsk compare-render test for
modifying the node (and do a matching change to
the reference image).
flip: negative scale flipping things horizontally
rotate: 90 degree rotation
repeat: 2x2 grid
In horizontal layout, we line up the baselines of all children to find
how much space we need above and below the box baseline.
In vertical layout, we need to pick one child to inherit the baseline
from, which is what the new GtkBoxLayout:baseline-child property is
about. It is the equivalent of GtkGridLayout:baseline-row.
When we are not doing baseline alignment, don't pass
a baseline to the allocated widget. This helps because
a number of widgets (GtkLabel, GtkEntry, etc) always
position their text on the given baseline.
Since we show them in GNOME shell, show them here too.
The comment that says "only show these in the a11y
theme" was still there, but we were always hiding them.
The Expose events following a ConfigureNotify may arrive at
a time that we did not resize the surface yet, making these
expose events a no-op. Even though gsk/gtk take care of the
window content itself, this might lead to unrendered portions
of the window shadow.
This may be seen with GSK_RENDERER=cairo and GDK_BACKEND=x11,
attempting to tile a window (e.g. gtk4-demo) left or right.
The window will show black rectangles or other artifacts in
the window shadow areas that correspond to the newly painted
portions (as the window needs to expand vertically).
In order to fix this with a similar behavior to Wayland,
consider ourselves the whole surface invalidated after resize,
in order to ensure everything is painted from scratch.
... when it is available.
Also introduce the new function gdk_rectangle_transform_affine(), which
looks like overkill for this purpose, but I'm about to use it elsewhere.
Drop the section that talked about main and how to update
local checkouts - its been 2 years, people should have gotten
around to it by now. Add some general git hints instead.
There's no need for EGL to do any timing, we do it in GTK already.
This fixes hangs in Mesa when we hide a surface after a SwapBuffers()
but before the frame callback arrives.
If we then reshow the surface and immediately render to it, Mesa would
still have a frame callback from before the hiding and forever poll()
waiting for the compositor to send the callback.
Fixes#5761
donʼt mention its renamed successor either, as that has its own section
later. We could have another sentence paragraph like ‘In the case of
GtkBox, the pack methods have been renamed to X and lost the trailing
arguments Y’, but that wonʼt help people prepare still on GTK3, which is
the point in the affected section… so just remove the misleading relic.
When adjusting allocations, treat BASELINE more like CENTER
than like FILL. The results are better, in particular for
controls like entries or switches, which we never want to
scale up vertically, but still want to align to the baseline.
A grid layout lets us get the baseline right in
vertical orientation, by setting a baseline row.
It would be nice if the box layout supported this
as well, but currently it doesn't, and adding that
feature isn't trivial.
Pass the GLsync object from texture into our
command queue, and when executing the queue,
wait on the sync object the first time we
use its associated texture.
Add a new function to TextureBuilder that takes a GLsync that
requires internal code to wait on before using the texture.
Somewhat sneakily, we don't take the sync if syncs are not supported by
the current GL context.
As public API has no code to query the sync for the destroy notify, this
is fine and it means we don't have to do the check every time we want to
call gdk_texture_get_sync() internally.
Building GL textures is complicated, so create an object to make them.
So far, this object just contains the functionality of
gdk_gl_texture_new(), but that will change in the future.
In particular, we want to get the GL version, when the Windows box/VM
has an unsuitable GL implementation.
This is somewhat helpful in analyzing failures to bring up GL on
machines where users claim GL does work.
This way, we can realize it and either print success information about
it or return NULL if that fails.
This makes it more likely that we fail early, which means we can then
initialize EGL.
This refactor achieves the following:
* check GL version against proper matching context version
In particular, for legacy contexts, we now actually check
* make sure the actual version is set, even for legacy contexts
* make sure set_is_legacy() is set properly
Now that all contexts do that, insist that they keep doing it.
And because they keep doing it, we can support querying the GL version
from gdk_gl_context_get_version() without requiring the context to be
made current.
The EGL spec states:
The context returned must be the specified version, or a later
version which is backwards compatible with that version.
Even if a later version is returned, the specified version
must correspond to a defined version of the client API.
GTK has so far been relying on EGL implementations returning a
later version, because that is what Mesa does.
But ANGLE does not do that and only provides the minimum version, which
means Windows EGL has been forced to use a lower EGL version for no
reason.
So fix this and try versions in order from highest to lowest.
Don't notify during destruction, notify afterwards.
This way we don't call into user code from a half-destructed node.
Note that this changes the order in which those notifies happen when
collapsing a large tree: From parent node before child nodes to child
nodes before parent node.
No actual use case for this, just thought it would be safer.
While we are collapsing a subtree, some signal handlers may not be
disconnected while we are doing this. By adding this check and not
giving those nodes no longer access to the model, we can stop it from
modifying it while we are trying to collapse stuff.
Fixes some crashes in gnome-builder.
... to backends.
That way, frame clocks can be constructed by the backends' surface
implementations and dont need to be passed in as construct arguments.
Also add an assertion that they are indeed constructed.
That way, it doesn't need a specific init function.
Also chain up last, so that the generic initialization code in
GdkSurface::constructed can access a fully initialized macos surface.
That way, it doesn't ned a specific init function.
Also chain up last, so that the generic initialization code can access a
fully initialized wayland surface.
This is also how regular buttons behave. Otherwise releasing on a
different menu item would register a click on the item that was
originally pressed. In these cases it is better to not register a click
at all.
Closes: https://gitlab.gnome.org/GNOME/gtk/-/issues/5760
[30/1038] Compiling C object gdk/win32/libgdk-win32.a.p/gdkmain-win32.c.obj
../gdk/win32/gdkmain-win32.c:146:1: warning: 'gdk_win32_finalize_ole' defined but not used [-Wunused-function]
146 | gdk_win32_finalize_ole (void)
| ^~~~~~~~~~~~~~~~~~~~~~
../gdk/win32/gdkmain-win32.c:113:1: warning: 'gdk_win32_finalize_com' defined but not used [-Wunused-function]
113 | gdk_win32_finalize_com (void)
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
A number of warnings are produced:
[23/1038] Compiling C object gdk/win32/libgdk-win32.a.p/gdkinput-dmanipulation.c.obj
../gdk/win32/gdkinput-dmanipulation.c: In function 'reset_viewport':
../gdk/win32/gdkinput-dmanipulation.c:354:11: warning: variable 'hr' set but not used [-Wunused-but-set-variable]
354 | HRESULT hr;
| ^~
Try to do something sensible instead.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
../gdk/win32/gdkclipdrop-win32.c: In function 'transmute_cf_shell_id_list_to_text_uri_list':
C:/msys64/ucrt64/include/glib-2.0/glib/gstring.h:72:5: warning: ignoring return value of 'g_string_free_and_steal' declared with attribute 'warn_unused_result' [-Wunused-result]
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
In file included from ../gdk/win32/gdkdrag-win32.c:201:
../gdk/win32/gdkprivate-win32.h:45: warning: "GDK_NOTE" redefined
45 | #define GDK_NOTE(type,action) \
|
../gdk/win32/gdkdrag-win32.c:40: note: this is the location of the previous definition
40 | #define GDK_NOTE(a,b)
Fixes: bc159207bd ("gdk: Drop old debug macros")
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Make `gtk_icon_theme_has_icon` and `gtk_icon_theme_has_gicon` also
consider unthemed icons. This makes their behavior consistent with the
actual (documented) lookup behavior.
Fixes: #5709 and makes the workaround in nautilus@b643a00b obsolete
This is needed for a query-tooltip handler, as mentioned in the
documentation, when there has been a hover timeout.
Maintain the previous behaviour when the link is clicked and follow the
existing documentation regarding selectable labels.
A notify::cursor handler can now also be used to retrieve the URI of the
link under the cursor.
BuilderListItemFactory isn't quite suited for our purposes, primarily
because you can't pass user data to BuilderListItemFactory. Because
we can't get the data we are using a workaround to get the
GtkFileChooserWidget ancestory, which used to work, but with the
recent list view changes no longer doesn't. Use GtkSignalListItemFactory
with the GtkFileChooserWidget as the user data.
It's not enough to sanitize values when starting an animation, as the
adjustment can reconfigure itself while the animation runs.
So as a simple way to handle this, we sanitize every value right before
setting it, too.
In the future we might also want to look at sanitizing start/end values
of the animation.
Fixes#5763
There are a lot of cases where properties are implemented in classes but
the getters for these exist in an interface that class implements.
A common Example is g_list_model_get_n_items() being the getter for
GtkWhateverListModel::n-items.
But also property implementations that don't use override_property()
(usually because they have a different default) are handled by this.
When adding mask nodes, I overlooked that
we have two separate functions for determining
what transforms a node supports without offlines.
Since we claim that mask nodes support general
transform, they must certainly support 2d transforms
as well.
GLES 2.0 version is fine now with current gtk according to B. Otte.
Let's use the same minimum requirement for all implementations.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
When using GDK_DEBUG=gl-egl, we end up using GL, but that is not well supported:
Creating EGL context version 3.0 (debug:no, forward:no, legacy:yes, es:no)
Created EGL context[0000000000000004]
OpenGL version: 0.0 (legacy)
* GLSL version: (NULL)
* Max texture size: -1059701680
* Extensions checked:
- GL_KHR_debug: no
- GL_EXT_unpack_subimage: yes
- OES_vertex_half_float: no
** (gtk4-demo.exe:14324): WARNING **: 19:16:41.468: Compile failure in
vertex shader:
ERROR: 0:7: 'gl_Position' : undeclared identifier
---8<---
Use GLES when EGL implementation is ANGLE.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Buttons under .toolbar were using for their 'hover', 'active'
and 'check' colors the default ones from %button_basic_flat
which are very dimmed, so we explicitly darken them.
Part of #5725
The test doesn't hold 2 references, it holds only one.
The reason one unref can cause a leak is that some backends - like X11 -
only destroy the surface once the DestroyNotify event from the X server
has come in.
X11 does add an extra reference to surfaces that gets released when the
DestroyNotify event arrives.
Wayland doesn't ave such an event, so that reference never gets
released.
This fixes a copy/paste error introduced in commit 590f3dfa1f.
We want to remove the event queue from the list of event queues, not the
surface.
Otherwise the freed queue stays in the list and the next time an event
comes in, we access invalid memory.
Fixes thinko introduced in commit 7fafa5133b.
Luckily, we leak all surfaces, so this problem never occured.
We want to support GLES 2, so make sure we test that support.
Also force-disable common extensions we don't explicitly check for and
don't want to accidentally use.
They're not needed and GLES doesn't technically support them, even
though GTK had been using them via epoxy sneakily using the
GL_OES_vertex_array_object extension behind our back.
Cache the last looked up item and use it for looking up the next item if
it's closest. This massively speeds up iteration over the model, because
each call to get_item() will be adjacent to the previous one.
Improves performance of the inspector quite a bit.
When the variant-editor emits a callback, it might not actually have
edited the value in question. Try to detect that by only emitting
signals if the value changed.
gsk_vulkan_render_download_target() currently resets the uploader
objects before downloading the image that it produces. This is
problematic because there might be unreleased buffers and images
in the command queue.
In particular, this can make validation layers complain about the
glyph atlas - of all things! - upload buffer being released while
still being used by the command queue.
Fix that by resetting the uploader after downloading the image.
For certain kinds of layouts, especially ones where one or both sizes of
a top level is constrained by physical limits, it's acceptable to have
buttons that rely on the minimum size of their contents, rather than the
natural size. It is left to the application authors, or the localization
teams, to ensure that things like translations and font sizes do not
result in a broken UI.
Our webdav server has a root which is davs://mynextcloud/remote.php/webdav
When once creates a GFile out of or out of a subdirectory, and one call
g_file_get_parent(), it recurses too far up and try to query
davs://mynextcloud/remote.php which fails, resulting in a broken pathbar.
To fix that, before querying the metadata of each element of the path,
I query the "enclosing mount", then use it's root to compare the GFile
against.
With the right GMount, we can also fix the icon drawing code in the
pathbar for network drives.
Check if the driver supports MAILBOX and prefer using it; in its
absense, checkif the driver supports IMMEDIATE and prefer using
it; finally, if neither of them are supported, use the guaranteed
to be supported FIFO mode.
Check the portal version number before trying to use
it. Most importantly, this will detect the case where
the interface isn't supported at all, since the proxy
will report a version of 0 in that case.
Fixes: #5733
We want to keep the wl_surface around, because surfaces create their
resources on construct and keep them until destroyed. See the HWND ond
Windows and the XWindow on X11.
This is relevant for graphics resources, where we want to have access
to the VkSurface and eglSurface while the GdkSurface is hidden.
We also want these surfaces to be permanent and not change during the
lifetime of the GdkSurface.
What we can - and must - destroy however are the xdg surfaces, because
those handle visibility on screen.
And we also need to ensure no buffer is attached, so that during the
next creation of the xdg surface we don't get a protocol error.
gdk_wayland_surface_maybe_resize() just calls
gdk_wayland_surface_update_size(), so make all callers call that one
instead.
The check that it does is done by the other function again.
This workaround - were it ever to trigger - is broken today. It destroys
the wl_surface and all associated structs but does not recreate the
xdg_popup or xdg_toplevel struct, so it would cause a hidden window.
The workaround looked a lot different when it was introduced in commit
83b54bab57, too - both in what it did and
in what the vfuncs did that it called.
.view does absolutely nothing in Default style since the whole box is
covered with a GtkNotebook which has its own background, and adds an
unwanted background onto the tab strip in Adwaita.
Some bindings (GJS!) could add temporary references to the GAsyncResult
argument that we return, and thus to the GTask, which may cause the
dialog not to close when the finish function is called (but at garbage
collection instead!).
To prevent this, just manually destroy the window (by removing the task
data), so that we are not bound to the GTask lifetime anymore.
Closes: https://gitlab.gnome.org/GNOME/gtk/-/issues/5741
If we click close enough between lines, and with the maximum distances applied
by GtkGestureClick we could jump between lines when handling double/triple
click for word/line selection.
Ensure that the whole operation stays in the same line and reset the
gesture/counter if we do move between lines, so we start from scratch in the
new line.
The date/time column relies on the filechooserwidget to format the date
properly. During bind, the filechoosercell, get the filechooserwidget
ancestor, but now due to changes in the listview, the cell isn't a
child of the filechooserwidget at that point. Since this is deeply
ingrained into the filechooserwidget, let's keep the same behavior,
but move it to filechoosercell in realize. Alternatively, we could have
used a signal factory (with the file chooser widget as the user data),
but that would have been a major overhaul.
The format of the type column depends on the the type_format, which
is stored in the filechooserwidget. We get that setting by looking
for the filechooserwidget ancestor, which no longer works after recent
changes to the list views (it was fragile to begin with). At one point,
the setting appears to have been dynamic, but now it is only loading
from GSettings, so let's simply do the same within FileChooserCell.
32247bc50e made several changes to account for the
fact that we no longer have a NULL editable at the beginning of the list
model. The commit mistakenly left out one change in remove_file(),
which causes the wrong file to be removed.
Now that the paint demo lets us test this, it has
become apparent that this condition is wrong, and
we don't get the expected events if stylus-only is
FALSE.
The current implementation of the glyph cache deals with atlases by
padding them with 1 pixel at the beginning, at the end, and between
each glyph.
That's cool and all, however, there's a very subtle problem with
this approach: the contents of the atlas are garbage, so this padding
is filled with garbage memory!
Rework the Vulkan glyph cache to draw each and every glyph in a
surface that has 1 pixel border of padding around it. Ensure the
surface is completely black by drawing a rectangle before handing
it to Pango to draw the glyph. Update tx and ty to pick the texture
position adjusted to the 1 pixel padding. The atlas now starts at
position (0, 0), since each glyph individually contains its own padding.
To improve legibility, add a PADDING define and use it everywhere.
Vulkan renders text using VK_BLEND_FACTOR_ONE_MINUS_SRC_ALPHA and
VK_BLEND_FACTOR_SRC_ALPHA, but that implies per-channel alpha
blending, which currently produces the wrong results when blending
glyphs with the images beneath them.
Use the default pipeline constructors, which implies using the
ONE and ONE_MINUS_SRC_ALPHA.
When determining double-clicks, don't use the distance
threshold for touch events. It is very hard to double
touch reliably within a few pixels of the same position.
Fixes: #5580
Typically, a popover gets mapped when shown and unmapped when
hidden. A situation there that breaks is where the popover gets
recursively unmapped/unrealized when its root is destroyed.
In that situation, the popover does however unmap (without being
hidden first), moving the GTK grab from show/hide to map/unmap
will handle the previous situations, plus this one.
Fixes things being unclickable if e.g. a modal dialog got a
popover popped up, then got closed via Alt-F4.
Closes: https://gitlab.gnome.org/GNOME/gtk/-/issues/5730
we were not checking the return gboolean of
gtk_action_muxer_query_action() which was
returning FALSE for the crash case, meaning
it didn't set the passed in GVariant, but
we were still using it as it was non-null.
Fixes#5729
Calling gtk_list_box_remove_all() is a no-op with a bound model; after
the introduction of the remove_all() method in 49e56fc7, we were left
with row widgets after the dispose() call chained up.
We could restore the explicit unparenting inside dispose() instead of
calling remove_all(), but since the bound list model is provided by the
user of GtkListBox, it's more appropriate to unbind it in the dispose()
implementation, to avoid any potential reference cycle (especially in
higher level languages that have no explicit reference acquisition).
We clean up the bound model, and its associated state, if any; and then
we remove all the row widgets that are left.
Basically what GL does, but without any debug or feature flag
to gatekeep it, since the Vulkan backend itself is experimental
already.
Ceil surface sizes, and floor coordinates, to the fractional scale
value.
The rects passed to the clip region are in buffer coordinates, and
must not be scaled. Consider the following scenario: Wayland, with
a 1024x768@2 window. That gives us a 2048x1536 raw image. To setup
the Vulkan render pass code, we'd scale 2048x1536 *again*, to an
unreasonable 4196x3072, which is (1) incorrect and (2) really
incorrect and (3) can lead to crashes at best, full GPU resets
at worst - and a GPU reset is incredibly not fun!
Now that we pass the right clip regions at the right coordinates
at all times, remove the extra scaling from the render pass.
This part of the Vulkan renderer is almost exactly equal to the GL
renderer, and the GL renderer already does that since at least
2a38cecd33. Copy that into the Vulkan renderer.
A nice side effect from this commit is that resizing a window now
actually works again.
Sneak in a trivial cleanup by using a variable to hold the draw
index.
This was a tricky one to figure out, but it's pretty simple to
understand (I hope!).
So, this AMD card I'm using requires buffer memory sizes to be
aligned to 16 bytes. Intel is aligned to 4 bytes I think, but
AMD - or at least this AMD model in particular - uses 16 bytes
for alignment.
When creating a a particular texture (I did not determin which one
specifically!) a buffer of size 1276 bytes is requested.
1276 / 16 = 79.75, which is clearly not aligned to the required
16 bytes.
We request Vulkan to create a buffer of 1276 bytes for us, it
figures out that it's not aligned, and creates a buffer of 1280
bytes, which is aligned. The extra 4 bytes are wasted, but that's
okay. We immediately query this buffer for this exact information,
using vkGetBufferMemoryRequirements(), and proceed to create actual
memory to back this buffer up.
The buffer tells us we must use 1280 bytes, so we pass 1280 bytes
and everyone is happy, right? Of course not. We pass 1276 bytes,
and Vulkan is subtly unhappy at us.
Fix that by passing the value that Vulkan asks us to use, i.e.,
the size returned by vkGetBufferMemoryRequirements().
This is what GL does, and for a reason: it can lead to width or
height for very small glyphs. Also, switch to dividing by a float
(1024.0) instead of an integer (1024).
This doesn't make any difference now, but will allow us to copy
subregions more easily. This is not obvious, but here's a quick
explanation:
Leaving 'bufferRowLength' and 'bufferImageHeight' implies that
Vulkan will assume the size passed in the 'imageExtent' field.
Right now, this assumption is correct - the only user of this
function is the glyph cache, and it only copies and uploads
exact rects. Next commits will change that assumption, so we
must pass 'buffer*' fields, and tell Vulkan, "this part of the
buffer represents an image of width x height, and I want the
subregion (x, y, smallerWidth, smallerHeight) of this image".
When creating an image using gsk_vulkan_image_new_for_framebuffer(),
it passes VK_IMAGE_LAYOUT_COLOR_ATTACHMENT_OPTIMAL.
However, this is a mistake. The spec demands that the initial
layout must be either VK_IMAGE_LAYOUT_UNDEFINED or
VK_IMAGE_LAYOUT_PREINITIALIZED.
Apparently this was an oversight from commit b97fb75146, since the
commit message even documents that, and all other calls pass either
VK_IMAGE_LAYOUT_UNDEFINED or VK_IMAGE_LAYOUT_PREINITIALIZED.
Create framebuffer images using VK_IMAGE_LAYOUT_UNDEFINED, which is
what was originally expected.
Removing all items from containers is a common use case.
Without this applications needed to implement this manually.
It makes senses to handle it here.
Fractional scaling with the GL renderer is
experimental for now, so we disable it unless
GDK_DEBUG=gl-fractional is set.
This will give us time to work out the kinks.
This commit combines changes in the Wayland backend,
the GL context frontend, and the GL renderer to switch
them all to use the fractional scale.
In the Wayland backend, we now use the fractional scale
to size the EGL window.
In the GL frontend code, we use the fractional scale to
scale the damage region and surface in begin/end_frame.
And in the GL renderer, we replace gdk_surface_get_scale_factor()
with gdk_surface_get_scale().
Whenn setting gtk_builder_set_allow_template_parents(), the builder
instance will accept
<template class="GtkWidget">
for a GtkBox instance.
It's going to be used with the new GtkColumnViewCell objects, so that
it's backwards compatible with ui file factories that use GtkListItem.
That way, local scrolling is available and the scrolling isn't random.
Recycling should now involve reordering the recycled widgets instead of
just keeping their order because all of them got recycled.
This allows setting a factory to toggle per-row properties.
Implemented are selectable, focusable and activatable.
These are meant to supercede the per-cell selectable and activatable
properties, which make no sense individually.
The focus property makes it possible to focus rows instead of cells,
which is the default behavior.
There is no way to set it yet, this is just to prove that it works.
It also changes the focus behavior of rows. They are now always
focusable - unless turned off by the factory once that is possible.
This makes the question if a listitem can be focused or not an explicit
decision by application developers.
Previously an item could be focused if it was selectable and no child or
grandchild was focusable - so if you put a label and icon into it, the
item was focusable, but if you put a GtkTreeExpander or a GtkButton into
it, the item wasn't. This needs to be decided explicitly now.
Technically this is an API break, because the previous behavior does not
exist anymore.
But I really don't want to make this a tristate (focusable, not
focusable, automatic), because then binding it to other things gets
hard, and because all the other focusable proeprties are booleans, too,
and working with them gets a lot harder.
Related: #3910
Cairo can do that, so just enable it:
* Create surfaces with the correct fractionally scaled size.
* Set the Cairo surface's device scale to that number.
Instead of setting the buffer scale via the buffer-scale command, set it
via the viewport.
This technically allows setting fractional scales, but we're not doing
that.
Instead of tracking a single scale, track x and y scales separately.
Factor out gsk_vulkan_render_pass_new() into a private function that
receives both scales, and pass 'scale_factor' for both.
April fools!
No, really.
The fractional scale protocol is just a way to track the surface scale,
but not a way to draw fractional content.
This commit uses it for that, so tht we don't rely on tracking outputs.
This also allows magnifiers etc to send us a larger (integer) scale if
they would like that, that is not represented by the outputs.
This is mostly a cosmetic change, and the goal is twofold:
1. Make it easier to spot unimplemented render node types; and
2. Prepare for a small rework
The implementation for each node now lives in specific functions,
like the GL renderer; unlike the GL renderer, however, we use a
node type vtable to map GskRenderNodeType → implementation. Render
node without an implementation map to NULL, and use the fallback
implementation. Render nodes that fail any check and return FALSE
also use fallback implementation.
The scrolling code assumes the adjustment values are up to date or
it crashes and before we've run size_allocate() we haven't update them.
Fixes a crash in the gtk-demo scrollinfo that would set the adjustments
with random values (via ScrolledWindow.set_child()) and then scroll in a
tick callback right before the (first) size_allocate().
The Lunarg validation layers seem to have been deprecated in favour
of the Khronos ones. There's no reason not to have both, to accept
loading both - simultaneously, even.
Instead of passing a single, potentially massive rectangle that is
just the extents of the damage rect, collect and pass all damage
rects individually.
Add a new flag to track whether buffer scale is dirty or not,
and centralize calling wl_surface_set_buffer_scale() in a single
place: gdk_wayland_surface_sync_buffer_scale().
gdk_wayland_surface_sync_buffer_scale() is only called by
gdk_wayland_surface_sync(), which itself is called by the GL,
Vulkan, and Cairo contexts, right before submitting a frame.
This ensure that each frame has an up-to-date buffer scale.
This mimics how opaque and input regions are tracked.
According to the at-spi2 docs, for a widget to be considered visible,
it needs both the showing and visible states. Many applications rely on that,
for example the flat review functionality of Orca.
this fixes#5194
Don't fudge around poking through the listview, trying to get a model
and selecting it directly. Instead, use the proper way and activate the
"listitem.select" action.
Instead of directly calling select_item(), trigger the select-item
action of the focused child.
We do this convoluted calling into the widget because that way
GtkListItem::selectable gets respected, which is what one would expect.
Plus, this code is usually triggered via keybindings, and this way the
ListBase keybindings work identical to the ListItem keybindings.
If we encounter a node or texture the 1st time and they are going
to be used again, give them a name.
Then, when encountering them again, print them by name instead
of duplicating them.
We extend the syntax for nodes from:
<node-type> { ... }
to
<node-type> { ... }
<node-type> <string> { ... }
<string>;
where the first is the same as before, the 2nd defines a named node and
the last references a previously defined node.
Or to give an example:
color "node" {
bounds: 0 0 10 10;
color: red;
}
transform {
bounds: 20 0 10 10;
child: "node";
}
This will draw the red box twice, once at (0,0) and once at
(20,0).
The intended use for this is both shortening generated node files as
well as allowing to write tests that reuse nodes, in particular when
dealing with caches.
We extend the syntax for textures from just:
<url>
to
[<string>] <url>
<string>
where the first defines a named texture while the second references a
texture.
Or to give an example:
texture {
bounds: 0 0 10 10;
texture: "foo" url("foo.png");
}
texture {
bounds: 20 0 10 10;
texture: "foo";
}
This will draw the texture "foo.png" twice, once at (0,0) and once at
(20,0).
The intended use for this is both shortening generated node files as
well as allowing to write tests that reuse textures, in particular when
mixing them in texture and texture-scale nodes.
If we map, reposition, unmap, remap, the reposition feedback from the
last time a popup was mapped might be received while we're dealing with
the new version of the popup. At this point, the old reposition token
has no meating, so lets drop it. Also reset the reposition tokens when
creating new protocol objects, so that the reposition token are as if
we're in the initial state.
This fixes an issue where we'd get stuck if repeatedly smashing a button
that'd create popups that'd immediately get dismissed by the compositor.
Since Wayland 1.15, it is now possible to use absolute paths in
"WAYLAND_DISPLAY".
In that scenario, having a valid "XDG_RUNTIME_DIR" is not a requirement
anymore.
For this reason we remove the "XDG_RUNTIME_DIR" check and we let
`wl_display_connect()` decide if our environment is correct.
Signed-off-by: Ludovico de Nittis <ludovico.denittis@collabora.com>
An inert gridview is a gridview that does not use the factory. This
allows faster updates because no calls into user code need to happen.
A gridview is inert when either:
- It is not rooted.
- It is not visible.
- No factory is set (that one is obvious)
The gridview does not need to be inert without a model, as that case is
handled by the item manager.
This should allow Nautilus to keep both the gridview and the columnview
around, and just gtk_widget_hide() the unused widget.
The code for now does not disable the item manager, as some
functionality of the item manager is required to allow setting scroll
positions and such.
But that is a place where more gains could be found if profiling showed
that was useful to do.
An inert listview is a listview that does not use the factory. This
allows faster updates because no calls into user code need to happen.
A listview is inert when either:
- It is not rooted.
- It is not visible.
- No factory is set (that one is obvious)
The listview does not need to be inert without a model, as that case is
handled by the item manager.
This should allow Nautilus to keep both the gridview and the columnview
around, and just gtk_widget_hide() the unused widget.
The code for now does not disable the item manager, as some
functionality of the item manager is required to allow setting scroll
positions and such.
But that is a place where more gains could be found if profiling showed
that was useful to do.
The widget would teardown the factory on unroot to avoid unnecessary
work when it isn't shown.
However, recycling may reposition widgets, and repositioning widgets
does a unroot/root.
We don't want to recycle widgets then.
The implementation lives (as always) in GtkListBase.
This is a feature request from the Nautilus developers, who currently do
some hacks to emulate that behavior and it apparently only breaks
sometimes.
We connect gtk_scrolled_window_update_use_indicators
as signal handler in realize(), but we were disconnecting
gtk_scrolled_window_sync_use_indicators in unrealize.
Spotted by Milan Crha.
Fixes: #5684
Just like GdkToplevel::compute-size, the size argument of the signal is
given to the handlers by GDK; it's not an out argument meant to be
allocated by the caller.
The size argument is passed to the signal by the GDK surface machinery,
as is: it's not going to be allocated by the caller (since it's a
signal), and it's not an out argument.
The cursor-theme-size setting is documented as
'0 means the default size'. Make it so by using
size 24 if we see a 0. Its better than crashing.
Fixes: #5700
The function is going away, and the computation
here was wrong anyway. Instead, add a helper that
properly computes the pointing-to rect in surface
coordinates and use it everywhere.
The widget paintable uses the widgets bounds
as intrinsic size, so we need to offset from
that to the allocation, which is what the
coordinates are relative to.
Text handles had the same problem as popovers.
They were interpreting their pointing-to rectangle
relative to the widgets bounds, when it is meant
to be relative to the widgtets allocation.
While we touch this code, rewrite it to use
gtk_widget_compute_point.
When we don't have a pointing-to rectangle, we want to place
the popover wrt to the parents bounds. But if we have a
pointing-to rectangle, it is relative to the widgets allocation,
which is different from the bounds.
We were not handling the second case correctly, leading to context
menus in the text view being mispositioned by the widgets CSS padding.
While we are touching this code, rewrite it to handle transforms.
Fixes: #5695
Since we are making GdkGLContext call the core wgl*() functions directly
instead of via libepoxy, drop the workarounds that we needed for notifying
libepoxy that wglMakeCurrent() outside of GDK/GTK was called.
This way, we clean up the code, and as a result, we can use the GstGL
APIs like the other platforms to query what GL api that is to be used.
For ensuring that things work between different threads, we now call
gdk_gl_context_clear_current() in place of calling wglMakeCurrent(xxx,
NULL), so that we make sure that there is no current GL context on a
thread outside of GstGL's thread, which Windows does not like.
We might be dealing with GL contexts from different threads, which have more
gotchas when we are using libepoxy, so in case the function pointers for
these are invalidated by wglMakeCurrent() calls outside of GTK/GDK, such as
in GstGL, we want to use these functions that are directly linked to
opengl32.dll provided by the system/ICD, by linking to opengl32.lib.
This will ensure that we will indeed call the "correct" wgl* functions that
we need.
This should help fix issue #5685.
When the GL texture already has a mipmap, we don't
have to download and reupload it to generate one.
We differentiate the handling for texture scale nodes,
where we do want to force the mipmap creation even if
it requires us to reupload the GL texture, and plain
texture nodes, where we just take advantage of a
preexisting mipmap to allow trilinear filtering for
downscaling, or create one if we have to upload the
texture anyway.
Make GdkGLTexture determine if the texture has
a mipmap, and provide private API to query this
information.
This check is done in gdkgltexture.c instead of
gskgldriver.c, since we're already binding the
texture here for other reasons, so it is easy
to query a few more things.
- grab_focus() on a row (happens with scroll_to()) keeps the focus
column intact if possible.
- <Tab> and <Shift-Tab> move through the cells in order, and move
to the next row when at the end.
- <Up> an <Down> move to the next/prev row, keeping the same column
focused.
- <Left> and <Right> move to the prev/next cell, if one is available.
If not, they stay where they are.
Make it move focus just like GtkWindow would.
Otherwise the listview will (try to) handle it and move focus between
cells - which doesn't do anything for lists and only works with grids.
This is a split of GtkListItemWidget into the generic parts of factory
using widgets.
On top of it there's GtkListItemWidget, which takes care of GtkListItem.
They're not used outside of GtkListBase, so no use to have them in the
header.
Requires moving one function up in the source now that the forward
declaration is missing.
create_at_context was confused - it stored a reference
to the newly created context in priv->context, but then
also returned a reference, and the caller stored that
in priv->context again.
Change it to only return a reference.
Fixes: #5690
Store texture coordinates for each slice
instead of assuming 0,0,1,1, and generate
overlapping slices to allow for proper mipmaps.
This almost fixes trilinear filtering with
sliced textures.
This one exhaustively tests reusing the same model as a child model for
many nodes.
This tracks that multiple items-changed signals emitted at the same time
(or multiple handlers for one such signal) doesn't put the treelistmodel
in an inconsistent state while it is handling all of them.
I'm not sure this (ab)use of treelistmodel should be officially
supported, but it works today, so let's test it to see if we can keep it
working.
If there is no other widget in the group that we can focus, don't focus
and activate ourselves.
Otherwise the arrow keys on checkbuttons toggle the checkbutton.
If a GtkImage is using an icon we use a gtk-icon-filter to se the icon
opacity when in insensitive state, however when using other kinds of
pictures we do not apply the same style leading to an inconsistent
result.
Closes: #5683
Items should be cleared when the node is discarded, not when the node's
children are discarded - which can also happen when a node is collapsed.
Fixes an error introduced in 9048e391b6Fixes#5681
In constrast to our other tests, these use
textures that are big enough to force slicing
with setting GSK_MAX_TEXTURE_SIZE, which we
will use in the following commits to improve
test coverage.
It is useless to have node files with references
to external files in the testsuite, so turn such
textures into data urls by doing a serialization
roundtrip.
No user knows that we have an internal function called
gtk_tree_list_row_destroy() that gets called when a row gets removed
from the treelistmodel.
So everyone was probably just making stuff up about what "destroy"
means.
Related: #5646
That way, we can return the item even after the row is removed. This is
particularly relevant in ListItemFactory::unbind callbacks because they
often use gtk_tree_list_row_get_item() and user code never tracks
changes to this property.
A side effect of this is that the item will survive until the row gets
destroyed, but that's what users expect anyway, so we can live with it.
Related: #5646
This is a good idea to avoid reentrancy problems when any child model
(or potentially more than one child model) has started emitting
items-changed but the emission hasn't arrived in this model yet.
At tat point, we'd get_item() the wrong item from those models.
We want to avoid such cases of reentrancy.
Related: #5646
We cheat and just set the texture parameters instead and hope nothing
explodes.
So far it didn't.
This is only needed to support GLES 2.0 so it's quite a limited set of
hardware these days.
Instead of uploading a texture once per filter, ensure textures are
uploaded as little as possible and use samplers instead to switch
different filters.
Sometimes we have to reupload a texture unfortunately, when it is an
external one and we want to create mipmaps.
When filtering changes for an already-cached
texture, we need to clear the render data
before setting the new one, otherwise it
does not take and we end up reuploading
the texture every frame.
Code above ensures that i is always in [0, n_columns - 1] range, so
the condition was always true, which resulted in filler tile always
being added to the grid. As the result, an empty row appeared at the
end of the grid if the number of columns divided the number of items.
Only add filler tile if last row is not full, i.e. when i > 0.
a11y: Fix the logic in gtk_accessible_get_next_accessible_sibling which decided whether we will use the overridden sibling on the context.
See merge request GNOME/gtk!5659
We were culling children based on the content box, but clipping via
overflow happens on the padding box, so we need to use that one instead.
Fixes issues with items not being visible / disappearing in Nautilus
when they are near the border.
Resolves#5380
The GtkUriLauncher calls into the openuri portal, which distinguishes
between files, directories, and URI. The GtkFileLauncher contains logic
to deal with this, because it can already handle the file and folder
differences.
If we have a file:// URI it's easier to create a GFile out of it, and
use the GtkFileLauncher API, while leaving the GtkUriLauncher API for
every other URI scheme.
Same fix as de3c1d0c73, for GtkLabel.
Fixes: #5671
The GtkUriLauncher calls into the openuri portal, which distinguishes
between files, directories, and URI. The GtkFileLauncher contains logic
to deal with this, because it can already handle the file and folder
differences.
If we have a file:// URI it's easier to create a GFile out of it, and
use the GtkFileLauncher API, while leaving the GtkUriLauncher API for
every other URI scheme.
Fixes: #5671
Otherwise GL surfaces that redraw without changing the hotspot have it
applied on top every frame and quickly slide away.
The cairo path and the X11 backend do not have this bug.
The GL Wayland drag surface code path has a bug where it does not reset
the hotspot, so if a GL-backed draw surface redraws without resizing or
resetting the hotspot, it moves away. The next commit will fix that, but
this commit adds a test for that.
Allow to set max texture size using the
GSK_MAX_TEXTURE_SIZE environment variable.
We only allow to lower the max (for obvious
reasons), and we don't allow values smaller
than 512 (since our atlases use that size).
GdkDragSurface-backed widgets are not parented to an existing widget,
unlike popovers, and like toplevels. This means that there's nobody to
actively call gdk_drag_surface_present() to update the size, and
GdkDragSurface should do it on its own, just like GdkToplevel.
This commit implements this for the Wayland backend.
Compute our size when requested by the backend. This makes GtkDragIcons
actually recompute their size when it changes, instead of getting stuck
with the first size and potentially underallocating.
Similarly to GdkToplevel, GdkDragSurface's compute-size should be called
by backends to query the current surface size, and should be connected
to by widget implementations (like GtkDragIcon) to report the current
size.
GdkDragSurface-backed widgets are not parented to an existing widget,
unlike popovers, and like toplevels. This means that there's nobody to
actively call gdk_drag_surface_present() to update the size, and
GdkDragSurface should do it on its own, just like GdkToplevel.
When fatal warnings were turned on, the developer would never see which
widgets were left as children to the widget that triggered the warning as
those were printed in separate g_warning calls.
Print a single warning with all the info so runs with fatal warnings
aren't left without any info.
For whatever reason, meson decides to use custom
target names in the file system, and on Windows,
the ':' is causing trouble here. So avoid it.
Fixes: #5280
This allows dropping or copy/pasting rendernodes into apps that accept
SVGs.
Not sure how useful this is because we advertise text/plain from
rendernodes already and we prefer that.
Doing it on hide() is not enough, since in some edge cases we didn't
ever actually map, we just attempted to compute the size, e.g. in
response to a ConfigureNotify event, then the window was destroyed.
Related: https://gitlab.gnome.org/GNOME/mutter/-/issues/2678
... and make the tile finding code use distance.
This also changes how gtk_list_item_manager_get_tile_at() finds the
right tile, so this is a custom commit for bisectability.
gtk_list_item_manager_get_nearest_tile() isn't used yet.
This way, listview and gridview don't need to check if the rect is out
of bounds and nothing is selected, a quick rectangle_intersect() does
the job for them.
Just get the position right and give them a height of 0px, that should
be good enough.
If we don't do that, code will think the item doesn't exist, which is
not what we want.
If the `GtkRecentInfo` represents a directory, simply use it, and
do not try to find its parent in `_gtk_file_chooser_extract_recent_folders()`.
For example, there is an entry in my recently-used database
from the Amberol music player about the folder I have opened
with it, but the folder is not listed on the "Recent" tab of
the file chooser widget, only its parent. After this change,
the directory itself is shown.
Native widgets get allocated via their surface,
so can skip them here. This avoids criticals when
re-mapping a popover for the second time, as can
be seen e.g. in the 'Selections' demo in gtk4-demo.
allocate() should not be calling into ensure_allocate(), they do a similar job.
In the end, the code does the same work, but it should be easier to follow now.
Currently the GtkSearchEngine is torn down every time the search
is stopped, which also means between typed characters. This
prevents any of the optimizations that the GtkSearchEngine can
do in the long run.
Let the GtkSearchEngine stay around for longer, and only be
disposed after search is cancelled, the filechooser moves
onto a different mode than search, or is otherwise unmapped/disposed.
While at it, remove an unused struct field.
Again on massive filesystems, the very first character
is likely to bring a likewise massive amount of search
results that we need to maybe query info for, then create
icons and widgets for. While it's impressive we can do
that, it's also expensive and likely pointless, for the
first character.
Typing a second character is however very likely to
considerably reduce the amount of items to categorize and
show. So start actually searching from there.
Testing on a filesystem with 1434099 files indexed, trying 5
semi-random 1 character searches (n, h, t, i, o) returns on
average 168K items (min. 78771, max. 331471), trying 5
semi-random 2 character searches (no, he, th, in, on)
returns on average 34K items (min. 11133, max. 94961),
which is a more approachable set.
Doing this is enough that typing on a filechooser search
entry feels completely fluid.
The search provider should make it sure there are some
specific GFileInfo fields set. Fix the mimetype extraction
from the query, and use that to fill in the missing gaps
the best we can.
When starting a search over a very populated filesystem, it
is possible that typing the first chars will return a too
high number of results. Even though iterating through the
cursor is in itself very fast, extracting the GIO information
from those many files at once is not going to be as fast.
In order to increase interactivity (i.e. not make things
possibly sluggish) iterate the cursor in an idle function
and add search results to the filechooser model little by little.
If the user keeps typing (as it is likely will happen), there
will be better chances to cancel and proceed to the next
query timely. If not, the results will be there soon enough.
This state is used for visited link-like widgets.
It has no ARIA equivalent, e. g. can not be set programmatically, but it
exists in the browser environment as well.
Error out if introspection is requested,
but g-ir-scanner isn't found.
And if introspection isn't explicitly disabled
but is required for building the docs, build it.
As fancy as property paths are, recursive resolution of files
to a location increases the big O complexity enough that it's
not a great option on large homedirs with many indexed files.
Ensure the files are from the right location through a URI
prefix match, which does hits an index. This may dramatically
improve performance on large indexed trees.
Testing this query in an isolated testcase with a total
1434099 indexed files shows that it can run more than 1500 times
per second in this computer (an average of 15200 queries in
several 10 second runs), which presumably is a tad faster than
anyone can type.
Closes: https://gitlab.gnome.org/GNOME/gtk/-/issues/4133
In certain scenarios, address the issue where gnome.compile_resources
fails to transmit the present source directory. This is most notably
visible with MSBuild.
width-request already ensures it's above the minimum width, so avoid an
extra queue_resize() when setting size request to (-1, -1).
This is the same way as GtkDropDown works. This also unbreaks
GtkComboBox after the recent allocation fix in
75a417e337.
Incidentally, this also makes GtkComboBox actually resize its popup as
intended (that was broken before).
I don't think this is ultimately the final fix, sometimes I still get
allocation warnings. But the proper fix will probably involve changing
some more allocation machinery around popovers. This is good enough for
now.
All the other signal handlers are connected in
realize and disconnected in unrealize, but the
::compute-size handler was forgotten.
This was notices in !5597.
The current definitions of the g_io_module_*() symbols do not build on
Visual Studio when building against GLib earlier than 2.75.0 due to the
way how these symbols are decorated in the GLib headers, as Visual Studio
does not allow symbols that were previously marked with 'extern' (or so)
to be marked with anything that is symantically different later.
As a result, if we are using Visual Studio and glib-2.74.x or earlier,
override _GLIB_EXTERN as appropriate in the modules/media sources before
including the GIO headers. This sadly, means that we need a
configure-time check as it would have been too late if we checked the
GLib version using G_VERSION_CHECK macro, as the GIO headers would have
been included already.
There are similar items in the print backends, but we will not attempt
to update these files as they are not meant to be built for Windows.
In derivable classes, the widget's class can be different from the one
dispose_template() was called for, which can lead to failing the
template != NULL check at best, undefined behavior at worst.
Since we already pass the correct GType into the function, just use that
instead.
The problem here is that new windows appear in the list before the
window's dispay gets set and we don't update the filter when the
display changes (would need watches support for the filtermodel).
So add this somewhat hacky method.
The split-up of gdksurface-wayland.c introduced a protocol violation
when it didn't make sure xdg_surface was destroyed after the role
objects (xdg_popup / xdg_toplevel). Fix that.
Fixes: 2a463baed0 ("wayland: Rearrange the surface code")
Don't misinform the observing listmodel that CSS nodes were removed that
weren't actually removed, but just moved. Otherwise the observer would
think it has run out of items when it really hasn't.
That stupid space in the bottom right when n_items isn't a multiple of
n_columns needs its own tile, or we'll get errors about not finding a
tile.
So make one.
Otherwise, when removing the columns, each column will trigger a
sorter::changed signal emission.
And because sorters are often still connected to a sortlistmodel, we
can't skip that emission and need to do it.
But we only need to do it once.
The previous check does not longer work.
When a model gets all items deleted, there will still be existing tiles
until the next time garbage collection is run.
So do that before checking if the list is empty.
Instead of making it 2 vfuncs for getting horizontal and vertical area,
make it one vfunc to get the area.
Also rewrite the implementations to use the tile's area instead of
trying to deduce things with fancy math.
Instead of randomly changing tiles, the listitemmanager gains a split
vfunc that listview and gridview implement so they can keep their tile
areas intact. The listitemmanager will now conform to these rules:
1. Never delete a tile.
This ensures that all areas stay intact.
2. Never change the n_items of a tile other than setting them to 0.
This causes "empty" areas to appear, but listview/gridview can
easily check for them by checking for tile->n_items == 0.
gtk_list_tile_gc() will get rid of them.
3. Adding items always creates new tiles that are added with empty area.
That way they don't interrupt any existing machinery until the next
allocation.
4. Adding/removing widgets has no effect on areas
This is useful in particular when scrolling where new widgets are
moving between tiles. When the manager moves the widgets, it may
split some areas, but will not remove any existing tiles, so the
whole area stays intact and the list can deal with further scroll
events before an allocation.
This improve the situation for #3334
Instead of the custom size property, use the new tile size.
Also introduce the ability to split tiles, so that gridview can split a
layout that would look like (question mark denoting cells without a
widget, which in this case would be a single tile)
█ █ █ ? ?
? ? ? ? ?
? ? ? ? ?
? ? ?
into 3 rectangular tiles like so:
█ █ █ A A
B B B B B
B B B B B
C C C
This of course also means we need to be able to merge those tiles again
when cells got added/deleted or the gridview was resized. For that job,
gtk_list_tile_gc() exists now, which removes tiles without items and
merges adjacent tiles without widgets.
... and use it to handle ListView allocations.
Nothing spectacular, just proof of concept.
The code introduces the idea that every tile stores its area (others
would call it "allocation", but I avoided that because tiles aren't
widgets). This should allow moving lots of code into gtklistbase.c and
not require special handling inside ListView and GridView.
And that in turn hopefully makes it easier to add more features (like
sections and so on.)
* Instead of using a gpointer to refer to it, use the GtkListTile type.
* Use gtk_list_tile_get_foo() instead of
gtk_list_item_manager_get_tile_foo() naming.
GLib 2.75 started checking if a GFileInfo was created with the attribute
we're querying, instead of failing silently and leaving us in an
inconsistent state.
Turns out that GtkFileChooserWidget, GtkFileSystemModel, and GtkPathBar
trip the newly introduced check.
The GL renderer was creating sripes for nodes that were scaled in
particular ways, probably due to rounding errors.
This testsuite focuses on one of those stripes to make sure they are
gone.
This test fails if we naively create fullscale
intermediate offscreens. This was fixed in the
previous commits.
This tests the fixes in 22ba6b1f33 (for
cairo) and 3a0152b65f (for GL).
Use the same approach and only create an offscreen
that is big enough for the clipped part of the scaled
texture.
If the clipped part is still too large for a single
texture, we give up and just render the texture without
filters (using the regular texture rendering code path
which supports slicing).
The following commit will add the texture-scale-magnify-10000x
test which fails without this fix.
Scale nodes can use large scale factors and we don't want to create
insanely huge Cairo surfaces.
A subsequent commit will add the texture-scale-magnify-10000x
test which fails without this fix.
Cairo surfaces are created transparent.
And even if they weren't, overdrawing with transparency wouldn't erase
what's in the surface because it's a no-op.
It would require CAIRO_OPERATOR_CLEAR or CAIRO_OPERATOR_SOURCE.
GtkAccessible implementations in C can get away returning objects just
by shuffling pointers around, but higher level languages prefer using
full ownership transfer in virtual functions.
Fixes: #5615
This reverts commit 40d4441fd8.
The accessible parent of the child widget in a GtkStackPage is cleared
when the GtkATContext gets disposed, so we don't need to unset it
ourselves. This also avoids a temporary vivification of the GtkATContext
during dispose.
If the early return path in `emit_property_changed()` is taken, and
`value` is floating, it will be leaked. Fix that by sinking `value` on
entry to the function.
Spotted by asan:
```
Direct leak of 128 byte(s) in 2 object(s) allocated from:
#0 0x7f44774ba6af in __interceptor_malloc (/lib64/libasan.so.8+0xba6af)
#1 0x7f44764c941a in g_malloc ../../source/glib/glib/gmem.c:130
#2 0x7f44764f6d8a in g_slice_alloc ../../source/glib/glib/gslice.c:252
#3 0x7f447654655d in g_variant_alloc ../../source/glib/glib/gvariant-core.c:565
#4 0x7f447654664c in g_variant_new_from_bytes ../../source/glib/glib/gvariant-core.c:608
#5 0x7f4476536ed5 in g_variant_new_take_string ../../source/glib/glib/gvariant.c:1307
#6 0x7f4475c75ada in gtk_at_spi_context_state_change ../../source/gtk4/gtk/a11y/gtkatspicontext.c:1112
#7 0x7f44758ee194 in gtk_at_context_update ../../source/gtk4/gtk/gtkatcontext.c:694
#8 0x7f44758dbfcf in gtk_accessible_update_property ../../source/gtk4/gtk/gtkaccessible.c:326
#9 0x7f4475b5abe3 in gtk_widget_set_tooltip_text ../../source/gtk4/gtk/gtkwidget.c:9740
#10 0x58439d in gs_updates_page_update_ui_state ../../source/gnome-software/src/gs-updates-page.c:302
#11 0x5857dc in gs_updates_page_set_state ../../source/gnome-software/src/gs-updates-page.c:403
#12 0x5879f1 in gs_updates_page_load ../../source/gnome-software/src/gs-updates-page.c:636
#13 0x58822d in gs_updates_page_reload ../../source/gnome-software/src/gs-updates-page.c:678
#14 0x50ff48 in gs_page_reload ../../source/gnome-software/src/gs-page.c:731
#15 0x5491ce in gs_shell_reload_cb ../../source/gnome-software/src/gs-shell.c:830
#16 0x7f4477363f54 in g_cclosure_marshal_VOID__VOID ../../source/glib/gobject/gmarshal.c:117
#17 0x7f447735e0ad in g_closure_invoke ../../source/glib/gobject/gclosure.c:832
#18 0x7f4477391f3f in signal_emit_unlocked_R ../../source/glib/gobject/gsignal.c:3802
#19 0x7f4477390c13 in g_signal_emit_valist ../../source/glib/gobject/gsignal.c:3555
#20 0x7f4477391324 in g_signal_emit ../../source/glib/gobject/gsignal.c:3612
#21 0x7f447705b3c3 in gs_plugin_loader_reload_delay_cb ../../source/gnome-software/lib/gs-plugin-loader.c:1538
#22 0x7f44764bd140 in g_timeout_dispatch ../../source/glib/glib/gmain.c:5054
#23 0x7f44764b9eb1 in g_main_dispatch ../../source/glib/glib/gmain.c:3460
#24 0x7f44764bb72c in g_main_context_dispatch ../../source/glib/glib/gmain.c:4200
#25 0x7f44764bba15 in g_main_context_iterate ../../source/glib/glib/gmain.c:4276
#26 0x7f44764bbbfa in g_main_context_iteration ../../source/glib/glib/gmain.c:4343
#27 0x7f44769ef655 in g_application_run ../../source/glib/gio/gapplication.c:2589
#28 0x4f2da5 in main ../../source/gnome-software/src/gs-main.c:49
#29 0x7f4474e4a50f in __libc_start_call_main (/lib64/libc.so.6+0x2750f)
```
Signed-off-by: Philip Withnall <pwithnall@endlessos.org>
When the folder changes, do not select the first item in the list,
so if the user enters a folder and then clicks the accept button,
the current folder is returned instead of the selected one.
To maintain consistency with the previous implementation, when a
folder in the path bar is clicked the previously-entered folder is
selected, and when the file chooser is in open file mode the first
item is always selected.
See #5438
This particular relation was not exposed to at-spi2.
Exposing it required adding a missing at-spi2 relation variant, but it was introduced
in at-spi2-core 2.26, so that's likely safe as well.
It was, for some reason, mapped to ATSPI_ROLE_SECTION, and GTK_ACCESSIBLE_ROLE_SECTION was mapped to
ATSPI_ROLE_FILLER, so the mapping is reversed. So, reverse it and make it correct.
Setting this attribute after querying, but before receiving the
results, can lead to inappropriate behaviour. This can be reproduced
by dragging the scrollbar very quickly in a large directory; after
going up and down a few times, some thumbnails will be wrong.
Without this branch, "wrong" means they'll show the completely wrong
icon or thumbnail, e.g. a folder icon in a video file. With previous
commit, "wrong" means they'll be empty even when there is a thumbnail
available.
The sequence of events that triggers this is as follows:
1. GtkListItem receives a GFileInfo object and passes it to
GtkFileThumbnail via expressions
2. `get_thumbnail()` is called, doesn't find a thumbnail
3. `filechooser::queried` is not set yet, so it is set to TRUE
and we call `g_file_query_info_async()`
4. **Before `thumbnail_queried_cb` is called**, a new GFileInfo
is set, and we cancel the query initiated in the previous
step
5. We now have a GFileInfo with `filechooser::queried` set to
TRUE, and no thumbnail!
This commit fixes that by only setting the `filechooser::queried`
attribute after the icon is queried. We need to set it in two
situations: when the query is successful, or when the error is
not G_IO_ERROR_CANCELLED. That's because the query was cancelled,
we didn't really perform it!
Unset the image if we fail to find the appropriate icon, regardless
of the reason of the failure. Prevents the thumbnail to misrepresent
the GFileInfo it's supposed to represent.
Currently nested custom tags work only as long as the element names differ
from the root one. If it's same, for example:
<condition type="any">
<condition type="max-width">600</condition>
<condition type="max-height">600</condition>
</condition>
then it will fail. Meanwhile the same tags wrapped into <conditions> would
work.
The problem is that custom tag parsing is considered finished as soon as we
encounter a closing tag with the same element name. So instead, track the
nesting level.
Reset alloc_needed_on_child *before* allocating the children. This is
because some child's size_allocate() may call queue_allocate(), which
will bubble up alloc_needed_on_child. An example of this happening is
with GtkScrollable implementations, which are supposed to configure
their adjustments in size_allocate(), which will cause GtkScrollbar's
GtkRange to notice and queue_allocate() on itself.
If we reset alloc_needed_on_child after this happens, then our children
will have a lingering alloc_needed_on_child and will not receive an
allocation.
This commit fixes widgets occasionally losing an allocation when this
scenario happens.
Programmatic changes to the entry contents should
not become part of the undo history.
Sadly, the editable implementations are also used
in the code paths that we use for user-initiated changes,
so we have to be careful to only set them as
irreversible if we are not already in a user action.
Fixes: #5622
Keep a separate boolean for enable-undo, and
disable the history if it is false, or the entry
is not using visible text, or isn't editable.
Related to: #5622
Previously, it was mapped to ATSPI_STATE_INVALID. However, that state
is used for some internal errors, and not user errors, so use the correct
one for that purpose.
GtkButton still has some code checking if the instance passed to
gtk_button_set_label() is a GtkCheckButton; GtkCheckButton is not a
GtkButton any more.
In 32247bc50e node_get_for_file() was
changed to return GTK_INVALID_LIST_POSITION rather than 0 when the file
is untracked. Most call sites were updated accordingly, but this one was
missed.
Fixes#5619
Up until now, toggle buttons were presented as regular push buttons.
That's the approach used by the ARIA specification, however, our platform
accessibility backend, at-spi2, can not represent accessibe states with values,
so we can not represent the design pattern precisely enough for screen readers.
If, in future, the a11y backends gain this capability, we might consider again
removing this role.
Showing a destroyed window might cause an application to
behave in an unexpected manner. For example, showing a
dialog after it has been closed by the user might cause the
application to freeze. The warning will help developers to
track down the issue.
Signed-off-by: Aaron Erhardt <aaron.erhardt@t-online.de>
Reuse a better to read would_drop() from ./testsuite/reftests/gtk-reftest.c
in ./tools/gtk-builder-tool.c
Fixed wrong indentation in ./testsuite/reftests/gtk-reftest.c
I encountered this issue where I casted user_data to my self type, but it
showed me they were actually swapped when I set the "object" signal attribute.
After checking the source code which confirms this, it is a good idea to
properly document that convenient behaviour.
2022-02-24 23:04:20 +01:00
1830 changed files with 83512 additions and 50199 deletions
GSK render nodes can be serialized and deserialized using APIs such as `gsk_render_node_serialize()` and `gsk_render_node_deserialize()`. The intended use for this is development - primarily the development of GTK - by allowing things such as creating testsuites and benchmarks, exchanging nodes in bug reports. GTK includes the `gtk4-node-editor` application for creating such test files.
The format is a text format that follows the [CSS syntax rules](https://drafts.csswg.org/css-syntax-3/). In particular, this means that every array of bytes will produce a render node when parsed, as there is a defined error recovery method. For more details on error handling, please refer to the documentation of the aprsing APIs.
The format is a text format that follows the [CSS syntax rules](https://drafts.csswg.org/css-syntax-3/). In particular, this means that every array of bytes will produce a render node when parsed, as there is a defined error recovery method. For more details on error handling, please refer to the documentation of the parsing APIs.
The grammar of a node text representation using [the CSS value definition syntax](https://drafts.csswg.org/css-values-3/#value-defs) looks like this:
Each node has its own `<node-name>` and supports a custom set of properties, each with their own `<property-name>` and syntax. The following paragraphs document each of the nodes and their properties.
Each node has its own `<node-type>` and supports a custom set of properties, each with their own `<property-name>` and syntax. The following paragraphs document each of the nodes and their properties.
When serializing and the value of a property equals the default value, this value will not be serialized. Serialization aims to produce an output as small as possible.
To embed newlines in strings, use \A. To break a long string into multiple lines, escape the newline with a \.
# Names
### Nodes
Nodes can be given a name by adding a string after the `<node-type>` in their definition. That same node can then be used further down in the document by specifying just the name identifying the node.
### Textures
Just like nodes, textures can be referenced by name. When defining a named texture, the name has to be placed in front of the URL.
# Nodes
### container
@@ -183,7 +193,7 @@ Creates a node like `gsk_linear_gradient_node_new()` with the given properties.
@@ -68,7 +68,7 @@ in a selector, widget names must be prefixed with a # character.
| E:focus-within | [CSS Selector Level 4](https://drafts.csswg.org/selectors/#focus-within-pseudo) | Set on all ancestors of the focus widget, unlike CSS |
| E:focus-visible | [CSS Selector Level 4](https://drafts.csswg.org/selectors/#focus-within-pseudo) | Set on focus widget and all ancestors, unlike CSS |
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.
