i.e. when wrap-width > 0. This was only being done for non-grid cases.
So, ComboBoxes in grid mode did not indicate their selection when popped
up and required users to keynav from ‘nothing’ (at the top-left) to the
item they wanted to select. By selecting the active item in advance, now
it’s highlighted & acts as the starting point for keynav around the grid
This previously only mentioned its effect on the displayed value, and
even after the previous commit, its rounding of the actual value upon
change still reads like too much of an afterthought. Worse, it wasn’t
mentioned at all in the doc for the @digits parameter. Change this to
emphasise rounding always occurs and the displayed value is secondary.
Whether it should is an open question, but for now, the documentation
should clearly indicate that currently rounding is only applied upon
changes to the value, not to the existing value when ::digits changes.
This is already clear in the doc for the underlying Range::round-digits.
https://bugzilla.gnome.org/show_bug.cgi?id=358970
The documents state that gtk_scale_set_digits() “causes the value of the
adjustment to be rounded off to this number of digits, so the retrieved
value matches the value the user saw.” Note the lack of any condition.
But in fact, if draw-value was false, rounding was disabled on the base
Range, so values that weren’t displayed weren’t rounded. This made the
docs wrong and made an apparently cosmetic detail alter functionality.
Fix by ensuring the number of digits set on Scale is always propagated
along to gtk_range_set_round_digits(), thus rounding to it in all cases
when the value changes, regardless of whether the value is displayed.
This doesn’t address the other idea from Bugzilla: that changing the
number of digits should clamp the _existing_ value if it’s more precise.
This contradicts digits docs in the base Range, but the above from Scale
can be read as implying it’ll happen. For now, that’s an open question.
https://bugzilla.gnome.org/show_bug.cgi?id=358970
GtkFileChooserButton installs a handler for the popped-up signal, which
refilters the menu, in order to hide the “(None)” item from the popup
if it was previously selected in the ComboBox. This oddity means that:
• Until recently, this item would be selected in the menu shell, which
would then be popped up and change the selection away from that item.
This was therefore redundant (more on which below!) but benign.
• After the patch for https://bugzilla.gnome.org/show_bug.cgi?id=771242
however, this causes a critical assertion fail, as now we stash the
originally selected item in a pointer so that it can be selected only
after realisation/popup – but by that stage, the model has just been
refiltered and the previous pointer no longer refers to a valid item.
This commit works around this problem by, after popping up the menu,
getting the active item again, in case a popped-up handler has gone and
invalidated the pointer to the active item that we saved before popup.
If a handler does this, everything done to find/use the original item is
pointless. But this avoids the ugly critical in FileChooserButton, while
not harming every other ComboBox that doesn’t mess with its model while
popping up (hopefully the vast majority), and it’s very difficult to
imagine a way to check if the active item is /going to/ be hidden later)
Previously, for compatibility with GTK 3.0, we allowed specifying
numbers without units and interpreted them as pixels, even when the CSS
specification didn't.
Remove that now that we can break API.
This reverts commit 4875c689a0.
This was a thinko. Writable is not actually settable from the
application side, but only for the user, from the backend side.
Elsewhere we already go through the keymap to get modifiers so we
should do the same here. In fact, this was relying on xkb modifier
mask values being bitwise compatible with GdkModifierType which isn't
necessarily true.
https://bugzilla.gnome.org/show_bug.cgi?id=770112
Gtk+ treats MOD1 as a synonym for Alt, and does not expect it to be
mapped around, so we should avoid adding GDK_META_MASK if MOD1 is
already included to avoid confusing gtk+ and applications that rely on
that behavior.
https://bugzilla.gnome.org/show_bug.cgi?id=770112
When a subsurface is used as a parent of a popup, GDK needs to traverse
up to the transient-for as the next parent, to properly find the parent
used by the popup positioner. This is because the parent of a popup
must always either be an xdg_popup or an xdg_surface, but traversing
the "parent" (in GDK terms) upwards from a subsurface will end up on
the fake root window before we hit the actual parent (in Wayland terms).
https://bugzilla.gnome.org/show_bug.cgi?id=776225
Instead of having 3 different shaders for the different clipping
versions, just have one shader and use a preprocessor define to use
different clip functions.
That preprocessor define is set in the Makefile.
Also use foo.frag and foo.vert as the file extensions instead of using
foo.frag.glsl and foo.vert.glsl, as that's what glslc suggests as
extension.
That way we don't need to move the clip rounded rect manually through
the vertex shader into the fragment shader but can just look at the push
constants.
Simplifies shaders a lot.
Passing a rectangle with zero width or height to xdg_shell-v6
set_anchor_rect() will cause a protocol error and terminate the client,
as with gedit when pressing the Win key.
Reason for this is because the rectangle used to set the anchor comes
from gtk_text_layout_get_iter_location() which uses the pango layout
width/height, which can be empty if there is not character at the given
location.
Make sure we don't use 0 as width or height as an anchor rectangle to
avoid the protocol error, and compensate the logical position of the
given rectangle if the size is changed, so that the actual position
remains as expected by the client.
https://bugzilla.gnome.org/show_bug.cgi?id=777176
Images with just an aspect ratio, but without a size, should be scaled
to be fully visible in the given area.
But we scaled them to completely cover the given area, which made them
partially invisible.
Reftest included.
gtk_snapshot_pop() => removed
gtk_snapshot_pop_and_append() => gtk_snapshot_pop()
So now there is no way to get a rendernode out of the snapshotting API
until you gtk_snapshot_finish().
... and use it.
The function is a bit awkward because it requires 2 calls to
gtk_snapshot_pop(), but once you accept that, it's very convenient to
use, as can be seen by the 2 implementations.
This is a free-form tab that can contain information about the
system environment. To see it, set GtkAboutDialog::system-information
to a non-NULL value.
https://bugzilla.gnome.org/show_bug.cgi?id=776604
The Mesa Vulkan drivers need XInitThreads() being called, because their
implementation has to use threads.
And I don't want to make the call depend on if Vulkan is compiled in
because that makes GTK's X11 behavior depend on compile-time flags, so
it's always called.
This way, we ensure that files that are built during make always get
properly listed. And we ensure that creating the resources actually
depends on them.
We already take ints when setting the translation, so it can't
currently take any other values. Additionally, I was seeing large
costs in int -> double -> int for the rects in
gtk_snapshot_clips_rect(), as all callers really are ints (widget
allocations) and the clip region is int-based.
This change completely cleared a 2% rectangle_init_from_graphene from
the profile and is likely to have nice performance effects elsewhere
too.
The width/height/aspect getters are called a lot, and almost all
callers already verify it from _gtk_css_image_get_concrete_size (),
so just skip these checks.
This means we allocate the collect data with the state, avoiding
an extra allocation. Also, a union means every state object
is the same size and we could reuse the state objects.
This was showing up quite high on the profiles, and there is
no real reason for copy to normalize, as the source is a
GskRoundedRect which should be normalized already unless
you did something very strange (and then you should have normalized
manually).
Simgle image cross-fade opacity was computed the wrong way, which caused
weird fade-in/out animations, for example in flat buttons.
I messed this up when porting cross-fades to snapshot().
Since the demise of theme engines, we can no longer hit
the case of id >= GTK_CSS_PROPERTY_N_PROPERTIES. So don't
check for this in a very frequently called function.
Using an image() fallback from svg to png doesn't make too
much sense, since the svg is always used (unless librsvg is
not present), while the png icon is faster and cheaper to
load and thus preferable.
Also, "ie" wasn't very clear, but fixing that to "i.e." would cause
truncation of the summary when processed by bindings using doxygen. So,
I replaced it with "in other words", which is no _less_ clear, at least.
It was suggested that the project files to be moved to win32/, so that we can
have one less layer of directories we need to go down into to reach the project files.
Tell people about what happens when generating projects when Visual
Studio 2013 or later is required, and mention that the .headers are
only needed when headers need to be copied.
So we can set the css name of a widget to something that's not related
to the class name. If the css-name property is set to NULL, we will
still fall back to the one set using gtk_widget_class_set_css_name which
is alwasys non-NULL since GtkWidget itself sets it to "widget".
Instead of relying on --generate-dependencies and the resource file,
actually list the resources in Make variables.
Fixes make not building new shaders because they're not inside the
resource file.
See the implementation of gtk_entry_create_layout():
pango_attr_list_splice() is used to add the PangoAttrList of the preedit
string. And that is done *after* applying the PangoAttrList of the
"attributes" property.
https://bugzilla.gnome.org/show_bug.cgi?id=776868
We now have GTK_CSS_AFFECTS_CONTENT for properties that have an effect
on content rendering.
Using GTK_CSS_AFFECTS_ICON is wrong for icon-transform and icon-filter
as they don't change the icon, just how the icon is rendered, so we use
GTK_CSS_AFFECTS_CONTENT for those.
We also introduce GTK_CSS_AFFECTS_POSTEFFECT for opacity and filter -
properties that affect the whole drawing of the widget by applying an
effect after everything is said and done.
...which treats the first '.' in doc comments as the end of the summary.
So, e.g., in gtkmm, get_kinetic_scrolling() is currently summarised as
"Changes the behaviour of @scrolled_window wrt." Not very informative!
No need for a period there & anyway, the phrase "wrt to" is superfluous,
and we have space to actually say "with regard to", so just do that now.
Instead of
-gtk-icon-effect: dim;
-gtk-icon-effect: hilight;
we now use
-gtk-icon-filter: opacity(0.5);
-gtk-icon-filter: brightness(1.2);
respectively.
This node essentially implements the feColorMatrix SVG filter. I got the
idea yesterday after looking at the opacity implementation.
It can be used for opacity (not sure if we want to) and to implement a
bunch of the CSS filters.
Since the status of the GDK broadway backend is more or less unsupported,
drop the projects that build gtk4-broadwayd and gdk-broadway, and update
the projects to not to refer to them.
However, keep the Broadway configs for now as we will later transform
them to become configs for Vulkan, so bascially besides "installation"
parts and output settings, they will do the same as their Release|Debug
counterparts with no support for Broadway.
...but disable them for now. Configs will be added for the projects to
support Vulkan-enabled builds which will then enable the builds of these
sources. Extra commands and items will be needed for the GSK resources
along with ensuring GSK_RENDERER_GSK being defined for the build of GDK,
GDK-Win32 and GSK so that the builds of Vulkan-enabled builds can be done
properly.
Filter out the Vulkan sources from the 'dist hook' rules in
gsk/Makefile.am as we don't want to in turn include them twice in the
projects when the 'make dist' is performed on a system with Vulkan
builds enabled.
If the signal handler ends up changing the label text,
the link is no longer around to update the css node.
Check for this possibility to avoid a crash here.
One cannot use #if...#endif within macro calls in Visual Studio and
possibly other compilers, and there are more uses of VLAs that need to be
replaced with g_newa().
There were also checks for the clip type in gskvulkanrenderpass.c which
were possibly not done right (using the address of the type value to check
for a type value), which triggered errors as one is attempting to compare
a pointer type to an enum/int type.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
This adds support to the GDK Win32 backend so that we can support Vulkan
context creation for use in the GSK Vulkan renderer, so that we can test
it on Windows platforms as well.
https://bugzilla.gnome.org/show_bug.cgi?id=776544
Use g_newa() instead of VLAs, as VLAs may never be supported by some
compilers as it became optional in C11 and there are concerns about their
implementations in compilers that do support it.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
Forces a full redraw every frame.
This is done generically, so it's supported on every renderer.
For widget-factory first page (with the spinner spinning and progressbar
pulsing), I get these numbers per frame:
action clipped full redraw
snapshot 0ms 7-10ms
cairo rendering 0ms 10-15ms
Vulkan rendering 3-5ms 18-20ms
Vulkan expected * 0ms 1-2ms
GL rendering unsupported 55-62ms
* expected means disabling rendering of unsupported render nodes,
instead of doing fallback drawing. So it overestimates the performance,
because borders and box-shadows are disabled.
It's faster to render once for every rectangle in the clip region than
rendering the outline of the clip region.
Especially because this reduces the time necessary to build up the frame
data.
In widget-factory (where we have 3 rectangles), this leads to a 5x
speedup in the rendering time rendering alone.
Snapshotting time goes from 10ms to ~1ms, which is another huge
improvement.
Note: We interpolate premultiplied colors as per the CSS spec. This i
different from Cairo, which interpolates unpremultiplied.
So in testcases with translucent gradients, it's actually Cairo that is
wrong.
Homogeneous branches repeated the calculation/assignment of the initial
space available to children. This avoids that by shuffling some code.
Perhaps more importantly, in doing that, I ended up with some ambiguous
names, and Company and I realised how vague the pre-existing naming was.
"size" becomes "extra_space", as this is what it represents. Conversely,
"extra" becomes "size_given_to_child" (albeit still given out in two
different ways depending on whether the Box is homogeneous). My hope is
that these sections of code are now somewhat less baffling than before!
This is now tracking the clips added by the clip nodes.
If any particular node can't deal with a clip, it falls back to Cairo
rendering. But if it can, it will render it directly.
Now that every call to GtkCellArea is a snapshot call and no more cairo
calls are left, move the actual differentiation between Cairo and
Snapshot down to the cell renderer.
Little tool that creates a bunch of test files to throw add the
rendernode binary.
They should really be part of a testsuite, but we have none, so OI just
put them here.
Only keep the version that calls gsk_render_node_draw() if people
specify the --fallback option.
The actual renderer selection works just as for regular GTK. The easiest
way to influence it is setting the GSK_RENDERER environment variable.
... and implement it for the Cairo renderer.
It's an API that instructs a renderer to render to a texture.
So far this is mostly meant to be used for testing, but I could imagine
it being useful for rendering DND icons.
That code doesn't do anything.
And what the code should be doing (clearing the abckground) isn't
necessary as cairo drawing is guaranteed to clear the surface.
This does a conversion to/from GBytes and is intended for writing tests.
It's really crude but it works.
And that probably means Alex will (ab)use it for broadway.
I had originally thought I'd use GskShadow for box-shadow, but didn't in
the end.
So now it's only used for text-shadow and icon-shadow, and those don't
have a spread.
VLAs are not supported by Visual Studio and possibly other compilers that
are supported by GTK+-3.90+, and probably never will be, although it is a
C99 specification, and it became optional for C11. It is also not a part
of the newer compiler requirements that are listed out for GTK+-3.90.x.
There exist concerns about the implementation of VLAs in compilers that
support them as well, so change it to a g_newa() approach.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
Instead of a separate allocation for any arrays in the render node
we allocate these as part of the render node itself, using C99
flexible arrays.
This leads to less allocations, which is nice, but the major reason
for this is that it allows us to change the allocation scheme further
in the future. For instance, we want to do stack-like allocation so
that all the render-nodes for an entire frame are allocated in one
(or a few) chunks.
Instead of constantly recalculating this (especially recursively for
parents!) we do it only on construction, because everything is
immutable anyway. Also, most nodes had a bounds already and can
use the new parent member instead.
We also do direct access to the node bounds rather than calling
gsk_render_node_get_bounds in various places, which means
we do less copying.
This causes the snapshotting algorithm to dump all widget nodes into
their own container node. We then name that group accordingly (ie
"GtkSwitch<0xdeadbeef>") so you can easily see which node belongs where.
The feature is toggleable in the inspector's visual tab.
There's a few problems with it, becuse GtkSnapshot optimized container
nodes away if they are not needed, so we are losing some widgets...
When the first/last color stop is not at 0%/100%, we need to start the
repeating at their offsets and not at 0%/100%.
Attached reftest demonstrates the problem.
Instead of making people intiialize a rectangle and then applying border
radius manually, provide a constructor that does it for them.
While doing that, also allow people to instead request the padding box
or the content box.
Refactor all relevant code to use this new constructor.
... and make the icon rendering code use it.
This requires moving even more shadow renering code into GSK, but so be
it. At least the "shadows not implemented" warning is now gone!
The node draws a solid CSS border, which can be used to cover everything
but dashed and dotted borders (double, groove, inset, ...).
For different border styles, we overlay multiple nodes and set their
colors to transparent for sides with non-matching styles.
This way we can pass the command pool around.
And that allows us to allocate and submitcustom buffers.
And that is necessary to make staging images work.
It turns out, some simple getters - such as
gdk_drawing_context_get_clip() - love copying things before returning
them.
I guess somebody has to burn cycles...
When the background-clip of the background is smaller than the
background-clip of blended images, not pushing a group is wrong.
Test testing exactly that included.
This uses the new push()/pop() mechanism to its fullest extent when
implementing transitions. It's fun to inspect the results in the
inspector.
Crossfades don't work yet, they continue using a Cairo fallback.
A side effect of the stack conversion is that widget-factory now uses
snapshots for a lot more things.
It is now possible to call push() subfunctions for simple container
nodes with just a single child. So you can for example
gtk_snapshot_push_clip() a clip region that all the nodes that get
appended later will then obey.
gtk_snapshot_pop() will then not return a container node, but a clip
node containing the container node (and similar for the transform
example).
This is implemented internally by providing a "collect function" when
pushing that is called when popping to collects all the accumulated
nodes and combine them into the single node that gets returned.
To simplify things even more, gtk_snapshot_pop_and_append() has been
added, which pops the currently pushed node and appends it to the
parent.
The icon rendering code has been converted to this approach.
This code makes renderers fall back to Cairo rendering if they don't
know how to handle a render node's type.
This allows adding new render nodes with impunity.
Instead of appending a container node and adding the nodes to it as they
come in, we now collect the nodes until gtk_snapshot_pop() is called and
then hand them out in a container node.
The caller of gtk_snapshot_push() is then responsible for doing whatever
he wants with the created node.
Another addigion is the keep_coordinates flag to gtk_snapshot_push()
which allows callers to keep the current offset and clip region or
discard it. Discarding is useful when doing transforms, keeping it is
useful when inserting effect nodes (like the ones I'm about to add).
Instead of having a setter for the transform, have a GskTransformNode.
Most of the oprations that GTK does do not require a transform, so it
doesn't make sense to have it as a primary attribute.
Also, changing the transform requires updating the uniforms of the GL
renderer, so we're happy if we can avoid that.
I'm about to move children handling to the container node, which means
the generic code can no longer assume children APIs existing.
So rewrite the treemodel to work without it.
gsk_render_node_get_bounds() still exists and is computed via vfunc
call:
- containers dynamically compute the bounds from their children
- surface and texture nodes get bounds passed on construction
In the brave new world of refactored render nodes, this function doesn't
really make any sense anymore. We could turn it into a vfunc, but I
don't think it's useful.
Especially because even in the brave old world, this function was
causing a vastl overallocation of nodes when the GL renderer needed render
targets.
If we ever feel, we need this function again, we can readd it later.
But nobody is using it other than for overriding opactiy. And you can
just override opacity directly if you care.
Creating render nodes is fire-and-forget, so all one should do is create
a container, append, append, append and then send it off to the
renderer. So there's no need to replace, insert between or anything
else.
We want to split nodes into containers and nodes that do actual drawing.
So pushing nodes that do drawing is exactly the wrong thing.
Also fix up GtkPopover. There's no need for it to push anything.
When we generate the Visual Studio 2013 projects, we need to remove the
*.vs12.sourcefiles and *.vs12.sourcefile.filters that are generated during
the process, so that 'make distcheck' won't complain about leftover files.
When running uninstalled tests with GtkApplication on an autobuilder with
a fake session bus, warnings will cause the tests to abort. The GNOME
session manager, the Xfce session manager, and the Inhibit portal are all
not needed for normal operation of GTK, so we should not log warnings if
they are not found.
As well as not being present on a fake session bus, it's also not
expected that they'll be present on all platforms.
https://bugzilla.gnome.org/show_bug.cgi?id=774784
... with gtk_list_box_get_row_at_y. It would be nice to avoid the
'find' versus 'get' discrepancy since we are planning to expose it as
public API.
https://bugzilla.gnome.org/show_bug.cgi?id=776187
Fix the build after the branch wip/alexl/simplify-gdkwindow was merged, as
there are some changes that broke things in the Windows backend, namely:
-gdk_win32_input_shape_combine_region() should not be removed at this
point (though it is a stub--otherwise GDK/Win32 will crash)
-Some more code need to be removed due to the removal of items in the
above-mentioned merged branch
Also, like the X11 backend, do not allow the creation of native child
windows, and stop checking for subsequent child windows
(GDK_WINDOW_CHILD), so that we can clean things up a bit.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
It does weird clipping that
(a) nobody likes
(b) is hard to support in the new rendering world.
So we take the easy way out.
The actual frame is now drawn by the frame node around the label.
GtkCellView has a gadget, so peopl can do all their shenanigans with
CSS.
And the original use case (overriding the background so that the
cellview's GdkWindow shares the background color of the combobox) is
outdated since we have transparent backgrounds.
We're not currently using this, and dropping it allows us to loose
a bunch of code which leads us towards the goal of having GdkWindow
only for toplevels (and reparenting makes not sense for toplevels).
We can't really support these on e.g. wayland anyway, and we're trying
to get rid of subwindow at totally in the long term, so lets drop this.
It allows us to drop a lot of complexity.
For subsurfaces, the new state which includes the input shape is not
applied by the compositor if the subsurface is in effective synchronous
mode.
So we need to apply the input shape once parent surface is in effective
desynchronized mode, which is when it's committed, otherwise the input
shape may never be applied if the widget is not using being_paint() /
end_paint() to draw on its subsurface, like clutter does.
We do that only for empty input shape as those won't need update when
the subsurface is resized, for all other non-empty input shape, the
client still has to use begin_paint()/end_paint() for the input shape to
be applied.
https://bugzilla.gnome.org/show_bug.cgi?id=774534
It's using a GtkCssPositionValue, even though that name is wrong. But
the functionality of managing 2 lengths is exactly what we want.
Nobody is using this yet.
- Recognize "gl" as well as "opengl" for the GL renderer
- GSK_RENDERER=help now works
- g_warning() for an unrecognized renderer (typo detection!)
- g_print() the actual renderer that is used (and error messages when
selecting) when a GSK_RENDERER is given, so you'll notice if your
renderer isn't taken.
Previously, code would work fine with --disable-vulkan if the Vulkan
headers were installed - code would happily just use them as they're
installed in /usr/include.
By creating unlimited render objects, we would never wait on the GPU.
This would mean that if the GPU was the bottleneck, we would fill its
queue with render commands faster than it could process them.
And because the nvidia binary driver and my code work surprisingly well
and bugfree, this lead to exhaustion of RAM. I had 50GB of swap
configured and my hard disk was quicker as swap storage than my GPU was
at processing the commands, so stuff still filled up.
At that point my computer became rather unresponsive and I decided to
reboot it, so I that could write this patch.
Add SURFACE and TEXTURE operations. This way, we actually render more
than one node every frame because not everything is a fallback node
anymore that gets composited with its children into a cairo surface.
Instead of pushing the root matrix, push the world matrix for the
current node. That way, the bounds we emit as vertices are actually
properly transformed.
First, we collect all the info about descriptor sets into a hash table,
then we use its size to determine the amount of sets and allocate those
before we finally go ahead and use the hash table's contents to
initialize the descriptor sets.
And then we're ready to render.
We can let the GPU do its stuff without waiting. The GPU knows what it's
doing.
Which means we now get a lot of time to spend on doing CPU things (read:
we're way better in benchmarks).
The old behavior is safer, so we want to keep it around for debugging.
It can be reenabled with GSK_RENDERING_MODE=sync.
And move the actual rendering code there.
A RenderPass is a collection of operations on the same target that
get executed one after another. It roughly targets VkRenderPass or
rather the subpasses of a VkRenderPass.
For now, only the infrastructure is there. No real stuff is happening.
This is refactoring work.
GskVulkanRender is supposed to be the global object for a render
operation, ie GskVulkanRenderer.render() will create this object for
what it does.
The object will be split into stages that perform the operations
necessary to create a drawing.
Instead of using a staging iamge, we require the final image to be
linearly allocated and have host-visible memory.
This improves performance quite a bit.
The old code is still there and can be enabled with a simple change
to a #define in gskvulkanimage.h
Instead, complain if somebody calls gdk_x11_window_get_xid() on a
non-native window.
We cannot make random windows native anymore because there's no GSK
renderer associated with them, so we cannot draw them.
We can now upload vertices.
And we use this to draw a yellow background. Which is clearly superior
to not drawing anything.
Also, we have shaders now. If you modify them, you need glslc installed
so they can be recompiled into Spir-V bytecode.
1. Output Vulkan status in summary
2. Add missing "test" call
3. Check for glslc
The glslc check will be necessary later for the code that automatically
compiles the Vulkan glsl source to Spir-V.
Nothing happens if glslc is not available - unless you modify the glsl.
gdk_window_create_vulkan_context() now exists and will return a Vulkan
context for the given window. It even initializes the surface. But it
doesn't do anything useful yet.
Adds the gdk_display_ref_vulkan() and gdk_display_unref_vulkan()
functions which setup/tear down VUlkan support for the display.
Nothing is using those functions yet.
These only exist for the window dragging which does not exist anymore
currently. It will be reintroduced later in a form that does not require
these handlers.
I read the code as if (use_gl) instead of if (!use_gl) and commented it
out in bddfd7bb41. That broke drawing on
Wayland without OpenGL completely.
Whoops.
Now it's back.
There were some parts that need some updates after the refactoring in
GDKGL, so that the code will continue to build and run.
For gdkwindow-win32.c, comment out the parts where we check for use_gl
(which was removed), since we are going to move all drawing to OpenGL,
but don't remove/disable the whole portion as that transition is not
complete at this point.
There a is new GDKGL function that checks for the damaged area of the back
buffer, but since the notion of "damage" is for *NIX (GLX/EGL for
Wayland/mir), meaning that there is no such extension for Windows in this
regard, so we can't support this on Windows as-is, at least for now.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
This is a way to query the damaged area of the backbuffer.
The GL renderer uses this to compute the extents of that damage region
(computed via buffer age) and use them to minimize the area to redraw.
This changes the semantics of GL rendering to "When calling
gdk_window_begin_frame() with a GL context, the area by
gdk_gl_context_get_damage() needs to be redrawn and every other pixel of
the backbuffer is guaranteed to be correct.
After gdk_window_end_frame() on a GL-drawn window, the whole backbuffer
must be correct.
We can always glXBufferSwap() now because of this.
... instead of a gl context.
This requires some refactoring in the way we mark the shared context as
drawing: We now call begin_frame/end_frame() on it and ignore the call
on the main context.
Unfortunately we need to do this check in all vfuncs, which sucks. But I
haven't found a better way.
That way we can capture both the actual changes (clip region) and the
area that was redrawn (render region), which in OpenGL might not be
identical.
Nothing shows the render region yet though...
Reenable GL drawing, but do it without Cairo.
Now, the context passed to gdk_window_begin_draw_frame() decides how
drawing is going to happen. If it is NULL, Cairo is used like before.
If a context is passed, Cairo may not be used for drawing and
gdk_drawing_context_get_cairo_context() is going to return NULL.
Instead, the GL renderer must draw to the GL backbuffer and
end_draw_frame() is then swapping that to the front.
The GskGLRenderer has lost the texture it used to render to and adapted
to render directly to the backbuffer instead.
The only thing missing is for GtkGLArea to gain back a performant way to
render. But it didn't have one since the introduction of GSK, this
patchset doesn't change anything about it.
The new rendering avoids two indirections (the GSK renderer's texture
and the GDK double buffering surface).
It improves icon count in the fishbowl demo by 30%.
This way, we can query the GL context's state via
gdk_gl_context_is_drawing().
Use this function to make GL contexts as attached and grant them access
to the front/backbuffer for rendering.
All of this is still unused because GL drawing is still disabled.
No visible changes as GL rendering is disabled at the moment.
What was done:
1. Move window->invalidate_for_new_frame to glcontext->begin_frame
This moves the code to where it is used (the GLContext) and prepares it
for being called where it is used when actually beginning to draw the
frame.
2. Get rid of buffer-age usage
We want to let the application render directly to the backbuffer.
Because of that, we cannot make any assumptions about the contents the
application renders outside the clip area.
In particular GskGLRenderer renders random stuff there but not actual
contents.
3. Pass the actual GL context
Previously, we passed the shared context to end_frame, now we pass the
actual GL context that the application uses for rendering. This is so
that the vfuncs could prepare the actual contexts for rendering (they
don't currently).
4. Simplify the code
The previous code set up the final drawing method in begin_frame.
Instead, we now just ensure the clip area is something we can render
and decide on the actual method in end_frame.
This is both more robust (we can change the clip area in between if we
want to) and less code.
This is a temporary switch-off of the GL dawing code that will make
things keep running. All GL related code (like the GSK renderer or
GtkGLArea will now fall back to software.
Wayland subsurfaces can have other native window parents, but those need
to be destroyed along with the rest of the window hierarchy otherwise
an assert() is reached.
https://bugzilla.gnome.org/show_bug.cgi?id=774915
GtkListBox is not a windowed widget anymore so we can't use
gtk_widget_get_window. Just directly access priv->view_window instead to
get the right window.
For a menu mode CB with wrap_width == 0 and an active item, that item is
selected in gtk_combo_box_menu_popup. Selection causes the MenuShell to
activate and hence take a grab. This was done before the menu was popped
up. A patch distributed in Debian sid - after being proposed on our BZ -
revealed that on the 1st popup of any such ComboBox, within grab_add,
the MenuShell's toplevel's GdkWindow is NULL. This causes a Gdk-CRITICAL
assertion fail on the 1st time opening any such CB, on Debian and if
that patch were merged to GTK+. By selecting after popup, we ensure the
MenuShell is realised before its grab_add and so avoid the critical.
https://bugzilla.gnome.org/show_bug.cgi?id=771242
This way, we don't spam criticals when GL is not available. Instead, we
print a useful debug message to stderr and continue with the Cairo renderer.
Signed-off-by: Emmanuele Bassi <ebassi@gnome.org>
and remove gsk_renderer_get_for_display().
This new function returns a realized renderer. Because of that, GSK can
catch failures to realize, destroy the renderer and try another one.
Or in short: I can finally use GTK on Weston with the nvidia binary
drivers again.
Signed-off-by: Emmanuele Bassi <ebassi@gnome.org>
Instead of having a gsk_renderer_set_window() call, pass the window to
realize(). This way, the realization can fail with the wrong window.
Signed-off-by: Emmanuele Bassi <ebassi@gnome.org>
gdk_window_get_toplevel() walks up the windows tree looking for the
corresponding toplevel window, but needs to account for subsurfaces as
well on Wayland.
https://bugzilla.gnome.org/show_bug.cgi?id=775319
We were producing org.symbolic.png from org.gnome.Recipes-symbolic.svg,
which is not useful. Look for the last dot in the original name, to
produce the expected org.gnome.Recipes-symbolic.symbolic.png instead.
gskrenderer.c includes gdk/wayland/gdkwayland.h and as a consequence
we need to be able to locate wayland's headers in case they are not
in standard location.
https://bugzilla.gnome.org/show_bug.cgi?id=775038
Now that subsurfaces can be created as child of another GdkWindow (and
not just the root window), they must be placed according to the location
of their parent, i.e. the abs_x/abs_y must be updated and taken int
account when placing and moving subsurfaces under Wayland.
https://bugzilla.gnome.org/show_bug.cgi?id=774917
Since at-spi-atk commit 96621a5e95 fixed PropertyChange notifications
for AccessibleParent, setting the parent will result in a call to
ref_state_set() which assumes that the object is fully initialized.
https://bugzilla.gnome.org/show_bug.cgi?id=774939
While GtkEventController implementations today are all GtkGesture, it is
possible to create a GtkEventController manually. This is an extrac check
to ensure we only add gestures to the list.
https://bugzilla.gnome.org/show_bug.cgi?id=774760
We need to unrealize the children manually for that to happen, but so it
goes.
The order is necessary because we want the renderer to still be alive
while children are unrealizing.
Only attempt to initialize Wintab after the display manager announces
that the first default display has been set. Fixes a segfault during
initialization of specific tablet drivers' wintab32.dlls. Add assertions
and verbose comments explaining this nonsense because this stuff is a
pain to have to keep fixing.
https://bugzilla.gnome.org/show_bug.cgi?id=774379
Move the orientation sanity-checks into the packet decode func.
Rationale: the packet handling func may otherwise read beyond the end of
device->last_axis_data.
Also expand them to cope with my test Huion's weird reporting.
Also correct the azimuth angle to align with GDK's presentation.
Most importantly, fix annoying comment typo.
https://bugzilla.gnome.org/show_bug.cgi?id=774265
We no longer have GtkPlug nor GtkWin32EmbedManifest for GTK+-4.x, and it
is not entirely clear at this point what would be the "best" replacement
for them, but this issue here prevents GTK+-3.89.x building on Windows.
As a result, this is a fast port to avoid using APIs that have been
removed for 4.x, and things seem to work properly (the print.c page
printed).
https://bugzilla.gnome.org/show_bug.cgi?id=773299
Fix a regression introduced in 4ce6d10601
which causes devices with an odd-numbered zero-based index in the list
to be passed over incorrectly. This might present as yet another "device
does not send pressure" bug for ~50% of devices out there.
This commit also closes off another potential segfault for wintab_devices
lists which have an odd length.
https://bugzilla.gnome.org/show_bug.cgi?id=774699
When checking if a rectangle is contained by the rounded box, the code
will refuse a rectangle which is the exact size as the one backing the
rounded box, since it checks for greater or equal width and height.
Check for greater only instead.
https://bugzilla.gnome.org/show_bug.cgi?id=774114
The functions gdk_pixbuf_get_from_window() and
gdk_cairo_set_source_window() are unreliable and depend on the windowing
system (they work great on X11 and Win32, less so on Quartz and Wayland).
With the switch to new drawing API and OpenGL, we can definitely no
longer support a generic way to snapshot windows.
People should either write windowsystem-specific code or draw their
widgets directly - like with gtk_widget_draw() - if they need to get a
rendering.
The tests read a nonexisting colorprofile, try to convert stuff read
from the window into it, do things that gdk-pixbuf should test and
then aren't even integrated into the testuite.
Sheesh.
- Make the rows larger
- Display the elapsed time between renderings
- Display if it was a full or a partial redraw
- Add a toggle button to display profiler info
Empty doc comments make gtk-doc complain about undocumented
functions, even though these functions are not supposed to
be documented in the first place.
Under Wayland, a subsurface can have another surface as parent, but
gdk would not allow native windows if the parent is not the root window.
Allow native subsurface for all parent under Wayland, not just for the
root window.
https://bugzilla.gnome.org/show_bug.cgi?id=774475
This can be triggered on workspace switches, and on hidpi results in
the scale factor being reset to 1 while the window is not in the
current workspace.
https://bugzilla.gnome.org/show_bug.cgi?id=774476
Just to avoid having to do NULL checks when calling
widget_class->snapshot. We were crashing with drawing areas who don't
have a draw or a snapshot vfunc (woot!).
We need so subtract the allocation from the clip to get the clip offset,
not the other way around.
This was screwing in particular with marks on GtkScale, because GtkScale
mark clip computation is broken and always returns (0,0) which makes
scales have a waaaaay too large clip.
But that's another bug.
And use this to cull widgets and gadgets that are completely outside the
clip region.
A potential optimization is to apply this clip region to cairo contexts
created with gtk_snapshot_append_cairo_node(), but for that we'd need to
apply the inverse matrix to the clip region, and that causes rounding
errors.
Plus, I hope that cairo drawing becomes exceedingly rare so it won't be
used for the whole widget factory like today (which might also explain
why no culling happens in the widget factory outside the header bar.
The equivalent to cairo_matrix_multiply (a, b, c) is
graphene_matrix_multiply (c, b, a).
graphene_matrix_multiply (a, b, c) may not be called with b and c being
the same matrix.
Grips have long been unused in GTK, so remove all support for them.
This removes the GTK_STYLE_CLASS_GRIP and the special
gtk_render_handle() code for drawing those grips.
This allows renderers (or anyone really) to attach "render data" to
textures. Only the first render data sticks.
You can gsk_texture_set_render_data() with the key you will use to
look the data up again, and if no data has been set yet, yours will be
set.
You can retrieve this data via gsk_texture_get_render_data() later on.
If your data has been cleared, NULL will be returned.
When gsk_texture_clear_render_data() is called (which the texture will
call when it is finalized), your destory notify will be called and you
have to release your render data.
The GL driver uses this to attach texture ids to GskTextures.
We do no longer bind textures to a renderer, instead they are a way for
applications to provide texture data.
For now, that's it. We've reverted to uploading it from scratch every
frame.
The snapshot vfuncs must only append at most a single node,
otherwise things are going to break if the widget is the root node.
Unfortunately there is no code that can check this in a generic fashion,
so we'll have to debug this on a case-by-case basis.
This happens in regular code paths for example when trying to render the
empty text string. We don't want to store a surface on the render
node in such a case (so actual rendering isn't slowed down), but we do
want to return a working cairo context that is not in an error state
(so the cairo rendering can continue without error messages).
We want to unrealize the renderer only after all widgets have been
unrealized. Otherwise, the widgets cannot release rendering resources
like textures.
Note that this implementation does not respect GDK windows at all. If
your widget requires respecting them, you should write your own
snapshot implementation and not chain up.
We now look at which of get_render_mode, draw or snapshot vfuncs is the
latest to have been overwritten in the class tree and then use that one.
This allows GtkContainerClass and GtkBinClass to override all of them
for without screwing things up.
and gtk_snapshot_render_frame() to be direct replacements for the
old gtk_render_*() functions.
Use them to replace Cairo usage completely in gtk_window_snapshot().
We now try to emulate cairo_t:
We keep a stack of nodes via push/pop and a transform matrix.
So whenever a new node is added to the snapshot, we transform it
by the current transform matrix and append it to the current node.
Unlike other container widgets, GtkStack would allocate its children
prior to moving its windows, which might prevent further valid size
allocation signals to be emitted.
Re-order the size allocation of child widgets to be performed after
moving the GtkStack windows.
Thanks to Owen for spotting the real issue here.
https://bugzilla.gnome.org/show_bug.cgi?id=767713
As in the last commit on gdkdisplay-win32.c, we need to define that to be
0x0600 (Vista) or later so that the items needed in the Windows headers be
activated.
See: https://bugzilla.gnome.org/show_bug.cgi?id=768081#c62
... to be for Vista (0x0600) or later. This is so that the necessary
items in the Windows headers be activated so that the code will build
properly on mingw-w64, and we already require Vista or later for GTK+.
Thanks Ting-Wei Lan for pointing this out.
See: https://bugzilla.gnome.org/show_bug.cgi?id=768081#c62
This fixes a DOS where any app can cause all running gtk apps
to use arbitrary amounts of memory.
Originally reported against mate-panel, where running a big slideshow
in eye-of-mate caused increasing RAM usage in mate-panel.
v2: Hardcode the value
Signed-off-by: Lauri Kasanen <curaga@operamail.com>
https://bugzilla.gnome.org/show_bug.cgi?id=773587
Making sure the surfaces are using the same scale factor makes it more
likely a fast path will be used when pixman gets involved, as pointed
out by Benjamin Otte.
https://bugzilla.gnome.org/show_bug.cgi?id=772075
We are currently truncating job names to 255 bytes, because that's the
maximum allowed length of job-name attribute in CUPS. This is a CUPS
limitation that GtkPrintOperation shouldn't need to know, and it
shouldn't affect other backends, that might have other limitations or
even no limitation at all. This has another side effect, that what you
set as GtkPrintOperation:job-name could be different to what you get if
the property is truncated, this is not documented in
gtk_print_operation_set_job_name(). So, I think the job name should be
truncated by the CUPS backend, right before setting the job-name
attribute.
https://bugzilla.gnome.org/show_bug.cgi?id=774097
The GApplication platform data may contain a startup ID that on X11
is used to set the startup notification ID when activated. Do the
same on the wayland backend to make startup notifications work for
DBus-activated applications where the DESKTOP_STARTUP_ID environment
variable is not set.
https://bugzilla.gnome.org/show_bug.cgi?id=768531
For wayland clients, the startup notification ID is currently only set
from the DESKTOP_STARTUP_ID environment variable. As that variable is
only set for clients launched via exec(), startup completion is not
indicated correctly for DBus-activated applications unless an explicit
ID is specified - usually that is not the case, as the default handling
uses gdk_notify_startup_complete().
To address this, we need API to set the startup notification ID from GTK
as we have on X11.
https://bugzilla.gnome.org/show_bug.cgi?id=768531
The spec says:
"If <shape> is omitted, the ending shape defaults to a circle if the <size>
is a single <length>, and to an ellipse otherwise."
Make it so.
It wasn’t clear that gtk_style_context_get[_valist]() behave like
g_object_get() — i.e. pointer-based types are returned newly-allocated.
Clarify that.
https://bugzilla.gnome.org/show_bug.cgi?id=773954
GtkLevelBar supports adding custom offsets as style classes, and they
are applied whenever the :value property matches. The current code,
however, only updates any CSS nodes when an offset is found, causing
it to not update when a discrete value changes but no custom offset
is added.
Fix that by always updating the CSS nodes.
https://bugzilla.gnome.org/show_bug.cgi?id=773799
This way we can recommend that applications use the
fullscreen_on_monitor() API on both X and Wayland otherwise they'd
have to keep a path for each backend to achieve this functionality.
https://bugzilla.gnome.org/show_bug.cgi?id=773857
This enables HiDPI support for GTK+ on Windows, so that the
fonts and window look better on HiDPI displays. Notes for the current
work:
-The DPI awareness enabling can be disabled if and only if an application
manifest is not embedded in the app to enable DPI awareness AND a user
compatibility setting is not set to limit DPI awareness for the app, via
the envvar GDK_WIN32_DISABLE_HIDPI. The app manifest/user setting for
DPI awareness will always win against the envvar, and so the HiDPI items
will be always setup in such scenarios, unless DPI awareness is disabled.
-Both automatic detection for the scaling factor and setting the scale
factor using the GDK_SCALE envvar are supported, where the envvar takes
precedence, which will therefore disable automatic scaling when
resolution changes.
-We now default to a per-system DPI awareness model, which means that we
do not handle WM_DPICHANGED, unless one sets the
GDK_WIN32_PER_MONITOR_HIDPI envvar, where notes for it are in the
following point.
-Automatic scaling during WM_DISPLAYCHANGE is handled (DPI setting change of
current monitor) is now supported. WM_DPICHANGED is handled as well,
except that the window positioning during the change of scaling still
needs to be refined, a change in GDK itself may be required for this.
-I am unable to test the wintab items because I don't have such devices
around.
https://bugzilla.gnome.org/show_bug.cgi?id=768081
Now that GTK+ is built as a single DLL, and the .lib that is built is
gtk-4.lib, we need to update the autotools sections in generating the
NMake Makefile snippets so that we can have the correct commands and flags
for building the .gir files, which will all now link to gtk-4-vsXX.dll (or
so).
As with the autotools builds, use gtk-4 as the name of the .lib file that
is produced from the build.
Actually this is already done with GTK-3.x with the autotools builds,
but this update is not done there as gtk-3.0.lib/gdk-3.0.lib/gailutil-3.0.lib
was used for such a long time that changing it there might have caused
trouble for people there.
Update the project configs to build GDK/GSK as a static lib and include
them into the GTK+ DLL as a monolithic DLL, which is in line with what is
done in the autotools builds, since the code changes needed for Windows
builds for a monolithic build are now in place.
Now that the autotools build folded the GDK/GSK bits into the main GTK+
DLL, there are some updates that need to be done for this. We need to:
-Fold the DllMain() of GDK-Win32 into the main GTK+ DllMain(), as we need
the HINSTANCE to register the window. We can't have two DllMain()'s in a
single DLL.
-Remove the GDK rc(.in) files, as that is not used anymore. Make the GTK+
.rc(.in) file load the gtk.ico GTK+ logo file instead so that we still
get the GTK+ logo for the application icon by default. Update the
autotools build files as well.
-Revert commit b9f9980 as LRN pointed out in comment 25 in bug 773299, as
GTK+ is now a monolithic DLL, and we ought not to export this private
function.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
gdk_wayland_window_attach_image() is normally called from
gdk_window_end_paint() to notify the compositor of newly staged drawing.
If any of the drawing code inadvertently dispatches the wayland event
loop (for instance with a gdk_flush() call), then it's possible that by
the time gdk_window_end_paint() is called, the staged drawing is already
destroyed.
This commit bypasses the attach_image call in scenarios where the staged
drawing is prematurely dropped.
https://bugzilla.gnome.org/show_bug.cgi?id=773274
Commit d249e77 (API: screen: Remove gdk_screen_is_composited()) attempted
to update the GDK-Win32 for the removal of the API, but some parts were
missed. This updates the code so that things continue to build and run.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
gtk/inspector/rendernodeview.c calls this private function from GSK, so we
need to ensure that this function is exported so that GTK+ can link
properly on compilers that do not support automatic exporting.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
This is a problematic struct, and giving direct access to it
has kept us from making improvements to GtkTextView. Drop it
from the public API, together with the auxiliary APIs. If
it turns out that this functionality is needed, we should add
individual getters.
We now record all render operations and display them.
Warning: This is very brute force, you can't clear the recordings or
turn recording off. And this thing easily records 25MB per recorded
frame, so be careful to not run out of memory and get your browser
killed. ;)
This one introduces the Recording object which is essentially a single
instance of something that happened.
The RenderRecording is an instance of an actual rendering operation.
The GLSL versions are:
OpenGL 2.1: #version 110
OpenGL 3.0: #version 130
OpenGL 3.2: #version 150
OpenGLES 2.0: #version 100
OpenGLES 3.0: #version 300 es
So we need to check the version of the GdkGLContext if we want use the
appropriate version, especially for legacy OpenGL contexts, which can be
both 3.x and 2.x.
This ensures that the drawing does not extend the actually drawn area.
It also ensures that our math is sane, because the math assumes the clip
area cannot extend the window. After all, before GTK4 it always was like
that.
Fixes a bunch of drawing bugs when the clip area does indeed extend too
far.
We want to have the coordinate system of the created cairo surface to be
identical to the coordinate system of the node's bounds. For that, we
need to translate the cairo surface by the bounds' origin.
In that case, we can't just rely on the stack allocation being big
enough. Especially, the child can actually be bigger than the current
stack allocation, so take that into account when positioning it.
The typical UI file has a lot more <property> tags than it has
<requested> or <interface> tags, etc. so order the string comparisons
according to this expected case.
We need an overridable entry point for GskRenderer to create Cairo
surfaces.
Implementations of GskRenderer can override create_cairo_surface() to
create efficient surfaces, possibly with zero copies involved, depending
on the GDK backend.
Switch code to use gdk_display_is_composited() instead.
The new code also doesn't use a vfunc to query the property but rather
requires the backend to call set_composited()/set_rgba() to change the
value.
Also add properties for those two properties.
The first property is equivalent to checking if an RGBA visual exists,
the 2nd is equivalent to gdk_screen_is_composited().
This reverts commit 8e29222d95.
This needs more work - spin buttons need to be converted at
the same time, and we should make sure that text still appears.
Update the GDKGL implementation:
-Allow legacy contexts to be created.
-Use finer-grained attributes to ask for a pixel format when possible,
which also adds support for anti-aliasing
In fact the changes here are required for GTKGL to work properly on
Windows for 4.x.
Note that creation of gles contexts is not done here, as the system does
not support such contexts directly on Windows, but only through means such
as ANGLE, which is a totally different issue here.
https://bugzilla.gnome.org/show_bug.cgi?id=773528
Add the needed custom build steps to generate the GResource and
enumeration sources that is needed for the build, and make sure that the
build is able to find the Graphene headers and lib.
Also add the necessary CFLAGS needed for building GSK.
This follows what happens in the autotools builds.
Also fix some project namespaces and the install project where there is a typo
in a project dep, which may cause a problem during the build.
Includes the ability to turn on updates in the inspector. Animations are
now run via a tick function which allows us to neatly overlay a
semi-transparent red rectangle and fade it out over time.
It also probably enables way more, but somebody with more UI neatness
than me needs to figure out what it eanbles first...
Before, we would immediately invalidate the GdkWindow of the widget, now
we call the parent's GtkWidgetClass.queue_draw_child() function.
This allows the parent to track redraw queueing of children.
By default GtkWidgetClass.queue_draw_child() will again chain up to its
parent while respecting the GdkWindow hierarchy for clipping.
GtkWindow is then the only widget actually invalidating the GdkWindow.
This essentially moves redraw queueing from GDK to GTK.
For some reason we end up allocating the colorplane widget
before it is realized, and then never initialize the surface.
Fix this by explicitly doing it on realize.
https://bugzilla.gnome.org/show_bug.cgi?id=773474
This merged gtk, gdk and gsk into one library, making it possible to
have internal private APIs between gtk them, as well as producing more
efficient code.
https://bugzilla.gnome.org/show_bug.cgi?id=773100
This adds the initial MSVC build items needed to build GSK under Visual Studio,
this is part of it that is required, we need to add items to the property sheets
to generate the code that is generated via glib-mkenums and glib-compile-resources.
This set includes, with the autotools scripts for the complete:
-GSK project files, which is integrated into the gtk+-4.sln.
-The NMake snippets to build the introspection files for GSK.
-The .bat files to call glib-mkenums to generate the enumeration sources.
We now need C99 features from the compiler which are only supported by
Visual Studio 2013 and later, so drop the MSVC 2008~2012 projects, and make
the baseline supported Visual Studio version be 2013. Update the build files
as a result.
GTK+-3.89.0 and later will require C99 features that is only supported on
Visual Studio 2013 and later, so prepare build/Makefile.msvcproj for this.
We still keep the 2008~2010 stuff here as this strives to be a shared module,
and there are projects using this that still supports building on pre-2013
Visual Studio.
These complicate a lot of GdkWindow internals to implement features
that not a lot of apps use, and will be better achieved using gsk.
So, we just drop it all.
Add a new ::measure vfunc similar to GtkCssGadget's that widget
implementations have to override instead of the old get_preferred_width,
get_preferred_height, get_preferred_width_for_height,
get_preferred_height_for_width and
get_preferred_height_and_baseline_for_width.
Now that the use_es field is an int with a possible negative value, we
cannot use it its truth value directly; we need to check if it's a
positive value, instead.
GDK defaults to asking for an OpenGL 3.2 Core Profile, but if we get a
legacy profile from the underlying windowing system, the OpenGL version
will be fixed to 3.0. If that happens, we need to set the legacy bit on
the GdkGLContext, since that bit will be used to determine the version
and type of GLSL shaders that will be used by application and toolkit
code alike.
With best-effort, try to use gdk_window_move_to_rect() more often, when
all pieces fit together. For the non-legacy paths to be triggered for
when gtk_menu_popup_for_device() or gtk_menu_popup() were used, the
following conditions must be met:
1) There is no custom positioning function specified
2) The menu is attached to a widget (using gtk_menu_attach_to_widget())
3) There is a associated grab device
https://bugzilla.gnome.org/show_bug.cgi?id=772922
xdg_shell v6 allows grabless popups, whose behavior is not that
different from override redirect windows with no grab to take
keyboard input (and pointer events outside).
This means we can relax the requirement to have a grab before
creating an xdg_popup. The warning is still useful to have so
people stop relying on gdk_window_show();gdk_device_grab() being
an ok pattern to popup a window, it's been moved to wayland
implementation of gdk_device_grab() instead, so we warn if trying
to grab a GDK_WINDOW_TEMP window that's already visible.
https://bugzilla.gnome.org/show_bug.cgi?id=771694
Newer versions of Automake warn about forward compatibility when the
build uses sources in sub-directories without the subdir-objects option.
Both GTK+ and GDK have an almost-but-not-quite non-recursive Automake
layout, with sources in sub-directories contributing to the build of a
top-level object.
In theory, just adding subdir-objects to AM_INIT_AUTOMAKE would be
enough, but the test suite references sources in a different top-level
in order to build tests that verify the implementation of private data
structures. This is not really allowed when using subdir-objects and out
of srcdir builds. In order to fix this case, we require some ad hoc
rules to create symbolic links in the appropriate build directory.
the darker bottom border used on buttons looks bad on circular ones
so now a gradient clipped on the border-box and a transparent
border is used in that partcular case.
See https://bugzilla.gnome.org/show_bug.cgi?id=771205 for details.
$button_fill contains the background-image property value of
buttons, having it readable outside the drawing mixin allows, for
example, stacking background images in an easier way.
This updates all the projects files to be be named appropriately as we move from GTK-3.x to 4.x,
and updates the autotools files so that things are distributed and generated properly.
Also remove deprecated/gtkstatusicon-quartz.c from gtk/Makefile.am, as that was causing 'make dist'
to fail as that file has been removed.
This fixes 'make dist' with the updated existing project files in proper order.
Note that this does not include the new GSK, which will be added later, so the project files do
not yet build the whole stack on Visual Studio at this point.
Make the Makefile.am targets for generating the Visual Studio projects re-generate the
project files and the header listings whenever the Makefile.am's that include
build/Makefile.msvcproj changes, so that whenever a source/header is added, they will
be reflected in the projects and in the property sheets that are used to copy the
headers.
Also ensure that these are applied to the vs11, vs12 and vs14 projects when this
happens, as they are copied and processed from the Visual Studio 2010 projects.
While porting GTK to GskRenderer we noticed that the current fallback
code for widgets using Cairo to draw is not enough to cover all the
possible cases.
For instance, if a container widget still uses GtkWidget::draw to render
its children, and at least one of them has been ported to using render
nodes instead, the container won't know how to draw it.
For this reason we want to provide to layers above GSK the ability to
create a "fallback" renderer instance, created using a "parent"
GskRenderer instance, but using a Cairo context as the rendering target
instead of a GdkDrawingContext.
GTK will use this inside the gtk_widget_draw() implementation, if a
widget implements GtkWidgetClass.get_render_node().
We're going to need to allow rendering on a specific cairo_t in order to
implement fallback code paths inside GTK; this means that there will be
times when we have a transient GskRenderer instance that does not have a
GdkDrawingContext to draw on.
Instead of adding a new render() implementation for those cases and then
decide which one to use, we can remove the drawing context argument from
the virtual function itself, and allow using a NULL GdkDrawingContext
when calling gsk_renderer_render(). A later commit will add a generic
function to create a transient GskRenderer with a cairo_t attached to
it.
Renderers inside GSK will have to check whether we have access to a
GdkDrawingContext, in which case we're going to use it; or if we have
access to a cairo_t and a window.
GskRenderNode is, at its core, a write-only API; you're supposed to set
up the render nodes instead of querying them for state.
Querying render nodes is left to the GskRenderer implementation.
Add a should_propagate function for render nodes. Eventually,
this is meant to avoid creating render nodes for child widgets
that are outside the parents clip area. Since we don't have
that information available right now, just filter out nondrawable
children for now.
Change get_render_node to return nodes that are sized to the clip
area and expect to be placed at the clip position; change
gtk_container_propagate_render_node to place child render nodes
accordingly, and change gtk_css_gadget_get_render_node to return
nodes that are sized accordingly as well.
We store the vertices in (unscaled) window coords (but the item size
is still scaled to match the texture size). Also, the
projection/model-view multiplication order is switched so that the scale
is applied at the right place.
The renderer will always use nearest-neighbor filters because it renders
at 1:1 pixel to texel ratio.
On the other hand, render nodes may be scaled, so we need to offer a way
to control the minification and magnification filters.
If we already have a GL texture we definitely don't want to use
gdk_cairo_draw_from_gl() to draw on a Cairo context if we're going
to take the Cairo surface to which we draw and put it into an OpenGL
texture.
The details of the modelview and projection matrices are only useful for
the GL renderer; there's really no point in having those details
available in the generic API — especially as the Cairo fallback renderer
cannot really set up a complex modelview or a projection matrix.
Instead of using the background as the gadget's node, we add a
non-drawing node that can be used to apply offsets; all other nodes are
children of the "box" node.
Just like we reuse texture ids with the same size we can, at the expense
of a little memory, reuse vertex buffers if they reference the same
attributes and contain the same data.
Each VAO is marked as free at the end of the frame, and if it's not
reused in the following frame, it gets dropped.
The child-transform is useful only if we also provide clipping to the
parent nodes, otherwise children will just be drawn outside of the
parent's bounds.
We'll introduce child transforms either at a higher layer, or once we
add clipping support to GskRenderNode.
This is the first example of indirect rendering involving
a box gadget. For now, we iterate the child gadgets manually,
and rely on gtk_container_propagate_render_node for the
child widgets. Eventually, we may want a better solution
here.
...and implement it for GtkCssGadget and GtkCssCustomGadget.
This allows us to decide on a per-object basis if a custom
gadget needs a render node for content or not.
The custom gadget draw function has the side effect of informing
the gadget machinery wether to draw focus or not. Bring the
draw function back, just for its boolean return value. We may
want to find a better solution for this.
I don't think this should stay in the code long-term, but it
is useful for debugging. It helped me track down some suspicious
placements of render nodes.
Give all nodes the same detail about the owner widget.
This reveals that every GtkCssCustomGadget gets a
DrawGadgetContents node, even if their draw_func is NULL.
We may want to come up with a better solution for that.
When creating the GskRenderNodes for the gadgets we should not translate
the coordinates inside the Cairo context, but we should tweak the
coordinates of the anchor point.
This is still not enough to get an appropriate rendering, as the result
is still slightly offset to the left.
Instead of passing the size of the buffer, we should pass the number of
quads; we know what the size of a single quad structure is, so we can do
the multiplication internally when creating the VAO.
This allows us to print the quads for debugging purposes.
GtkWidget.create_render_node() sets up a GskRenderNode appropriate for
rendering the contents of a widget, including its bounds,
transformation, and anchor point.
The naming is consistent with other scene graph libraries, as it
represents an additional translation transformation applied on top of
the provided transformation matrices.
We can also simplify the implementation by applying the translation when
we compute the world matrix.
We keep the textures used inside a frame around until the end of the
following frame; whenever we need a texture with the same size, and
it's not marked in use, then we just reuse the existing texture.
We were allocating a surface thats big enough for the clip, and
we were setting the transform for that, but then GtkContainer
was overriding the transform with the one for the allocation.
Also, we were drawing at the clip position, not the allocation
position.
This was overwhelming other useful debug output, so make it
opt-in. We print the render items for both opengl and transforms,
since the matrices bleed into each other, otherwise.
Since we use an FBO to render the contents of the render node tree, the
coordinate space is going to be flipped in GL. We can undo the flip by
using an appropriate projection matrix, instead of changing the sampling
coordinates in the shaders and updating all our coordinates at render
time.
We need to apply a scaling factor whenever we deal with user-supplied
coordinates, like:
- when creating textures
- when setting up the viewport
- when submitting the scene
Render nodes need access to rendering information like scaling factors.
If we keep render nodes separate from renderers until we submit a nodes
tree for rendering we're going to have to duplicate all that information
in a way that makes the API more complicated and fuzzier on its
semantics.
By having GskRenderer create GskRenderNode instances we can tie nodes
and renderers together; since higher layers will also have access to
the renderer instance, this does not add any burden to callers.
Additionally, if memory measurements indicate that we are spending too
much time in the allocation of new render nodes, we can now easily
implement a free-list or a renderer-specific allocator without breaking
the API.
We cannot implement GtkWidgetClass.get_render_node() in GtkContainer
without breaking the fallback path that renders a widget to a single
render node rasterization. For GtkContainer subclasses we should provide
a simple API, similar to gtk_container_propagate_draw(), that gathers
all the render nodes for each child.
If a node is non-opaque and has a non-zero opacity we need to paint its
contents and children first to an off screen buffer, and then render the
resulting texture at the desired opacity — otherwise the opacities will
combine and result in the wrong rendering.
We're not going to use separate rendering lists soon, and the way we
render items is less similar to a gaming engine and more similar to a
simpler compositor. This means we don't need to perform a two pass
rendering — opaque items first, transparent items later.
Use appropriate names, and annotate the names with the types — 'u' for
uniforms, 'a' for attributes. The common preambles for shaders are split
from the bodies, so we need some way to distinguish the uniforms and the
attributes just from their name.
We want the GL driver to cache as many resources as possible, so we can
always ensure that we're in a consistent state, and we can handle state
transitions more appropriately.
Drop the texture parameters handling from the texture creation, and
associate them with the contents upload. Also, rename the function to
something more in line with what it does.
We want to add the list of FBOs tied to a texture; this means we cannot
trivally copy the Texture structure when adding it to a GArray. We're
also going to have more textures than VAOs, so it makes more sense to
use a O(1) access data structure for them.
We can use the GL_ARB_timer_query extension (available since OpenGL
3.2, and part of the OpenGL specification since version 3.3) to query
the time elapsed when drawing each frame. This allows us to gather
timing information on our use of the GPU.
The clip rectangle may have non-zero offsets, so we need to ensure that
the GskRenderNode associated to the rendered area is translated by those
same offsets.
For the root node we do not need to use blending, as it does not have
any backdrop to blend into. We can use a simpler 'blit' program that
only takes the content of the source and fills the texture quad with
it.
We should use ShaderBuilder to create and store programs for the GL
renderer. This allows us to simplify the creation of programs (by moving
the compilation phase into the ShaderBuilder::create_program() method),
and move towards the ability to create multiple programs and just keep a
reference to the program id.
We should keep the ShaderBuilder around and use it to query the various
uniform and attribute locations when needed, instead of storing those
offsets into the Renderer instance, and copying them. This allows a bit
more flexibility, once we have more than one program built into the
renderer.
The GL renderer should build the GLSL shaders using GskShaderBuilder.
This allows us to separate the common parts into separate files, and
assemble them as necessary, instead of shipping one big shader per type
of GL API (GL3, GL legacy, and GLES).
GskShaderBuilder is an ancillary, private type that deals with the
internals of taking GLSL shaders from resources and building them,
with the additional feature of being able to compose shaders from a
common preamble, as well as adding conditional defines (useful for
enabling debugging code in the shaders themselves).
Using GObject as the base type for a transient tree may prove to be too
intensive, especially when creating a lot of node instances. Since we
don't need properties or signals, and we don't need complex destruction
semantics, we can use GTypeInstance directly as the base type for
GskRenderNode.
We need a virtual function to retrieve the GskRenderNode for each
widget, which is supposed to attach its own children's GskRenderNodes.
Additionally, we want to maintain the existing GtkWidget::draw mechanism
for widgets that do not implement get_render_node() — as well as widgets
that have handlers connected to the ::draw signal.
This commit changes the way GskRenderer and GskRenderNode interact and
are meant to be used.
GskRenderNode should represent a transient tree of rendering nodes,
which are submitted to the GskRenderer at render time; this allows the
renderer to take ownership of the render tree. Once the toolkit and
application code have finished assembling it, the render tree ownership
is transferred to the renderer.
Whenever the render tree changes we want to drop the RenderItem arrays,
as each item contains a pointer to the GskRenderNode which becomes
dangling once the root node changed.
The surface-to-GL upload logic has become more complicated with the
addition of the GLES code paths; it's more logical to have a public
utility function that can be called from GDK users, instead of copy
pasting the whole thing multiple times.
GSK is conceptually split into two scene graphs:
* a simple rendering tree of operations
* a complex set of logical layers
The latter is built on the former, and adds convenience and high level
API for application developers.
The lower layer, though, is what gets transformed into the rendering
pipeline, as it's simple and thus can be transformed into appropriate
rendering commands with minimal state changes.
The lower layer is also suitable for reuse from more complex higher
layers, like the CSS machinery in GTK, without necessarily port those
layers to the GSK high level API.
This lower layer is based on GskRenderNode instances, which represent
the tree of rendering operations; and a GskRenderer instance, which
takes the render nodes and submits them (after potentially reordering
and transforming them to a more appropriate representation) to the
underlying graphic system.
This is an attempt to get rid of gdk_window_new() for more specific use
cases. These 2 are for client-side windows - regular ones and input-only
ones resepectively.
So far all those functions just call into gdk_window_new().
On windows you might not have a theme installed by default which
means that when trying to create the context quark it will fail.
If then we try to replace a NULL key in the hash table it will crash.
https://bugzilla.gnome.org/show_bug.cgi?id=769859
Just like GLib, GTK+ would benefit from getting warnings and errors from
the compilers.
We check various, common warnings, especially for a future use of C99;
additionally, we promote some warnings to errors, in order to ensure
that simple mistakes are caught during the development phase, before
they are submitted to the code repository.
Use a static array for the known icon sizes, now that we don't allow
registering custom icon sizes any more. This allows us to cut a one-off
allocation that makes Valgrind sad.
The update tracking code was ugly and using deprecated drawing APIs. It
was also in the wrong place.
So instead of trying to keep it working, I'll remove it. We need to find
a better way to put it and make it work there.
This API was only used in GtkModifierStyle and GtkStyleProperties and
they are both on their way out.
CSS properties must now be set using strings via the regular parser API.
And replaces its usages in GtkTextView/GtkStyleContext with a hard-coded
0.04 which was the default value for cursor-aspect-ratio. Also remove
the public gtk_draw_insertion_cursor which used draw_insertion_cursor
which in turn looked up cursor-aspect-ratio
And with it, gtk_widget_get_visual() and gtk_widget_set_visual() are
gone.
We now always use the RGBA visual (if available) and otherwise fall back
to the system visual.
The cursor was set using gdk_window_set_cursor() even in
gdk_window_new().
So instead of having yet another flag, just make the users of that flag
call gdk_window_set_cursor() directly after the window was created.
X11 was the only backend to support it and people can just override it
using XSetClassHint() directly.
The docs already advertised the function as "Do not use".
Keep the existing call to XSetClassHint() in place, so that we keep
setting the same values as in GTK3.
... and gdk_screen_get_window_stack().
Those functions were originally added in
5afb4f0f11 but do not seem to be used as
they are not implemented anywhere but in X.
As GDK is not meant to fulfill window management functionality I'm going
to remove these functions without replacements.
... and gdk_screen_get_width_mm() and gdk_screen_get_height_mm() and
the shortcut counterparts that call these functions on the default
screen.
Modern display servers don't provide an ability to query the size of a
screen or display so we shouldn't allow that either.
Nobody ever does a NULL check there so all that causes is crashes. So
we better return a non-primary monitor than NULL.
Fixes gdk-wayland always returning NULL.
gtk_widget_destroy() removes widgets from their container. However
_internal_ widgets must be unref'ed using gtk_widget_unparent() instead.
This is symmetric with the fact that these widgets were ref'ed by direct
call to gtk_widget_set_parent(). It's also the method that was used in
gtk_headerbar_destroy().
https://bugzilla.gnome.org/show_bug.cgi?id=772859
> Due to Gtk+ keeping a reference to the window internally,
> gtk_window_new() does not return a reference to the caller.
> To delete a GtkWindow, call gtk_widget_destroy().
Caller(s) aren't expecting a need to delete help_overlay themselves
once they've installed it. (E.g. see gtk_application_window_added()).
I didn't notice any direct precedents, but there's a parallel in the
current implementation of gtk_container_destroy() which uses
gtk_widget_destroy() on any added widget.
This avoids leaking 100s of kB per window, when I tested nautilus.
https://bugzilla.gnome.org/show_bug.cgi?id=772859
This allows the use of a "text-direction" hint set to one of "none", "rtl",
or "ltr" to enforce the text direction of a "horizontal-buttons"
display-hint.
This is useful when a menu has buttons that map to physical space in the
UI and therefore must match the application widgetry.
https://bugzilla.gnome.org/show_bug.cgi?id=772775
ClutterEmbed on Wayland uses a subsurface and relocates it on configure
events, but when placed within a scrolled window, no configure event is
emitted and the ClutterEmbed subsurface remains static.
Emit a configure event for native windows in GdkWindow's internal
move_native_children() so that custom widgets relying on configure
events such as ClutterEmbed can relocate their stuff.
Similarly, when switching to/from normal/maximized/fullscreen states
which change the shadows' size and possibly shows/hides a header bar,
we need to emit a configure event even if the abs_x/abs_y haven't
changed to make sure the subsurface is size appropriately.
https://bugzilla.gnome.org/show_bug.cgi?id=771320https://bugzilla.gnome.org/show_bug.cgi?id=767713
to hilight drop target there is a wildcard selector which turns
the border and shadow to green, this clearly shouldn't happen when
the whole window is a drop target.
...by putting it in a stack. The busy_spinner and eject_button are
mutually exclusive, but only the latter was coded to ensure that its
visibility did not cause the rest of the row to reflow. By putting both
widgets in a stack and setting child_visible on that, the row allocates
enough space to show one - or none - at once, avoiding any misalignment.
https://bugzilla.gnome.org/show_bug.cgi?id=772345https://bugzilla.gnome.org/show_bug.cgi?id=772348
Calling eglGetDisplay forces libEGL to guess what kind of pointer you
passed it. Different EGL libraries will do different things here, and in
particular glvnd will do something different than Mesa. Since we do have
an API that allows us to explicitly type the display, use it.
The explicit call to eglGetProcAddress is working around a bug in
libepoxy 1.3, which does not understand the EGL concept of client
extensions. Since it does not, the normal epoxy resolver for
eglGetPlatformDisplayEXT would not find any provider for that entry
point, and crash when you attempted to call it.
Signed-off-by: Adam Jackson <ajax@redhat.com>
https://bugzilla.gnome.org/show_bug.cgi?id=772415
We currently beep when a character is appended at the end in
overwrite mode. That is obviously not right. Patch based on
a patch by Ian MacDonald.
https://bugzilla.gnome.org/show_bug.cgi?id=772389
# YourProject_HEADERS_EXCLUDES = ... # <list of headers to exclude from installation, separated by '|', wildcards allowed; use random unsed value if none>
#
# dist-hook: \ # (or add to it if it is already there, note the vs9 items will also call the vs10 items in the process)
if exist ..\..\..\MSVC_$(Configuration) goto DONE_GDKCONFIG_H
if exist ..\..\..\gdk\gdkconfig.h del ..\..\..\gdk\gdkconfig.h
if exist ..\..\..\GDK_BROADWAY_BUILD del ..\..\..\GDK_BROADWAY_BUILD
if exist ..\..\..\MSVC_$(Configuration)_Broadway del ..\..\..\MSVC_$(Configuration)_Broadway
if exist $(Configuration)\$(Platform)\bin\$(GtkDllPrefix)gdk$(GtkDllSuffix).dll del $(Configuration)\$(Platform)\bin\$(GtkDllPrefix)gdk$(GtkDllSuffix).dll
if exist $(Configuration)\$(Platform)\bin\gdk-$(ApiVersion).lib del $(Configuration)\$(Platform)\bin\gdk-$(ApiVersion).lib
if "$(Configuration)" == "Release" del ..\..\..\MSVC_Debug
if "$(Configuration)" == "Debug" del ..\..\..\MSVC_Release
<GenerateGtkDbusBuiltSources>cd ..\..\..\gtk & $(PythonPath)\python $(GDbusCodeGenCmd) & cd $(SolutionDir)</GenerateGtkDbusBuiltSources>
<GenerateGtkDbusBuiltSourcesX64>cd ..\..\..\gtk & $(PythonPathX64)\python $(GDbusCodeGenCmd) & cd $(SolutionDir)</GenerateGtkDbusBuiltSourcesX64>
if exist ..\..\..\MSVC_$(ConfigurationName) goto DONE_GDKCONFIG_H

if exist ..\..\..\gdk\gdkconfig.h del ..\..\..\gdk\gdkconfig.h

if exist ..\..\..\GDK_BROADWAY_BUILD del ..\..\..\GDK_BROADWAY_BUILD

if exist ..\..\..\MSVC_$(ConfigurationName)_Broadway del ..\..\..\MSVC_$(ConfigurationName)_Broadway

if exist $(ConfigurationName)\$(PlatformName)\bin\$(GtkDllPrefix)gdk$(GtkDllSuffix).dll del $(ConfigurationName)\$(PlatformName)\bin\$(GtkDllPrefix)gdk$(GtkDllSuffix).dll

if exist $(ConfigurationName)\$(PlatformName)\bin\gdk-$(ApiVersion).lib del $(ConfigurationName)\$(PlatformName)\bin\gdk-$(ApiVersion).lib

if "$(ConfigurationName)" == "Release" del ..\..\..\MSVC_Debug

if "$(ConfigurationName)" == "Debug" del ..\..\..\MSVC_Release

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.