This is meant to be used at class-init time,
and will replace bindings, eventually.
We are reusing GtkApplicationAccels here.
It should probably be renamed.
Add a facility to register and install actions
at class init time. The intended use for these
actions is for
a) context and other model-based menus
b) key bindings
Most of these actions are going to be stateless,
so add separate apis for the simple and stateful
cases.
We avoid creating an action group for these by
teaching the action muxer about these actions.
The action muxer also maintains the enabled
state for these actions.
Instead of iterating action groups manually,
just show what the action muxer provides. This
gives an accurate view of what actions are
available in a widgets context, and saves us
the trouble of juggling prefixes manually.
Instead of duplicating this code in multiple places,
add an api to look up an action group for a prefixed
name, and return the unprefixed name at the same time.
This gives us marks to track the duration of processing certain types of
GdkEvent. It also provides some basic struct information in cases where
having that information would likely be useful for debugging.
Page ranges entry can now be focused directly, and will automatically
select the page ranges button when doing so.
This avoids the sometimes counter-intuitive previous behavior where the
entry was automatically focused when toggling the radio button, but the
user may still find themselves clicking uselessly in the text entry
because they scheduled it in their mental model.
Instead, use a new title style class to let
themes influence title formatting. Note that
the theme style will be overridden if the
application uses markup for presentation,
such as <b> or <i>.
Port this from ::populate-popup to the new context
menu api. At the same time, fix some parts of the demo
that were broken since it was changed to use GtkStearchEntry.
Drop the ::populate-popup signal and implement
the new context menu api. Things are a bit more
complicated here, since we have different menus
on links and selectable text.
Menus traditionally don't have separate
hover and focus locations. Make the same
change here that we already did for
popover menubars: Track the active item
and set its selected state. Both keynav
and mouse change the active item.
The Wayland backend has a hack to work around
a race with popover mapping: If the surface size
changes before the initial configure, we hide and
show the surface. Unfortunately, the code was doing
this in a way that is externally observable (by
listening for surface state changes), and popovers
were observing it and hiding themselves in response.
Avoid this by not going through the GDK frontend
code for this.
This solves issues with parent widgets, like combobox
or entry, installing their own bindings for these
keys, overriding the activation behavior that
is implemented in gtkwindow.
This solves issues with parent widgets, like textview
or scrolled window or combobox installing their own
bindings for these keys, overriding the focus behavior
that is implemented in gtkwindow.
Add a convenience api to skip children
that should not be included in the layout,
and call gtk_native_check_resize on all
native children outside of the vfunc.
We can't call gtk_widget_destroy on something
that wasn't added as a child to a container -
it ends up calling gtk_container_remove, which
asserts if it doesn't find the child.
Make gtk_widget_real_focus do the right
thing for focusable widgets with children.
A case where this is (now) relevant is
an entry with a context popover.
Differentiate between wrapping around and
stopping at the end of the focus chain.
Update the existing tests, and add two
new ones where the difference matters.
Add a test that enumerates the focus chain by
emitting move-focus repeatedly, and compares
the result to expected output.
The test expects a ui file and a reference
file as input. The reference file can be created
using the --generate option.
The new rule for focus events from the windowing
system is: We only want them for toplevels. If you
put focus on popups, we don't want to know about
it, and you still need to deliver key events to
the toplevel.
This is expected menu keynav behavior: If the
focused item has a submenu, open it on right
arrow press. And if we are in a submenu title,
make left arrow press close it.
Widgets are supposed to call gtk_widget_child_focus.
Calling internal focus_move function directly makes
us skip the childs ::focus() implementation, which
is where the magic happens.
Make left/right cycle the active item
among the bars children. Separate the styling
for the active item from :hover, since it is
a separate state, and only mixed up in menus
for historical reasons.
Make gtk_popover_new_from_model() return a GtkPopoverMenu,
rename it to gtk_popover_menu_new_from_model() and add
a relative_to argument to gtk_popover_menu_new().
Update all callers.
We can't improve popover menus as long as we
have to be able to work with any old popover
we're given. Remove this, so we can make
gtk_popover_new_from_model return a subclass.
These need to be skipped in measuring,
and we need to call gtk_native_check_resize
on them during allocation.
This was showing up as the new-style
context menu in widget-factory forcing
its relative-to box to grow when its
shown.
We only want to reserve indicator size if
there are any checks or radios in the popover.
Unfortunately, GtkIcon has a hardcoded min-size
of 16, defeating this use. Work around by
wrapping each indicator in a box, and showing/
hiding the actual indicator.
When we are not given an explicit accel (as is
the case when the popover is constructed from
a model), then look it up from the GtkApplication
at map time.
Move checks to the left, and introduce a size group
to align things. The size group is provided by the
parent, using the new ::indicator-size-group property.
This gets us out of using direct presentational
markup like 'inverted' and 'centered' and will
make it easier to play with different layout.
Use the new role when creating popover
menus from models.
We don't need to cover every case with a va_marshaller, but there are a
number of them that are useful because they will often only be connected
to by a single signal handler.
Generally speaking, if I opened into a file to add a va_marshaller, I just
set all of them.
Use the same texture atlases to back both
the glyph and icon caches, and unify their
sizes and management. Store big glyphs
in separate textures, so all atlases have
the same size. Tweak some of the eviction
parameters.
We share the caches across all GL contexts
on a display, unless the GSK_NO_SHARED_CACHES
env var is set.
The logic here seems faulty. We want to keep
a timestamp that tells us when the glyph was
last used, so always update the timestamp.
And whenever we use a glyph, it turns 'young'
again, so remove it from the old pixels
accounting.
The (MAX_AGE, MAX_AGE+CHECK_INTERVAL) interval
is only relevant to prevent us from turning
a cached glyph old more than once, and that
is already taken care of.
It was not copying the terminating \0 in the string, breaking output
in spinbutton (didn't try scale). Fixes#3452.
(cherry picked from commit ae2ef1472c)
This adds specific marshallers for all of the locations where a generic
marshaller is being used. It also provides va_marshallers to reduce the
chances that we get stack traces from perf going through ffi_call_unix64.
This is forward ported from gtk-3-24.
# Conflicts:
# gtk/gtkeventcontrollerkey.c
# gtk/gtkeventcontrollermotion.c
# gtk/gtkgesture.c
# gtk/gtkgesturemultipress.c
If somebody does a transform like
scale(5) scale(10) translate(1,1) translate(5,0)
store it instead as
scale(50) translate(6,1)
This way, less memory is consumed and transforms are easier to read.
In particular, this simplifies the typical transforms we do in GTK,
which are just one translation after another.
We don't need to just look at the scale of the new modelview matrix, but
at the one we get when multiplying the new one with the current one.
Test case attached.
1) In the SetWindowPos() function (and the WINDOWPOS struct) the
"hWndInsertAfter" argument/field means the window that will be
directly above after the change, not the window that will be
directly below. MSDN says "precedes" for SetWindowPos(), but
WINDOWPOS documentation is more precise: this is the window
behind which the affected window will be placed. Apparently,
Z-axis goes back-to-front.
Therefore, logging should be reworded correctly.
2) When we switch away from the application and then switch back
to a transient window, we need to bring up its transient-owner
(and its transient-owner's owner and so forth) as well,
otherwise our transient (modal) window might be transient for
something that might not be visible.
3) When we bring up a window, we should bring all of its children
(popup windows) on top of it.
Because Windows doesn't provide a function to bring one window
on top of the other, we have to work around this by calling
SetWindowPos() twice, swapping the windows between the calls.
Make GtkMenuBar use a box as well,
and let GtkMenuShell get the items
from GtkMenuBar and GtkMenu via
a vfunc. Use that to fix the keynav
implementation in GtkMenuShell.
Similar to previous removals of g_cclosure_marshal_VOID__VOID we can remove
other marshallers for which are a simple G_TYPE_NONE with single parameter.
In those cases, GLib will setup both a c_marshaller and va_marshaller for
us. Before this commit, we would not get a va_marshaller because the
c_marshaller is set.
Related to GNOME/Initiatives#10
The CSS transform should operate on the border-box, not the margin box.
So we need to shrink the bounds by the margin before we apply the CSS
transform.
Use cairo-script-interpreter to parse the scripts that generate cairo
nodes.
This requires libcairoscriptinterpreter.so to work properly, but if
it isn't found we disable this (unimportant for normal functioning)
code and just emits a parser warning.
The testsuite requires it however or it will fail.
A new test is included that tests all of this.
This uses the new sysprof-3 ABI to implement the capture writer. It also
uses the statically linked libsysprof-capture-3.a that is provided with
Sysprof for the capture writing to ensure that we do not leak any symbols
nor depend on any additional libraries.
The GTK_TRACE_FD can be used to pass a FD for tracing into Gtk. Sysprof
uses this when the Gtk instrument is selected for recording.
We were deferring the reflow until map, but this
leads to the section initially having an enormous
height and the window picks up that size before
we have a chance to reflow, This could be seen
in the "Builder" demo in gtk4-demo.
Closes: https://gitlab.gnome.org/GNOME/gtk/issues/11
If we set c_marshaller manually, then g_signal_newv() will not setup a
va_marshaller for us. However, if we provide c_marshaller as NULL, it will
setup both the c_marshaller (to g_cclosure_marshal_VOID__VOID) and
va_marshaller (to g_cclosure_marshal_VOID__VOIDv) for us.
With gtk_window_set_position gone, we should never
come up with a new position to set in this code.
Leave a warning in place and remove the gdk_surface_move
calls.
Root coordinates are going away, so this
api does not make sense anymore. Use
gdk_surface_get_device_position instead.
We still keep this as internal api for
root-coordinate using backends.
The X backend was storing global coordinates
in surface->x/y, and keeping the parent-relative
positions in its own fields. Switch this around
to store the relative position in x/y, as is
expected by the frontend.
This function returns the position relative to
the surface parent, so will always return 0 for
non-popups. The out arguments don't need to
allow-none either - nobody passes NULL for these.
We maintain offsets for popups, so we can translate
coordinates between surfaces that are attached directly
or indirectly to the same toplevel. Add an api for that.
Make the transform (transfer full).
1. This makes sure we actually reference the transform. Previously we
did not.
2. Most callers create a new transform to pass to us. Now they don't
have to uref it anymore.
CSS does not do exponents, so printing numbers close to 0 as 1.234e-15
does not work.
Also up the accuracy to 17 digits because that's what everyone else
uses.
GTK4 doesn't have WS_CHILD windows anymore, so hWndParent argument
to CreateWindowEx() is always interpreted as the owner window,
not the parent window.
A window with an owner:
* is above the owner in Z-order
* is destroyed when the owner is destroyed
* is hidden when the owner is minimized
This is enforced by the OS.
GTK can only allow this for popup windows.
Desktop window must never[0] be an owner.
[0]: https://devblogs.microsoft.com/oldnewthing/20040224-00/?p=40493
On Windows that call resizes the native window immediately,
and the corresponding GDK event is emitted and processed
before the control is returned to gtk_window_move_resize().
Therefore, update freeze and configure_request_count increment
must happen before the call, not after it.
Popups can't be active or inactive, so emitting GDK events
in response to WM_ACTIVATE makes no sense for these kinds
of GDK surfaces.
The jury is still out on whether we should block (return 0)
or ignore (don't return anything) this message.
Blocking WM_NCACTIVATE (which we currently ignore) is definitely
not an option - it completely breaks input somehow.
1) Handle GDK_SURFACE_POPUP in RegisterGdkClass()
(for now pretend it's the same as GDK_SURFACE_TOPLEVEL)
2) Remove useless code from GDK_SURFACE_TOPLEVEL case in _gdk_win32_display_create_surface()
(now there's just GDK_SURFACE_TOPLEVEL there, no need for a type check)
3) Have a separate case for GDK_SURFACE_POPUP and ensure that
it doesn't get WS_CHILDWINDOW (and neither should GDK_SURFACE_TEMP).
We were queuing an allocation whenever
the popover is flipped over, unnecessarily,
since we don't change the size of the surface.
This was showing up as popovers being invisible
when flipped over, under X.
There is no need for popups to connect to the frame
clock to pause and unpause events on the display -
the toplevel already does it.
And don't connect to paint either - handle paint
on popups recursively.
Somewhat change the order of initialization (to be closer
to what Wayland backend does).
Also remove the wrapper field that is no longer needed -
it used to hold a pointer to the main GdkWindow instance,
which wrapped GdkWin32ImplWindow. Since impls are gone,
nothing is wrapping anything anymore.
Fix a substitution error, where wrong pointer was added
to the hash table. Added a comment to ensure that future readers
(including myself) won't be confused by the fact that we're
inserting a pointer instead of the handle itself.
Since we are now sharing frame clocks with multiple
surfaces, we can no longer dispose them unconditionally
when a surface goes away. Only do it if we are a
toplevel (without parent).
This was showing up as criticals on exit when opening
and closing any popover in widget factory.
Now that popups share the frame clock of their
parent, we have to be much more careful about
freezing the clock, since that may stop updates
for another surface.
This commit makes two changes that make the
X11 handling of the frame clock more similar
to the Wayland backend:
- Use gdk_surface_freeze_updates instead of
gdk_surface_freeze_toplevel_updates to avoid
affecting the frame clock
- Bail out early in before_paint/after_paint
if the surface is frozen, to avoid affecting
the frame clock
Together, these two make the X11 popup surface
type work without freezing updates for the toplevel.
With separate clocks, the phases are not coordinated,
which messes with GTKs size allocation machinery treating
the entire widget tree as a whole, and causes us to
run into assertion where popups get drawn before they
are allocated.
Call gtk_native_check_resize() from size_allocate,
as is required now. This gets volume buttons closer
to working again (dragging the slider still doesn't
work).
I can't think of a case where this is the desired
behavior. So, instead of setting an explicit cursor
on all popups, just stop walking the parents at
surface boundaries.
When a GtkNative widget is marked as resize_needed,
we need a current position for its parent and we need
the parent to be allocated (so we can position our
surface), but we don't need the parent to be marked
as resize_needed, since the parent size is entirely
independent of the popup size.
When we print warnings about a widget, using
gtk_widget_get_name() is slightly better than
G_OBJECT_TYPE_NAME(), since it will give us
the widgets unique name when available.
This api is meant to mimic xdg-popover.grab - we
show the surface, and dismiss it when we get events
on other surfaces. For foreign surfaces, the compositor
handles that for us; for our own, we check outselves
before delivering events to GTK.
Make them use o-r windows, and move
with their parent.
We do a sort-of ok job on stacking order
here - whenever the parent window gets a
ConfigureNotify, we just restack all popups
directly on top of their parent. This is good
enough to keep popups on top of their parent
while we drag it around, and it gets the popup
to disappear when raising another window on
top of the parent.
Store popup parents separately from transient-for
parents, since these are separate concepts with
different behaviors. And we need the parent in
the frontend, so we can use it in the fallback
move-to-rect implementation.
We don't need the complicated wrapper system anymore,
since client-side windows are gone. This commit moves
all the vfuncs to GtkSurfaceClass, and changes the
backends to just derive their surface implementation
from GdkSurface.
We want to use a gdk_surface_new_popup for popups,
and align the constructor names with the surface
types, so rename
gdk_surface_new_popup -> gdk_surface_new_temp
gdk_surface_new_popup_full -> gdk_surface_new_popup
The temp surface type will disappear eventually.
All the information in it is already contained
in the surface object we pass along, and none
of the backend implementations were using the
attributes at all.
We are not creating such surfaces anymore, and
they were only ever meaningfully implemented
on X11. Drop the concept, and the api for determining
if a surface is input-only.
We still need to keep the vfunc around, since the
fallback implementation of gdk_display_get_monitor_at_surface
uses it. So, a GDK backend must either have root coordinates
or always return a monitor from monitor_at_surface.
We still need to keep the vfunc around, since the
fallback implementation for move_to_rect uses it.
So, a GDK backend must either have root coordinates
or implement move_to_rect.
This lets us remove a use of GTK_WINDOW_POPUP,
which should eventually be going away.
We need to disable treeview search, since it
creates a toplevel that will disrupt our grabbing
popup, causing it to be dismissed.
We don't need to grab ourselves, since the popover
code does it for us. We don't need to reposition our
window, since the popover takes care of that too.
Start by adding a constructor. We have to call it
gdk_surface_new_popup_full for now, since gdk_surface_new_popup
is taken. This may be reshuffled later.
Now that roots can have parent widgets, we need to
carefully examine all calls of gtk_widget_get_toplevel,
and replace them with gtk_widget_get_root if we want
the nearest root, and not the ultimate end of the parent
chain.
... when a window gets hidden and later reshown.
The code now properly cleans up a window when it gets unmapped instead
of trying to retain previous updates information.
The 'documentation' option also guarded the man page build. Instead
if skipping the whole docs subdir skip the specific gtkdoc calls, so that the
man page build still works.
This brings it in line with the gtk3 meson build.
The “xdg-output” protocol provides clients with the outputs size and
position in compositor coordinates, and does not provide the output
scale which is already provided by the core “wl_output” protocol.
So when receiving the wl_output scale event, we should update the scale
regardless of “xdg-output” support, otherwise the scale will remain to
its default value of 1 and the surface will be scaled up by the
compositor to match the actual output scale, which causes blurry fonts
and widgets.
Fixes: https://gitlab.gnome.org/GNOME/gtk/issues/1901
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
The code interprets NULL to mean 'create a new Adjustment and use that,
deposing the old one', but we neither documented nor annotated that
ability, so users could be unsure how to "unlink" a specific Adjustment.
While users could pass their own Adjustment in if this wasn't nullable,
we already support NULL in the code, and it doesn't hurt to document
that this is available as a convenience while retaining the behaviour.
GdkAtom is a typedef to a pointer to an opaque structure. We need to
tell GTK-Doc how to override it, so that the documentation is accurate.
Fixes: #302
We blindly assume everywhere that a single glyph will definitely fit on
one atlas, but that's not always the case.
For now, don't crash or produce GL errors.
The GTK_DISPLAY_DEBUG_CHECK macro will cleverly only call the function
if any of the display debug flags are set, so in the common case it
won't even be executed.
Test that rendering empty nodes succees. For a lot of nodes the
resulting rendering isn't clearly defined, in those cases we overdraw
those regions (sometimes the whole image) with black.
- Remove remains of g_test_*() functions
We're not a glib test, we're a simple binary.
- Handle nonexistence of reference image properly
Don't assert, but create the output image and the error out.
Instead of only allowing for glyph indexes, allow ASCII characters as
replacements. So this glyph sequence
glyphs: 65 8, 66 8, 67 8
Can be replaced by
glyphs: "ABC"
provided that the glyph for "A", "B" and "C" are 65, 66 and 67
respectively and their advance is exactly 8.
x offset and y offset must always be 0 and every glyph must start a
cluster.
Update to the docs outlined in #1887.
In particular, the changes do:
1. Require no property, have a working default for everything
2. Be clear about what gets printed and how.
Tests ahve been adapted to still pass.
Base the rewrite on testsuite/css/parser/test-css-parser - we now
require the node file to match a reference node and track the errors it
triggers.
We also no longer use gtester.
When printing, behave the same way as when parsing:
Magically skip a container node if there is one - just like the
parser magically creates a container node to hold all the nodes
it parses.
We can't just pretend we have an identity matrix when we are actually
scaling. This fixes the node editor sometimes not drawing things when
rendering to a texture. We were mistakenly discaring render nodes
because the bounds transformation was wrong.
Ensure that the class structs of all subclassable
types have sufficient padding (standardizing on 8
slots, here).
GtkBox
GtkButton
GtkDrawingArea
GtkFixed
GtkFrame
Ensure that the class structs of all subclassable
types have sufficient padding (standardizing on 8
slots, here).
GtkApplication
GtkWidget
GtkContainer
GtkWindow
GtkDialog
GtkApplicationWindow
GtkToolItem
GtkBin
The change to keep some server resources around
until destroy was causing us to not recreate
the right things when a surface is hidden and
then shown again. Make sure to recreate everything.
The Wayland backend was dropping _all_ serverside
resources on hide, which is too early e.g. for
GtkGLArea which wants to use egl resources to
unload textures on unrealize.
Closes: https://gitlab.gnome.org/GNOME/gtk/issues/1485
This is not an api we want to propagate anymore.
If you need to, you can still emit the "clicked"
action signal on a button using g_signal_emit_by_name.
We need to store the region *before* adding our own damage area, because
we want to only store the changes of this frame, not the whole history.
So do it in the same place Vulkan does it.
Fixes#1900
Make GtkMenuButton a widget that has a
toggle button, instead of deriving from it.
We give it icon-name and label properties,
to let people do what they expect to do
with menu buttons.
We stuff both gl-drawn and cairo-drawn textures into the same cache, so
we can't really assume that we need to draw any of them flipped or not.
Fix this by drawing fallback stuff upside down and then using
upside-down vertex data for everything.
Fixes#1897
I don't want to install a new set of bindings every time someone
attaches a reproducer to a bug. I also don't want to rewrite sait
reproducer in C every time just to eliminate the possibility of broken
bindings.
We were adding incomplete frame timings to the
profile, which lead to occasional nonsense
numbers. Instead, only add timings to the profile
once we marked them as complete. This also
gives us an opportunity to add the presentation
time as a marker.
If we want the model items to be listmodels, we don't need to do a ==
comparison. We need to do g_type_is_a(). Implementations of listmodels
are obviously fine.
Besides requiring it at build time, require that the server the client
is running against exposes the XInput2 protocol. We no longer fallback
on a device manager for core events.
XInput2 is more than a decade old already, and the input improvements
there (and in every other backend really) make it untenable to have
support for X11 core input events dragging things behind.
This fixes a long standing bug in pid_get_parent on OpenBSD (which was mine
so... my fault). kp wasn't properly allocated and the function could return
random failures.
The key controller was consuming key events
for modifier keys, for no entirely convincing
reason, which leads to problems when somebody
actually listens for those, such as the simple
input method does for C-S-u processing.
Closes: https://gitlab.gnome.org/GNOME/gtk/issues/1884
There are two ways GTK can add a headerbar to a dialog:
- the dialog is constructed with the :use-header-bar property
- all windows should use client-side decorations
In the first case, the headerbar is added by GtkDialog with no
dedicated style class, and in the latter by GtkWindow with the
"default-decoration" style.
As a result, dialogs with plain titlebars can end up with clearly
distinct and inconsistent styles.
To address this, allow headerbars to track whether they should use
the "default-decoration" style and enable it for dialogs.
https://gitlab.gnome.org/GNOME/gtk/merge_requests/836
We don't want to return a GFile because GFile can't handle can't deal
with data: urls.
That makes the code a bit more complicated that doesn't deal with those
URLs, but it makes the other code actually work.
GtkCssImageUrl also now decodes data urls immediately instead of only at
the first load. So don't use data urls if you care about performance.
When we're at the end of a block and gtk_css_parser_get_token() returns
NULL, gtk_css_parser_commit_token() still consumed the next token.
It does not anymore.
This does not affect the CSS parser, but it exposes issues with the
render parser, which previously just consumed too many closing } tokens
in the past.
The calls used old bugzilla URLs and nobody cared about that.
So apparently they are very unused.
There's also a potential conflict between gitlab and bugzilla URLs and
what base bug to use there.
The old usages have been converted to comments.
Instead of encoding the raw data, encode the full image to a PNG.
And instead of stuffing that encoding into a string, use a full
data: url.
And then remove the width and height properties, because they're now
implicitly included in the data.
And then change the parser to match.
And because the parser now parses regular urls on top of data: urls, we
can now load any random file.
When looking for versioned theme files, we were
looking for directories names gtk-4.x for x
bigger than 14, which mades sense for GTK 3,
but we are starting out at 0 again, so remove
this check.
GdkSurface::set_startup_id() is NULL on Win32 and would cause a segfault
if called.
While the documentation of the main caller of set_startup_id(),
gtk_window_set_startup_id(), mentions that it's not implemented on
Windows it can still be automatically called via Glade and simply doing
nothing on Win32 is going to be less disruptive than a segfault.
If SYSPROF_TRACE_FD is set in the environment,
interpret it as an fd to write profiling data
to.
If GTK_TRACE is set, write profiling data
to a file with name gtk.$PID.syscap.
This is writing data in the capture format of sysprof,
using the SpCaptureWriter. For now, this is using a
vendored copy of libsysprof. Eventually, we want to
use the static library that sysprof provides.
The http* family of functions was deprecated after CUPS 1.7. We can
conditionally use it when built against a newer version of CUPS. The
additional parameters are taken directly from the fallback values
inside CUPS itself.
The inspector, and other tools introspecting the widget structure like
gtk-builder-tool and Glade, may very well want to access the default
layout manager used by a class, especially if there are layout
properties involved, without having a whitelist of widget/layout manager
associations.
Some widgets have a well-defined layout manager created alongside their
own instance; if they do, we can handle the layout manager creation at
the GtkWidget instantiation.
The default value of GtkWidget::visible changed
from FALSE to TRUE from GTK 3 to 4. Make --3to4
deal with this by ensuring the visible property
is explicitly set, before simplifying.
This adds a test tool gsk/node-parser that takes node files and parses
them.
A few of these node files have been added, for crashes I encountered while
developing the new parsing code.
Instead of the previous approach using GVariant, this new approach uses
human-readable text files as the serialization format for render nodes.
The format is a custom one, but it is inspired by QML and conforms to
the CSS syntax. Because of that, we can use the CSS machinery from GTK
to parse it, and in particular share code to parse properties that GTK's
CSS machinery also supports, such as colors.
This commit breaks all existing usages of node files - such as the
testsuite and various test tools - they will be fixed in further
commits.
We were using one Emoji chooser when triggered
via the context menu, and another one when
triggered via the icon.
Change things to always use the same Emoji
chooser instance.
We had code in gtkwindow.c that generated duplicate,
and defective, focus-change events, in the following
way:
- gtkmain.c generates a chain of focus-change events
for moving focus from one window to another
- gtkwindow.c catches a focus-in event in the middle
of this chain and sets itself as 'active'
- and then it proceeds to generate focus-change
events towards its own focus widget without a
related target
This is not necessary since we gtkmain.c already
generates a complete sequence of focus-change events.
So stop doing it.
This property has a 'smart' default that depends
on the class of the object we're creating. Take
that into account when deciding whether to omit
properties that are set to their default value.
If the recent section is empty, we want to hide
it, make its button insensitive, and select the
next section, initially. This was not working
properly, since nothing was ever setting the
section box to invisible.
It takes half a second on my system to initially
populate the Emoji chooser. That is too long. Do
the work in 8 millisecond chunks to give GTK a
chance to get some frames done.
Change the way we compute border color cutoffs to the same method that
browsers use. This method does not consider the corner sizes at all and
only looks at border-width.
Previously, when borders were too big - ie when a 100x100 rect had only
one 100x100 border, like the black part of ◔ - and then shrinking this
rect by 25px on either side, we'd end up with a 50x50 rect with a 75x75
border, and that's obviously not correct.
We were hardcoding that GtkApplicationWindow only
has a single action group with prefix 'win', but
that is no longer the case. Simply use the code
for the general widget case that can handle multiple
action groups.
I was a little overzealous when going
for the new default handling here. We
can't switch to gtk_widget_activate_default
before we actually handle the default.activate
action.
Display changes now happen exclusively through
the ::root and ::unroot vfuncs. Third parties
can observe display changes by listening
for notify::root.
Export gtk_widget_root/unroot privately,
make them work on roots, and use them in
gtk_window_set_display. This gets us to a
single way to listen for display changes,
the root property.
Currently, we sometimes emit display-changed
when the display changed, and sometimes we don't
since the display is changed via gtk_widget_root.
Listen for notify::root as well and update our
display. This is a temporary fix - all display
changes should go through gtk_widget_root,
eventually.
apis that takes multiple display-relative objects
should make sure that they are all from the same
display, or hard-to-track-down badness will happen
later on.
Add such a check for the surface and device arguments
of gdk_seat_grab. This helped in tracking down
critical warnings from combo boxes in the inspector.
The specification for border-radius goes
top-left top-right bottom-right bottom-left.
The css for the add button in the color
chooser got this wrong, and was showing
a broken top-right corner.
Closes: https://gitlab.gnome.org/GNOME/gtk/issues/1856
We are now getting focus-out and focus-in events
when the Emoji chooser is shown and hidden, and
this is causing the text to select-on-entry before
inserting the Emoji, which then deletes the selection.
Avoid this by saving and restoring the selection
when presenting the Emoji chooser.
That a property can't be set does not mean
its value can't change. This was showing up
as the cursor-position and selection-bound
properties in GtkText not showing their
current value in the inspector.
When a modal dialog is smaller than its parent,
we were keeping the resize cursor from the dialogs
edge all over the parent window, which looks
really irritating, since the resize cursors are
closely associated with the window edge. Fix
this by falling back to the default cursor
outside the grab widgets surface.
Closes: https://gitlab.gnome.org/GNOME/gtk/issues/23
We don't support that setting life-updating anyway. So, instead of
getting the value *every time we format a time*, get it once for the
filechooserwidget and reuse that value.
The spinbutton>button>image is currently blue when the image is clicked
and dark-ish when the button is clicked(but not the image). This was not
the case before since we didn't even propagate :active down to the image
child. Fix this by only applying the blue color to direct image children
of entries.
The old default of 148px doesn't work everywhere. Instead, pick a
default value of -1 and measure() the sidebar widget in the
filechooserwidget in that case. Other values >= 0 are still handled as
before.
Leave it to the ::hits-added handler to switch to the list of search
hits. This way we don't get a weird transition when the current search
didn't have any hits and the next one doesn't either.
Searches with hits still feel good.
Now that GdkSurface has properties, it makes
sense to turn the frame clock into one too.
This will make it easier to reshuffle some
of the surface constructors later.
Set the cursor on the surface of the target
widget, not the surface of some of its parents.
This does not make a difference currently.
But it will in the future, when we have
parented widgets with surfaces.
Make find_grab_input_seat return a GdkWaylandSeat
instead of a struct wl_seat, so we can use it and
avoid calling gdk_display_get_default_seat when
we need to get a serial later.
Floating point values cannot ever be compared for equality. GLib has a
G_APPROX_VALUE macro that lets us compare two value within a provided
precision, so we should use that instead.
Artisanal, homegrown, locally sourced, vegan reference counting has been
replaced by the appropriate API in GLib, which does small things like
saturation and type checking.
The default widget is mostly a dialog concept,
and does not really need this generic api.
If you need to mark a widget as default,
use gtk_window_set_default() directly.
We used to handle has-default specially in ui
files. It was awkward, so stop doing that. If you
need to influence the default widget in a window,
you can just set the default-widget property.
Instead, use the new way of activating default.
I think most of the default handling in
GtkFileChooserDialog should be dropped, but
for now this keeps things working.
Activating this action will replace other
activate_default apis. It is more flexible,
since intermediate widgets can intercept the
action and do their own handling.
This api wasn't used anywhere in GTK. And since
we've dropped the variant for the default widget,
this one should go too. If it is needed, it should
become and action too.
We used to handle has-focus in ui files specially.
It was awkward, so stop doing that. If you need
to influence the initial focus of a window, you
can just set the focus-widget property.
When hitting Escape in the location entry,
we were not moving the focus anywhere,
causing focus to be NULL, and key bindings
to stop working. The visible effect was
that Ctrl-L / Escape / Ctrl-L would not
get back to the location entry, as expected.
Closes: https://gitlab.gnome.org/GNOME/gtk/issues/1851
When hitting Escape, the file chooser will go
into search mode, because the search entry
consumes the key to emit the ::search-stopped
signal. Recognize this situation and avoid
switching to search mode in this case.
Closes: https://gitlab.gnome.org/GNOME/gtk/issues/1850
The primary icon is on the right in RTL, so don't offset the text child
to the right for it.
Fixes the text and icon overlapping in the second entry
in tests/testentryicons
It doesn't make sense to keep track of all the last_* values anymore now
that widgets only get allocated when their size changes anyway.
Remove all the associated (and thus now unused) flags as well.
We were forwarding key events to the search entry
and unconditionally considered search started
afterwards. That is not correct, since things
like a Ctrl key press should not trigger search.
Fix this by only switching to search mode when
the event was actually consumed.
Closes: https://gitlab.gnome.org/GNOME/gtk/issues/1829
The key capture was interfering with other
entries in the dialog, so be smarter about
when we want to capture keys and when we
don't.
Closes: https://gitlab.gnome.org/GNOME/gtk/issues/1842
Apparently genTextures and friends only "reserves names", initializing
them will actually create them. Using glObjectLabel on textures before
initializing them will throw a GL_INVALID_VALUE error.
Just measuring it (so the warning goes away) but then not using the
values will later underallocate the text widget.
Instead, always reserve space for the icon (which will inevitable be
visible as soon as the searchentry is actually being used).
Fixes#1831
When rendering to a texture, collecting the render ops might bind a
different framebuffer, so bind the one we want again before doing the
actual rendering.
This adds debug groups in various places, including the debug
nodes if those are in use. This makes the traces in tools like
renderdoc much easier to read.
GL keeps the unoform state per-program, but not per-frame. So, we can't
pretend that this works for us. Keep the RenderOpBuilder around for the
entire lifetime of the renderer instead.
Added two new private GtkWidget API:
* gtk_widget_add_surface_transform_changed_callback()
* gtk_widget_remove_surface_transform_changed_callback()
The intention is to let the user know when a widget transform relative
to the surface changes. It works by calculating the surface relative
transform during allocation, and notifying the callbacks if it changed
since last time. Each widget adds itself as a listener to its parent
widget, thus will be triggered if a parents surface relative transform
changes.
This is a meson test, not a GTest thing. So:
- Use g_print(), not g_test_message
That makes meson test --verbose print the actual log messsages.
- Don't g_assert() all the time
Instead, run tests through to the end and just return a non-0 exit
status.
This fixes rendering to a texture on intel hardware. The glClear calls
would throw a GL_FRAMEBUFFER_INCOMPLETE error here, because the
gsk_gl_driver_begin_frame() call in do_render() reset the framebuffer
object in use.
Save the information whether the cursor in use is the default one, and
don't create a new cursor object in that case.
We previously created a new cursor object every frame just to compare it
to the current cursor in use and then throw it away.
The skip-taskbar, skip-pager and urgency hints were
only ever implemented for X11, and are not very useful
with modern desktops. Relegate the functionality to
x11 backend api, and drop the GtkWindow api.
This fixed the reftest introduced in the previous commit.
I'm using a mesh gradient here instead of drawing 4 individual sides to
avoid artifacts when those sides overlap in rounded corners.
And update the surface accordingly (eg. scale on hidpi). The mechanism
that did that for wl_pointer has been made generic so it can be shared
with tablets too.
Closes: https://gitlab.gnome.org/GNOME/gtk/issues/1675
_gdk_wayland_cursor_get_buffer was not initializing
its out variables in the 'not found' case. This
was showing up in protocol traces as garbage hotspots
being sent to the compositor.
Closes: https://gitlab.gnome.org/GNOME/gtk/issues/1328
gtk_css_provider_get_named() is the old GTK3 style API to load themes.
Instead, export the function we currently use,
gtk_css_provider_load_named().
As a side effect we allow people to load a theme as often as they want
without conflicting with GTK's theme.
Previously, the GDK backend for Wayland would deduce the logical size
of the monitors from the wl_output size and scale.
With the addition of fractional scaling which advertises a larger scale
value and then scale down the client surface, the computed logical size
of the monitors in GDK would be wrong and confuse applications which
insist on using the monitor size and position (like Firefox).
The xdg-output protocol aims at describing outputs in a way which is more
in line with the concept of an output on desktop oriented systems by
presenting the outputs using their logical size and position appropriately
transformed.
Add support for the optional xdg-output protocol so that the size and
position of the monitors as reported by GDK is correct even when using
fractional scaling.
Fixes: https://gitlab.gnome.org/GNOME/gtk/issues/1828
These flags check for code that we don't want to write, so turn them
into error flags.
Variable length arrays should be replaced by malloc() - or explicit
alloca() calls if you know what you're doing.
Implicit fallthrough should be replaced by explicit fallthrough with the
usage of G_GNU_FALLTHROUGH.
This work inspired by Kees Cook's LCA2019 talk:
https://www.youtube.com/watch?v=FY9SbqTO5GQhttp://outflux.net/slides/2019/lca/danger.pdf
We should show all the possible result sections:
- passed
- skipped
- expected failures
- failures
- timed out
Even if we consider the first three to be successes.
My web design skills are terrible, but it's better than nothing; I'm
sure this will lead to somebody filing a merge request to make the
test report look a lot better.
While we're at it, let's include the reftest images inside the report
itself, so we don't have to hunt them down.
HarfBuzz 2.0 deprecated some API used by the GtkFontChooser, but since
we're still supporting older versions of HarfBuzz, we should disable the
deprecation warnings to avoid too much noise during builds.
The JUnit cover report is useful, but only up to a point; for instance,
it's not used unless it's part of a merge request. This means you don't
get a report if you're pushing to a branch that does not have an MR open.
With a simple Python script and some minimal templating, we can generate
an HTML report from the "I Can't Believe it's not JSON™" log that Meson
produces, and keep it as a CI artifact.
The nice thing about that is that we can then log messages about the
errors to the log.
And then we can read the logs of the CI machinery and actually know
what's going on.
before, code was using the "C" locale, but that one uses ASCII. Instead,
run in the "C.utf8" locale.
Nobody expects code to not support UTF8 and no end user runs their
machine in an ASCII setup, so it makes no sense to default to that.
We're passing integers without validating their size, and newer GCC are
very cross about it, with warnings like:
warning: argument 1 range [18446744071562067968, 18446744073709551615]
exceeds maximum object size 9223372036854775807 [-Walloc-size-larger-than=]
We should check we're not overflowing the allocation size, by limiting
the range of values we can use.
First of all, we need to use `gsize` instead of a random `int`, since we're
allocating data.
Additionally, we need to check that the multiplication that computes the
size of the allocation doesn't overflow the maximum value of a `gsize`.
They should be fixeed before 4.0 but the fixes are more involved. And we
want to start running the existing tests on CI, because they break
regularly and we want to catch that.
* :nth-child(first) => :first-child
* :nth-child(last) => :last-child
* Add semicolons at end of declarations
* Remove spaces between color functions (shade, alpha, ...) and args
That test was cool in 2011, but hasn't been updated or used since then
because its features are now part of widget-factory and the inspector.
So let's remove it.
Instead of just checking that the line of the error message is correct,
assert that start and end position are on the correct character offset.
Also fix all the tests to conform to this.
Make the test use an actual integer property that accepts negative
numbers (opacity) instead of one that wants units (margin-top) or
can't deal with negative numbers (everything else).
Emit all errors via the parser, don't try to have a custom error
handling machinery.
The only exception is the initial file load error - we need to do that
one directly, because there is no parser.
This commit is still way too big, but I couldn't make it smaller.
It transitions the old CSS parser to the new parser. CSS parsing is now
tokenized, everything else is probably still buggy.
This function is a (private) function to parse a GdkRGBA accoridng to
the CSS specs. We should probably use it for gdk_rgba_parse(), but that
would change the syntax we accept there...
This also introduces a dependency of libgdk on libgtkcss.
So far, no users for this function exist.
We can't try to get an integer because ultimately integer getters
support the same shenanigans that numbers and percentages do with calc()
and whatnot.
As part of that, adapt the syntax from
-gtk-scaled( [<image>, <int>?]# )
to
-gtk-scaled( [<image> <int>?]# )
because the commas should be used to separate distinct elements.
Note that almost nobody specifies the scale anyway.
This is ithe first step towards converting the parsing code to use
tokens. For now, the topken type is just a magic enum value that only
works as-needed.
Instead of an error vfunc, have the tokenizer vfunc take a GError
argument. Note that even when an error is returned, there is still a
token to be read.
This library is meant to be the new CSS library that gets used from GDK,
GSK and GTK for string printing and parsing.
As a first step, move GtkCssProviderError into it.
While doing so, split it into GtkCssParserError (for critical problems)
and GtkCssParserWarning (for non-critical problems).
Just like GtkContainer provides a default implementation of
GtkWidgetClass.get_request_mode(), we can do the same inside
GtkLayoutManager.
A default implementation preserves the behavior of existing widgets that
moved, or will move, to a GtkLayoutManager.
Since we're embedding text coming from the tests into the report, we
should specify an encoding for both the source JSON file and the target
XML file when opening them.
The default GtkWidgetClass.get_request_mode() is implemented by
GtkContainer; now that GtkBox uses a GtkBoxLayout, we need to implement
it inside the layout manager to preserve the same behavior as the old
GtkBox.
Fixes#1821
The current Meson releases have broken CMake support, meaning that it is
likely that HarfBuzz could not be located for Visual Studio builds
unless one handcrafts pkg-config files for it, which is both tedious and
error-prone.
Instead, use the existing mechanism for looking for the HarfBuzz headers
and libraries on Visual Studio first when it could not be found via
dependency(), and then use the fallback if it still could not be found.
We don't want to stop people from being able to debug GTK applications
by default.
The keybinding also runs last in event delivery, so it doesn't override
existing keybindings anywhere and is therefor safe to enable.
The setting of course should remain, so people who want to lock down
installations, like for kiosks, can turn this off.
Since commit 3b2f9395, the frame time may be set into the future, so
only ensure monotonicity, and don't store the offset. This prevents the
frame time from becoming out of sync with g_get_monotonic_time().
Fixes#1612
Instead of re-styling the border and radius of the linked buttons
depending on the position (middle, first, last, or only child), just
remove the border and radius in a specific direction when the button is
`:not(:first-child)` and/or `:not(:last-child)`.
This properly handles the style of linked buttons in all positions
-- middle, first, last, and only child.
Fixes#1294
This will be used to let the inspector and other users
pick insensitive widgets again. For now, update all
callers to pass no flags, preserving the current
behavior.
These were broken by the crossing event unification.
We are now generating some more crossing events, and
the treeview was not looking closely enough at the
ones it gets.
Closes: https://gitlab.gnome.org/GNOME/gtk/issues/1814
When a root is set on a widget, the style context may
already exist. We need to make sure that the style context
has the right display set.
This was showing up as "css spillover" in the inspector.
Closes: #https://gitlab.gnome.org/GNOME/gtk/issues/1817
Keep only the software gl setting for GL, and put it together
with the simulate touchscreen setting in a 'misc' box. This
keeps all the 'show' options nicely grouped.
We don't want to render focus rectangles on everything. With
the way focus is propagated nowadays, the theme has to selectively
render focus on certain widgets.
At the same time, we always want to render focus for this
theme, so use the focus pseudoclass, not focus(visible).
Closes: https://gitlab.gnome.org/GNOME/gtk/issues/1815
After we stopped untranslatable delivering events, menus stopped
working 'the second time'. After some painful debugging, it appears
that this is caused by the menushell code deliberately grabbing
on the menubar *after* grabbing on the menu, causing events to
be deleivered to the wrong toplevel. This did not use to matter,
but now we drop these events.
Absent a more thorough rewrite of menus, just don't grab on
the menubar. This makes menu activation work again and does not
appear to have ill effects, on either Wayland or X.
Closes: https://gitlab.gnome.org/GNOME/gtk/issues/1796
GitLab's CI will bail out at the first failure, which means the
JSON-to-JUnit conversion script won't run unless it's part of the same
script that we run for building an testing.
The `buttons` test for CSS nodes sets the second RadioButton as the
active one, whereas the first RadioButton is not set as active.
Nevertheless, the reference output says that the first radio button
should match the `:checked` selector, whereas the second radio button
should not.
The fact that the test currently passes is a mystery.
Getting the layout manager instance out of GtkGrid, and then querying
all layout properties can be tedious, especially for code that was
usually calling gtk_container_child_get().
To replace that, we can add a simple query function that returns the two
attach points and the spans.
Additional code improvements and fixes:
- Use g_regex_match_simple() instead of sscanf()
- Added spaces between function names and left parantheses
- Set always correct custom page size
- Added page_setup field to CupsOptionsData data structure
- Replaced tab indentions by spaces
- Moved #define out of add_cups_options() function, removed line breaks from regular expressions
We may avoid setting it on the paned widget depending on the pointer
position altogether, since the handle is now a widget. Also is more
likely to be correct as the implicitly grabbed widget will probably
be that one.
Fixes the paned losing the resize cursor after button press.
Instead of handle_event(), use set_key_capture_widget().
Also, use the fact that we now propagate key events throughout the
stage, and connect to the GtkInspectorPropList instead of the toplevel.
We only want the search bar to be active while that widget is mapped
and able to receive events.
And also, handle the "search bar" visibility on ::search-started.
Users of search entries usually handled visibility (when initially hidden)
by checking the return value of gtk_search_entry_handle_event(). This does
not pan out with gtk_search_entry_set_key_capture_widget() since the return
value is not directly seen by the caller.
Add a ::search-started signal to cater for it, which gets emitted when the
search entry went from empty to non-empty.
The hand-rolled stack combo we were using before
was looking at the visibility of the page itself
to show or hide items. Other stack switchers
expect us to use the GtkStackPage::visible
property for this.
This demo has everything:
- a GtkFixed inside another GtkFixed
- a cube made out of GtkFrame widgets
- an example of 3D transformations
And what's there, in the window once I launch it? The GTK logo made of
widgets.
This is mostly convenience API around GtkFixedLayoutChild, but it should
push people towards using transformations with GtkFixed instead of just
using fixed positioning.
The need of a specialised fixed layout container that can be placed into
a GtkScrolledWindow ceased to exist once GtkScrolledWindow gained the
ability to automatically interpose a GtkViewport when adding a child
that does not implement GtkScrollable.
All the other justifications that led to the existence of GtkLayout as a
separate widget from GtkFixed have been largely made irrelevant in the
20 years since its inception.
We are building against the 3.32 sdk, which has
a new-enough glib to default to the keyfile
settings backend.
Also, use fallback-x11, so we prefer Wayland.
We preiously did not apply the resizes and moves as they were previously
only done in the Cairo drawing context on Win32. Fix this by applying
this too in the GL drawing context.
Make gdk_win32_surface_get_queued_window_rect() and
gdk_win32_surface_apply_queued_move_resize() not static functions, as we
want to use them in gdkglcontext-win32.c, to fix resizing and moving.
As in commit d45996c, the x and y coordinates passed into begin_drag and
begin_move are no longer root coordinates but are now surface
coordinates.
Use the x and y surface coordinates to acquire the root x and y
coordinates so that resizing and moving can work as expected.
As we are building the gtkreftestprivate and reftest test libraries as
DLLs, we need to export the symbols in there so that things will link.
Decorate the symbols with G_MODULE_EXPORT for this purpose.
This makes the search entry show up again
when I type. There is still some misbehavior
where the entry loses focus again, and Escape
does not work to exit search.
We don't want the new transform while drawing things on a texture.
Instead, only apply the new transform matrix when adding the final
texture drawing ops.
This fixes the stack cube rotation transition to at least look somewhat
better.
I've come to the conclusion that we should keep
this state, since not all backends support per-edge
information. Updated the docs to explain how the
tiled state relates to the per-edge states.
The `install` argument for configure_file() was introduced in Meson
0.50, and was ignored in earlier versions.
Since we're still using Meson 0.48 as a baseline, and since it doesn't
cost us nothing to use a conditional in the only place where we used the
`install` argument, let's drop it. This avoids a warning in newer
releases of Meson.
This is nice when you want to make a "screenshot" by using save-as.
Its not going to perform as well though, so you have to enable it
by adding ?datauri to the url
This is not ideal because we report the time of a full roundtrip, rather
than the presentation time, but its better than nothing, and i'm not sure
how the browser time should be reconciled.
Otherwise the module and gtk-reftest will each have their own copy and
that'll mean all symbols - and inhibiting the shutdown - will exist
twice. Not good.
Run gtk-builder-tool --3to4 over it. As the test cares about the
"initial" and "inherit" CSS keywords, the actual widgets aren't that
important. It's just important to have many of them.
The old reftests drew an opaque image, the new image is transparent.
This test drew the reference image as black and the test as transparent
black, and those are now different.
Instead of waiting for the first invalidate-contents signal, wait until
we get a render node. This will break spectacularly for reftests not
drawing anything at all, but we just hope that won't happen.
This way, we don't get an abort once the first test gets a warning.
We also can use meson test to run individual tests.
Unfortunately, only ~60% of tests pass.
If we want to inspect the type of layout properties exposed by a
GtkLayoutManager, we need a way to connect the layout manager type to
the GtkLayoutChild type it creates. In order to do so, we can set the
GtkLayoutChild type on a field of the GtkLayoutManagerClass structure.
Storing the GtkLayoutChild type on the class structure of the layout
manager also allows us to implement a default create_layout_child()
virtual function.
The appstream-util check performed by Flatpak on recent GNOME SDKs has
become more stringent, and now it requires a <release> tag in the
AppData XML file. If we don't have it, the Flatpak bundles of gtk-demo
and gtk-widget-factory will fail on our CI infrastructure.
This is a very old X session management thing, and you
will be hard-pressed to find a session manager that can
make use of it, and even harder-pressed to find apps
using it to their advantage.
...and the setter/getter for it.
This is a very old X session management thing, and you
will be hard-pressed to find a session manager that can
make use of it, and even harder-pressed to find apps
using it to their advantage.
Add a propagate-text-width property, which, when set,
makes the entry request a natural width that is just
enough to fit the content, within the limits given
by width-chars and max-width-chars.
Closes: https://gitlab.gnome.org/GNOME/gtk/issues/1721
Use the same sizing approach we use for GtkEntry:
ignore icons when measuring. This ensures that
search entries don't change size as icons come
and go.
Use the same sizing approach we use for GtkEntry:
ignore icons when measuring. This ensures that
password entries don't change size as icons come
and go.
Change the all the begin_drag and begin_move apis in
GdkSurface and GtkWindow to expect surface coordinates.
Update the x11 implementation to translate to root
coordinates where it matters. Wayland is ignoring the
coordinates anyway.
GtkLayoutChild instances are created on demand once we have a widget, a
GtkLayoutManager, and a child widget. This makes testing their creation
fairly tricky.
Let's skip them, for the time being.
The GtkWidget::parent-set signal was removed in ff6cd8f7.
Instead of removing GtkLayoutChild instances associated to a widget
using notifications when the widget's parent changes, we can have
gtk_widget_unparent() call a method on GtkLayoutManager to remove any
eventual GtkLayoutChild instances associated to the widget.
When sending render nodes from the client to the daemon we add an id,
and whenever we're about to re-send the entire tree node we instead
send the old id. We track all the nodes for the previous frame
of the surface this way.
Having the id on the daemon side will allow us do to much better deltas.
We want to delay some rendering, and to make that safe we need to correctly
refcount the use of blob uris for the textures so that we don't unref
it while something is scheduled to use it.
- Rename GtkLegacyLayout to GtkCustomLayout
- Use for() to iterate over children in GtkBinLayout
- Whitespace fixes for code imported from GtkBox
- Store the GtkLayoutChild instances inside LayoutManager
- Simplify the GtkLayoutManager API by dropping unnecessary arguments
- Fix the ownership model of GtkLayoutManager
If a widget has a LayoutManager instance, then we want to parse layout
properties in UI description files; the grammar is similar to packing
properties in GtkContainer:
<child>
<object ...>
<property name="...">...</property>
<layout>
<property name="pname">value</property>
</layout>
</object>
</child>
The properties are applied after a child has been added to its parent,
to the parent's layout manager property should be set.
GtkLegacyLayout is a layout manager for the transitional period between
the introduction of layout managers and the removal of GtkWidget virtual
functions for the size negotiation.
Layout managers needs a way to store properties that control the layout
policy of a widget; typically, we used to store these in GtkContainer's
child properties, but since GtkLayoutManager is decoupled from the
actual container widget, we need a separate storage. Additionally, child
properties have their own downsides, like requiring a separate, global
GParamSpecPool storage, and additional lookup API.
GtkLayoutChild is a simple GObject class, which means you can introspect
and document it as you would any other type.
We can use a constructor property for existing container widgets with
a layout policy, and move the layout policy implementation out of the
widget itself and into a LayoutManager subclass.
We delegate the size request mode, the measuring, and the allocation of
a widget through a GtkLayoutManager instance, if one has been attached
to the widget; otherwise, we fall back to the widget's own implementation.
A base abstract class for layout manager delegate objects.
Layout managers are associated to a single widget, like event
controllers, and are responsible for measuring and allocating the
children of the widget they are bound to.
Otherwise we're getting warnings about allocating a widget we haven't
measured first, which is fair. The contents gizmo itself will later take
care about whether or not the real popover child is NULL.
We were not paying enough attention to detail when updating
hover and focus state while generating crossing events. The
invariant that we need to preserve here is that when a widget
has focus or hover, its parent does too.
We basically don't have child surfaces anymore (the last
use in popovers is on the way out). This really needs
to be done in terms of widgets, not surfaces. For now,
just stop walking parent surfaces.
ImmIsIME() doesn't work (always returns TRUE) since Vista.
Use ITfActiveLanguageProfileNotifySink to detect TSF changes,
which are equal to IME changes for us.
Also make sure that IMMultiContext re-loads the IM when keyboard layout
changes, otherwise there's a subtle bug that could happen:
* Run GTK application with non-IME layout (US, for example)
* Focus on an editable widget (GtkEntry, for example)
* IM Context is initialized to use the simple IM
* Switch to an IME layout (such as Korean)
* Start typing
* Since IME module is not loaded yet, keypresses are handled
by a default MS IME handler
* Once IME commits a character, GDK will get a WM_KEYDOWN,
which will trigger a GdkKeyEvent, which will be handled by
an event filter in IM Context, which will finally re-evaluate
its status and load IME, and only after that GTK will get
to handle IME by itself - but by that point input would
already be broken.
To avoid this we can emit a dummy event (with Void keyval),
which will cause IM Context to load the appropriate module
immediately.
When the window gets active / inactive, we
don't propagate events, but just send focus-in / -out
to the current focus_widget. Improve this by updating
its state flags as well.
We were walking the parent chain here, which now
always needs to consider whether it should stop
at roots. Like this one should.
The symptom was that a label with a popup attached to
it would end up with an unintentional focus ring that
would not go away.
Refactor the child allocation machinery, so that the complex allocation
paths are only run when the animation is running.
And in particular, ensure that when no animation is running, the
identity transform is allocated.
When dealing with subclasses of GtkEntry, we were not
getting the property offset that is stored on the GtkEntry
type.
This was showing up as criticals when trying to set
::width-chars on a GtkFileChooserEntry.
This is named gdkconstructor.h to avoid any possible conflicts. This fixes
the current usages of G_HAS_CONSTRUCTORS, as that header is not installed
by glib.
Add boolean properties, is-pointer-focus and
contains-pointer-focus, that track whether the pointer
is in the widget itself or in one of its descendants.
The ::enter and ::leave signals get emitted up and down the
connecting path between the old an the new pointer location.
The signals are less useful if you can't find out where along
the path you are. That is what crossing mode and detail are
about, so add those to the signals.
It does not exist anymore. I'm removing this code now
because our CI tests are using xim and fail due to this.
Eventually, this code should be ported to use a popover.
Check that we get the expected sequences of focus
change events for the nonlinear, inferior and ancestor
cases.
It would be nice to do the same checks for crossing
events, but we have no gtk_window_set_hover().
Emit focus change events in the same way as crossing events.
Also change the code to only emit focus change events for
the master keyboard - we only maintain a single focus location,
so sending multiple focus change events for different devices
seems confusing.
GtkWindow has a focus_widget that points to the current input focus.
GtkWidget has a focus_child that points to the child that contains
the input focus. Following the focus_child chain from the toplevel
always leads to the focus_widget. We never unset focus_child, we only
set it. We bubble focus change events.
Make the function that determines initial visibility
look at whether the class implements GtkRoot. That is
the eventual goal for this check. For now, allow
popovers in here as well.
It can be a bit confusing to have an indicator
and an action next to each other, and with the
peek icon, the need for the Caps Lock warning is
reduced, since you can just reveal the text to
see that it is capitalized.
Therefore, only show the Caps Lock warning if
the peek icon is disabled.
Add a ::show-peek-icon property and show a clickable
icon when it is set. Clicking it toggles the visibility
of the content. The same functionality is also accessible
via a context menu item.
This is a common feature of password entries.
We don't want it to appear clickable, but we still
need to keep it pickable for the tooltip to work,
so explicitly give it the same cursor that we use
for the text.
gsk/gskenums.h:181: Error: Gsk: multiple "@GSK_TRANSFORM_CATEGORY_2D" parameters for identifier "GskTransformCategory":
* @GSK_TRANSFORM_CATEGORY_2D: The matrix is a 2D matrix. This is equivalent
^
gsk/gsktransform.c:1342: Warning: Gsk: gsk_transform_to_2d: unknown parameter 'm' in documentation comment, should be 'self'
gsk/gsktransform.c:1368: Warning: Gsk: gsk_transform_to_2d: invalid return annotation
gsk/gsktransform.c:1461: Warning: Gsk: gsk_transform_to_translate: unknown parameter 'm' in documentation comment, should be 'self'
This reinstates diffing in the same way that it worked for offset nodes.
It would be possible to add diffing for affine transforms or even all
transforms, but I think this is unnecessary right now - and also quite
expensive to compute.
Make the API expect a tranform of the proper category instead of
doing the check ourselves and returning TRUE/FALSE.
The benefit is that the mai use case is switch (transform->category)
statements and in those we know the category and don't need to check
TRUE/FALSE.
Using the wrong matrix will now cause a g_warning().
... instead of computing it every time we need it.
This should be faster and we want to use it a lot more prominently.
Also, we have the struct memory available anyway.
In particular, check that to_matrix() and to_2d(), to_affine() and
to_translate() return the same values.
This also requires a recent Graphene version or the tests will fail.
In particular, add a per-category querying API for the matrix:
- gsk_transform_to_translate()
- gsk_transform_to_affine()
- gsk_transform_to_2d()
- gsk_transform_to_matrix()
This way, code can use the relevant one for the given category.
In case the theme doesn't set a height/min-height for the treeview
separator the treeview drawing gets confused and draws rows on top of each
other depending on the redraw area.
This is due to gtk_tree_view_get_row_height() assuming that a node with a
height <= 0 is not set and not a separator and it will default to the
expander size.
Ideally gtk_tree_view_get_row_height() would know if it operates on a separator,
but there are too many calls/levels, so just make sure the separator height
is at least 1 (Adwaita already sets "min-height: 2px", so no change there)
Cherry-picked from !614 to master
The tests were added when we thought we had to align memory allocations
for structures including a Graphene type in their members. Graphene
added alignment annotations for its types, and we never really used the
symbols we set after testing for allocations being aligned out of the
box with malloc(), and for aligned allocators.
Picking is done by drawing a line along the parent's z axis and picking
at the intersection with the child's z=0 plane.
However, the previous code was casting a ray along the child's z axis.
This patch actually transforms the line to pick into the target's
coordinate system and then computes the corrrect intersection with the
z=0 plane.
Using graphene_point3d_interpolate() to compute the final intersection
point is a bit of abuse of that function, but I found no better way in
Graphene to achieve the same thing.
Since we can do partial redraws, dropping every shadow that's been
unused for one frame happens too fast. This is also a problem when a
shadow gets drawn on a texture for a few frames.
This can happen for certain transform nodes. The transform node's
child's bounds are fine, but the transform node bounds are all nan.
Just ignore those bounds since we can't meaningfully render them anyway.
If the given matrix is explicitly of category IDENTITY, we don't need to
do anything, and in the 2D_TRANSLATE case, just offset the child bounds.
Those are the two most common cases.
While *some* systems alias python to python3 nowdays, this is
not true for eveything. Especially systems that can potentially
offer both python2 and python3.
According to both PEP 394 and PEP 441 its recommended to always
add the 3 in the shebang.
After considerable discussion, we came to the conclusion
that the convenience of this API wins over the correctness
of gtk_window_present_with_time(), in particular since we
don't have a good mechanism to carry timestamps from the
events to the places where we present windows.
When 0 or GDK_CURRENT_TIME is passed to gtk_window_present_with_time(),
print a warning so that the application developer knows that this isn't
a supported use of the function, but carry on working for now.
Instead of using a grab on a GtkInvisible, use
a hook in the GTK event propagation machinery to
get events.
The only downside of this approach is that we
lose the crosshair cursor. But we get rid of
the last use of GtkInvisible.
Change gdk_surface_get/set_user_data to private
API and rename them to get/set_widget.
Also remove an unused associated function.
The last two places where the surface API is used
are in gtkroot.c and gtkwidget.c. Make them
use the private api.
This was broken by the change in 01f7f255b5 which
caused the inspector to not get any events anymore.
Revert that part, even though it may be technically
correct.
This is a first cut at updating the drawing model chapter
for the way we do things now. It introduces the scene graph and
render nodes, explains node caching and tree diffing, and removes
sections about subwindows.
This avoids invalidating the size of all widgets when updating CSS
transforms.
In theory, we don't even have to allocate the widget itself, because we
didn't change its size. But we have no way to track that.
Instead of gtk_snapshot_offset(), provide a full set of functions
kept in sync with GtkTransform APIs.
On top of that, add gtk_snapshot_save() and gtk_snapshot_restore()
mirroring cairo_save()/restore() that allow saving a snapshot's
transform state.
The code didn't change, it was just shuffled around to make the
with_bounds() versions of the text rendering unnecessary and instead
pass through the generic append_node() path.
Instead of just tracking 2 integer translate_x/y coordinates, tracka a
full GtkTransform.
When creating actual nodes, if the transform is simple enough, just
create the node in a way that makes use of the transform. If the
node, can't represent the transform, just push a transform node instead
and automatically pop that node with the next gtk_snapshot_pop() call.
They were a neat idea while they lasted. But now, it's time for
categorized transform nodes, where matrices with
GSK_MATRIX_CATEGORY_2D_TRANSLATE are the exact replacement.
Renderers have not been adapted for this purpose, so they (continue to)
run slow paths.
This is a new object (well, boxed type, but I'm calling it object) for
dealing with transform in a more constructive way than graphene_matrix_t
by keeping track of how the transform was created.
This way, reasoning about the transform becomes easier, and we can create
better ways to print it or transition from one transform to another one.
An example of this is that while a 0 degree and a 360degree rotation are
both the identity matrix, doing a transition between the two would cause
a rotation.
Don't leave memory in an unitinialized case when returning FALSE from
gtk_widget_compute_transform().
We both know that people are going to call that function without
checking the return value.
Use a GtkText child, and delegate the editable functionality
to it. Also forward all the properties that are provided by
GtkText.
Some of the more internal APIs, such as layout and im context
access and caps-lock warning, are removed here, but we preserve
most of the plain GtkEntry API by forwarding it to the GtkText
child.
With transforms in the mix, checking if the coordinate is inside the
widget "allocation" makes even less sense. Just use gtk_widget_pick()
and walk up until we find a GtkFlowBoxChild.
This is the same as the old code since the transformation only contains
teh offset right now, but it will be different later where arbitrary
transformations are possible per widget.
The transform matrix is a translation matrix from the parent's origin to
the widget origin. We will later allow more transformations than just
translations.
Instead of style + rect_of_one_box, pass the new GtkCssBoxes object.
This has the nice side effect that when drawing background + border +
outline, we only compute all the boxes we need once.
Previously, those numbers stored the values relative to the margin box
of the widget. Now they store values relative to the content box,
thereby getting rid of the last remains of weird coordinate systems.
Split out the code for computing CSS boxes from given variables from the
background render code. This way, it can be shared between different
codebases.
Also, make that code completely be contained of static inline functions.
That ensures that it can be 100% inlined in cases where only parts of
the rectangle are needed (like in gtk_widget_get_width() in the future).
This will require some more patches to actually work, but those will
follow.
This way, we can compare with literally the previous allocation and the
size will not be influenced by an adjusted allocation.
But more importantly, we can now use the transform/width/height values
for other stuff.
It's not priv->transform (to be turned into a graphene matrix),
priv->width and priv->height.
The numbers are still the same.
The only difference is that unallocated widgets will now have x/y set to
0, not to -1.
Make items-changed never emit 2 signals, instead, always emit only one,
potentially by extending the range reported in items-changed.
And be a lot more exhaustive about autoselect tests.
1. Do not make position an inout variable
The function is meant to return a range for a given position, not modify
a position. So it makes no conceptual sense to use an inout variable.
2. Pass the selected state as an out variable
Using a boolean return value - in particular in an interface full of
boolean return values - makes the return value intuitively feel like a
success/failure return. Using an out variable clarifies the usage.
3. Allow passing every position value
Define what happens when position >= list.n_items
4. Clarify the docs about how this function should behave
In particular, mention the case from point (3)
5. Add more tests
Again, (3) needs testing.
When the user approaches a tablet tool to the screen we get a proximity-in event
and in this moment we need to check the surface output scale to find the scaling
to be applied to the cursor.
And the same should be done when the tool is detached or the monitors
configuration changes.
https://gitlab.gnome.org/GNOME/gtk/issues/1675
We need to remove the weak pointer, as the stack switcher can
keep the list model alive beyond the stack. This was observed
to cause crashes:
==16870== Invalid write of size 8
==16870== at 0x5168A4E: g_nullify_pointer (gutils.c:2284)
==16870== by 0x522C500: weak_refs_notify (gobject.c:2791)
==16870== by 0x50FE7BC: g_data_set_internal (gdataset.c:407)
==16870== by 0x50FECA7: g_datalist_id_set_data_full (gdataset.c:670)
==16870== by 0x5227EB4: g_object_real_dispose (gobject.c:1056)
==16870== by 0x522D295: g_object_unref (gobject.c:3309)
==16870== by 0x4AF849F: unset_stack (gtkstackswitcher.c:428)
==16870== by 0x4AF892E: gtk_stack_switcher_dispose (gtkstackswitcher.c:527)
Ironically, these properties are too good - they always
give you a proper value, which is unfortunately different
from the declared default value, which is NULL. So, don't
check these.
It is easy to emit wrong ::selection-changed signals,
and then bad things will usually happen later. Add
some sanity checks to gtk_selection_model_selection_changed
to make this easier to track down.
Make GtkStackSidebar and GtkStack communicate via
the selection model that GtkStack now exposes.
This is parallel to the GtkStackSwitcher changes
in the previous commit.
The first set of glyphs is created with a timestamp of 1. Later we
subtract the glyph timestamp from the cache timestamp, meaning we end up
with numbers ending in 9, e.g. 59. Now unfortunately !(60 <= 59), so we
do not end up incrasing the old_pixels count of the cache. Later we then
call lookup() and DEcrease the old_pixels count, which makes the
unsigned int wrap and cause a huge old_pixels value, which causes us to
drop the cache.
If the recoloring would end up multiplying the alpha component with 0
anyway, just skip drawing anything altogether.
This increases the icon count in the switch demo of the fishbowl from
~260 to ~280 on my system.
Specifically it is avoided to be toggled if:
- Just received focus (in order to preserve OSK state across focus changes)
- Moving cursor around. Still allow some jitter as perfect accuracy is not
possible.
Closes: https://gitlab.gnome.org/GNOME/gtk/issues/1277
We were looking at the wrong class names here, we need
to look at the owner type to match against our list.
This fixes problems where gtk-builder-tool simplify
inadvertedly loses hexpand or vexpand settings, messing
up layout, as recently happend in gtk4-widget-factory.
Instead of adding them and waiting for the changed signal to be emitted
in the main loop, there might be a race where the change signal is
emitted before we have a chance of spinning the loop.
Rewrite the builder-tool simplify command to have
a full parse tree around, and perform simplifications
on that tree. This lets us rewrite GtkStack and turn
child properties into child meta objects.
In addition to <property name="foo">bar</property> referring
to an object with ID bar, we now also parse
<property name="foo"><object>...
to specify a property 'inline'.
The change to make widgets visible by default broke GtkInvisibles
special-cased state handling and that in turn caused picking in
the inspector to break with another recent change.
This change makes the inspector pick button work again.
This is necessary to give back focus to the Broadway elements when
content is embedded in an IFrame.
Signed-off-by: Mickael Istria <mistria@redhat.com>
Since we position the tooltip window relative to the toplevel widget and
not actually relative to the effective_toplevel, we shouldn't get the
pointer position relative to the effective_toplevel.
We previously used the pointer position (relative to the
effective_toplevel) and the anchor rect (relative to the toplevel
widget) together to calculate x_distance. This leads to wrong values in
cases where get_surface (new_tooltip_widget) != get_surface (toplevel)
Fixes#1427 in master
The event_widget is the widget that the surface belongs to which got
this event. The target widget is the one that will receive the event.
The previous terminology was confusing.
The @filename@ directive will use the full path of the file being parsed
for enumeration types; we should use @basename@, instead, as it improves
the reproducibility of the build by using only the file name.
Commit bd71e744d2 removed
gtk_box_pack_end(), but it added a gtk_container_add() with an
uninitialised widget, and the compiler is very unhappy about it.
Without this, disabling a widget that's being hovered and is a child
widget of the widget we're disabling (e.g. the GtkImage child of a
GtkButton) will retain its :hover state even though it should be
insensitive to any sort of input now.
Change the reorder api to insert after a sibling,
so that moving to first place becomes reorder (... NULL).
And add a insert_after api that can replace the common
container_add / reorder_after (... NULL) combination.
Update all callers.
The position child property is problematic, since it
requires us to emit notification for all children when
inserting a child early in the list of children.
Remove the property from all ui files.
GtkWidget saves a widget list for us, so we don't need to keep track of
them ourselves. This is okay now that we don't have a pack-type child
property anymore.
Some of the flags got lost in the meson transition or were demoted from
error flags to warning flags.
This commit reintroduces them.
It also includes fixes for the code that had warnings with those flags.
The big one being -Wshadow.
Remove the unneeded is_platform() check and just go by extension point
priority.
Also g_error() out if no im module exists, because "simple" is compiled
in and should always exist.
All built-in backend modules get a priority of 0 because they are the
default ones.
GtkIMContextSimple gets a priority of G_MININT because it's the fallback
one.
This mirrors the media modules code.
The format of the printout will be suitable for addition as a new test to
testsuite/gtk/rbtree-crash.c
by just grepping the printouts from the relevant rbtree.
We need to tell the portal what filter is supposed to be selected by
default, or it will just pick the first one, which could be wrong and
annoying.
This will require updated xdg-desktop-portal and xdg-desktop-portal-gtk
to work properly.
Fixes#1492
gtk_file_chooser_set_filter() doesn't work for GtkFileChooserNative. The
code forwards added and removed filters to the delegate dialog, but
doesn't do anything to set the selected one, so the wrong one gets
chosen. So fix that.
This only fixes the fallback dialog. The portal will be fixed in a
subsequent commit.
Partial fix for #1492
Instead of using the INCLUDE directive inside the sections file, we can
specify the default C include in the gtkdoc-mkdb arguments, and override
it inside the C sources that need it.
This gives us a better way of choosing the color of the placeholder text
(and enabled general css styling on it of course).
Closes#378 (If you want to keep the placeholder on focused and empty
entries, just don't set the placeholder opacity to 0 in
entry:focus>placeholder. This is the default behavior but this commit
includes a rule in Adwaita to hide it.
Since we now have a widget whenever we query tooltips, we can as well
get the events target_widget if we have an event (which is what we do
when coming from gtkmain.c). This keeps us from searching the entire
widget hierarchy for the target event even though we've already done
that for pointing events in gtkmain.c
This reduced the work done in gtk_tooltip_handle_event in normal motion
events to basically nothing since we already did all the heavy lifting
when handling the pointing event in gtkmain.c
As stated by the documentation, this should be called when a widget gets
updated, but in that case, one can equally use
gtk_widget_trigger_tooltip_query.
Setting it as qdata on the object doesn't save any memory since we use
the user_data as the event target, which every event has set these days.
This way is also faster since just reffing the object doesn't do any
locking.
If the text style changes, or the display settings do, we need to update
the state labels to ensure that the glyphs are available in the font
we're using.
The entire color scale hack is still done in GtkRange, which draws the
color scale in the range gizmo. So, to correctly redraw the color scale
when setting a new color, we need to redraw the proper widget and that's
the trough widget.
Fixes#1453
Instead of recording the way up from the target widget to the grab
widget (or toplevel) and then walking that path upwards, just walk the
parent chain and look at the cursor.
Most of the time, the GtkSnapshot objects we create while snapshotting
widgets don't end up containing all that many nodes or states in their
respective node or state stack. This undermines the amortized allocation
behavior of the G(Ptr)Array we use for the stacks. So instead, use the
(until now unused) parent_snapshot GtkSnapshot* passed to
gtk_widget_create_render_node and reuse its node and state stack.
We do not avoid allocating a new GtkSnapshot object, but we do avoid
allocating a ton of G(Ptr)Array objects and we also avoid realloc'ing
their storage.
Even though the IEC power glyphs are part of Unicode 9.0 (released in
2016) not all fonts have them.
To avoid showing the hexbox of doom when the system font does not have
the glyphs we'd like to use, add a fallback pair, using the old glyphs
we suggested when the labels were translatable.
The refactoring of automatically updating tree->root when setting a
node's parent works very well - unless all nodes get removed and no
node's parent got updated.
The tree is not needed to walk around the nodes.
It is however still needed for anything that requires modifying the
tree.
There is no immediate benefit in changing this API, but there might be
situations in the future where we can avoid looking up the tree when we
just want to check some details about the node.
Store a link to the tree in the root node. This allows looking up the
tree in O(log N) from the node without any extra memory usage.
This is useful because code can just store a pointer to the node and
doesn't need to keep the tree pointer around. And that can (for large
trees) save quite a bit of memory.
Searching through the tree is too specific to use a general function.
All the existing code just copies and slightly adapts the same 20 lines
instead, so there's no reason to keep the complicated API.
This broke the overlay blur demoe when resizing the window to a size
that would completely move the image below a button, causing the
GtkSnapshot code to remove the clip node below the blur node.
Considering the operations that some of the rendernode constructors
do, nodes with width or height 0 (or both of course) are very well
possible. This would break in the rendernodepaintable when adding a
transform, which divides by width/height of the rendernode.
We'd like the rose picture to be bigger than 16×16. Also remove the
scrolledwindow since the GtkPicture now automatically scales down the
rose image. This also fixes the picture always being allocated at y=0.
This can happen whenever the ::activate-link handler sets different
markup on the label, causing all links to be recreated. In this case,
the GtkLabelLink* passed to emit_activate_link is garbage after the
g_signal_emit call and we shouldn't try to do anything with it.
Fixes#1498
Unicode 9.0 introduced glyps for the "on" and "off" power states, in the
form of:
- U+23FD POWER ON SYMBOL, or ⏽
- U+2B58 HEAVY CIRCLE, or ⭘
With `HEAVY CIRCLE` as "power off symbol" selected to avoid adding yet
another circle to the standard.
Since we moved GtkSwitch to always show glyphs instead of (translatable)
strings, asking the localisation teams to either come up with a suitable
short string to replace the English "ON" and "OFF", or to fall back to
Unicode glyphs, we should ensure we're using the appropriate symbols to
begin with.
See also: gtk!503 for the corresponding gtk-3-24 change.
Signal emittion was added in 6f857f87dc commit and it seems that
this is only place where selected_row is set after emitting signal.
Because of this gtk_list_box_get_selected_row currently returns NULL
as selected row if selection mode is set to GTK_SELECTION_BROWSE.
Using an empty `configuration_data` object to copy a configuration file
is deprecated since Meson 0.47 (released July 2018); the equivalent
behaviour is available by using `copy: true`.
The target position is irrelevant for determining if the child should be
visible. When the current position is 0, it needs to be hidden, period.
Fixes#1355
The previous fixes made it unnecessary to hardcode IM modules for
different display types. The code now automatically skips system IM
modules for other displays.
The code would technically allow loading the xim module when X11 support
was not compiled in.
This is probably an artificial concern, because it's pretty hard to
compile XIM support without X11 support, but it makes the code clearer,
so there we go.
Context IDs are dependant on the display - both because displays can use
different backends, but also because changing the GtkSetting is a
per-display operation.
So just remove the cache.
If it turns out we need a per-display cache, we can add one to
GtkSettings.
Calling the accessibility function `grab_focus()` on a `GtkCell` under
Wayland will cause the client to crash.
This is another case of `gdk_x11_get_server_time()` being called
regardless of the actual windowing backend used.
Closes: https://gitlab.gnome.org/GNOME/gtk/issues/1507
So it's able to operate properly with the DnD gesture set by
gtk_drag_source_set(). We usually just react on button release,
that's the right time to claim the gesture.
Closes: https://gitlab.gnome.org/GNOME/gtk/issues/1557
I was stuck in an X session and noticed that my resize corners
all got east or north cursors. It turns out that gnome-shell
does not properly advertise support for edge constraints under X11,
and the absence of that makes the code for determining the edge
under the cursor misbehave.
This change should fix that.
So we can check that the currently set clip is the first one and now
intersect with it. This first clip is always the entire viewport or the
entire render_area and we don't want to end up drawing things to a
texture because of it.
Link to the GitLab documentation, and clarify that if no single commit
in a merge requests closes an issue, you should add a reference to the
issue in the commit message anyway.
This is an important document for newcomers, so we should err on the
side of being more detailed on what kind of contributions we expect,
and how we expect them.
The text is heavily modelled on the contributing-template by Nadia
Eghbal available here:
https://github.com/nayafia/contributing-template
g-ir-scanner incorrectly evaluates macro definition that include
references to other macro definitions. Provide a correct value as an
annotation.
Differences in generated gir files:
```diff
@@ -19017 +19017 @@
- <constant name="PRIORITY_REDRAW" value="20" c:type="GDK_PRIORITY_REDRAW">
+ <constant name="PRIORITY_REDRAW" value="120" c:type="GDK_PRIORITY_REDRAW">
@@ -74229,3 +74229,3 @@
</constant>
- <constant name="PRIORITY_RESIZE" value="10" c:type="GTK_PRIORITY_RESIZE">
+ <constant name="PRIORITY_RESIZE" value="110" c:type="GTK_PRIORITY_RESIZE">
<doc xml:space="preserve">Use this priority for functionality related to size allocation.
@@ -106786,3 +106786,3 @@
<constant name="TEXT_VIEW_PRIORITY_VALIDATE"
- value="5"
+ value="125"
c:type="GTK_TEXT_VIEW_PRIORITY_VALIDATE">
```
See !472
If the revealer is told do animate and then unrealize itself, we do
(correctly) stop the animation, but used to do a shortcut where we
just set the target state as current.
Other things are dependent on the animation properly finishing though,
like the contained widget child visibility. This may lead to inconsistent
state where gtk_revealer_get_child_revealed() returns TRUE but the child
widget is unmapped, or vice-versa.
Fully finish the animation here, so the child state is coherent the next
time the revealer is mapped. We can also skip notifying on the property
since it will be handled by gtk_revealer_set_position().
Tools on the same physical item have the same serial number, so the eraser
and the pen part of a single pen share that serial number. With the current
lookup code, we'll always return whichever tool comes first into proximity.
Change the code to use the hw id in addition to the serial number, this way we
can differ between two tools.
Generic tools (Bamboo, built-in tablets) always have the same serial number
assigned by the wacom driver. This includes the touch tool when the wacom
driver handles the touch evdev node (common where users require the wacom
gestures to work).
When the first device is the touch device, a tool is created with that serial.
All future tools now return the touch tool on lookup since they all share the
same serial number. Worse, this happens *across* devices, so the pen
event node gets assigned the touch tool because they all have the same serial.
Since we don't actually care about the touch as a tool, let's skip any unknown
tool. This captures pads as well.
Any wacom device currently sets the tool type to UNKNOWN. The wacom driver has
a property that exports the tool type as one of stylus, eraser, cursor, pad or
touch. Only three of those are useful here but that's better than having all
of them as unknown.
Before this patch, imwayland would assume that text-input enter and leave events follow the general (wl_keyboard) focus, and was unable to handle the situation where they would not be provided at the same time.
Fixes terminal emulator misbehaviour as outlined in https://gitlab.gnome.org/GNOME/gtk/issues/1316, which was introduced in 49b17e6c. The original commit cleared preedit text by setting it to an empty string, which still counted as existing preedit. The fix sets preedit string to null, which is correctly understood as not present.
There may be situations where this might get called while the
currently focused context just went away (eg. after setting the
text widget unsensitive).
Closes: #1317
* We don't output spaces anywhere in the code, unlike the doc suggested.
* CSS explicitly forbids whitespace between function names and lparens:
https://stackoverflow.com/questions/13877198
Do not call _gtk_widget_captured_event(), in propagate_event_down(), or
gtk_widget_event(), in propagate_event_up(), when the widget has been
unrealized.
iter_init_common() is used on uninitialized GtkTextIter, and since neither it
nor its callers initiliaze its padding fields, they contain garbage.
This is a problem for Go - which checks that structs passed to C functions do
not contain pointers to Go-allocated memory - when the garbage happens to be
such a pointer. Although Go zero-fills all GtkTextIter that it allocates, this
does not help when GTK functions such as insert_pixbuf_or_widget_segment called
for gtk_text_buffer_create_child_anchor copy garbage from their stack-allocated
GtkTextIter into a clean iter. To work around this a GtkTextIter has to be
discraded after use in text buffer anchor inserting functions:
https://github.com/gotk3/gotk3/pull/307
We display a list of supported protocols in the server_addresses_popover.
However, this curated list contains protocols which may or may not be
available, depending on the respective gvfs backend being installed.
So, populate the list only with protocols which are available.
https://gitlab.gnome.org/GNOME/gtk/issues/1476
When the user types an address with a schema that is not supported,
the Connect button doesn't become sensitive, but there is no visible
feedback at all.
This feels unresponsive and leaves the user clueless.
While it doesn't help explain why the address doesn't work, this will
provide a hint that the input was acknowledged but doesn't work.
https://gitlab.gnome.org/GNOME/gtk/issues/1476
This makes apps use "Segoe UI 9" by default instead of whatever matches "Sans 10".
It also cleans up the code and uses some new pango API while at it.
This was previously disabled in 9e686d1fb5 because it led to a poor glyph coverage
on certain versions of Windows which don't default to "Segoe UI 9" (Chinese, Korean, ..)
because the font fallback list was missing in pango.
This is about to get fixed in https://gitlab.gnome.org/GNOME/pango/merge_requests/34
so enable it again when we detect a new enough pango version.
(See !436 for the original MR)
GTK widgets expect the scroll deltas to be 1 or -1 and calculate a scroll value from that.
Multiplying the delta by the Windows scroll line setting (which defaults to 3) results
in a much larger delta and vastly different behaviour for running a GTK app on Windows
vs on Linux. For example text view and tree view scroll by 9 lines per scroll wheel tick
per default this way while on Linux it is around 3.
Remove the multiplication for now.
See !426 for the gtk3 MR
Enables hinting, antialiasing and set the subpixel orientation according to the
active clear type setting. This ensures that font rendering with the fontconfig backend
looks similar to the win32 backend, at least with the default system font.
See !437
The existing post-install shell script will most likely not work on
Visual Studio builds as there is normally no shell interpreter installed
on the system where the build is done, but the build is normally done in
a standard Windows cmd.exe console.
Instead, use a Python script so that it will work on the platforms that
Python supports.
We need to override _GLIB_EXTERN to export the required symbols for the
GIO module on Visual Studio, so that the media modules can be
successfully loaded.
Build the .rc files for Windows so that one can track the version info
more easily for Windows, as well as giving GTK+ apps a default icon.
Also, move back the manifest embedding for the themed Windows print
dialog back into gtk-win32.rc.body.in, so that we just have one way of
embedding this manifest file, making things easier for ourselves, as
this is supported in the later Visual Studio compilers as well, which is
2013 and later.
Issue #1495 showed that the docs of GtkGrid retain outdated implications
that (as was once, but is no longer, the case) it is intended to replace
GtkBox, by discussing HfW and widget properties in a way that suggests
GtkBox can't handle them. But of course it does, and it's preferable for
simple single-row/column cases. Worse, we said GtkGrid “provides exactly
the same functionality” for the latter case, but the original point of
that Issues was that it doesn’t, at least for CSS positional selectors!
Box:
• Use an actually meaningful @Short_description.
• Remove unhelpful @See_also references to unrelated containers.
• Remove references to “rectangular area”: it might be another shape
via CSS, or “rectangular” might falsely imply 2 dimensions of children.
• Mention Orientable:orientation.
• Emphasise usefulness of :[hv]align for allocating in the other axis.
• Don’t say that Grid “provides exactly the same functionality” for a
single row or column, since (A) it is overkill for that case and (B)
said Issue proved that it *doesn’t* for CSS child order, for example.
Grid:
• Don’t dwell on widget properties and height-for-width in a way that
wrongly implies that Box can’t handle those (or Grid can better). In
fact, just get rid of that bit altogether: Box handles them fine, and
such wording was only needed years ago for migration from GTK+ 2 to 3.
• Point to GtkBox as being preferred for the simple row/column use case.
As per the spec:
> The back buffer can
> either be reported as invalid (has an age of 0) or it may be
> reported to contain the contents from n frames prior to the
> current frame.
So a buffer age of 1 means that the buffer was used in the last frame.
We were handling buffer_age==1 the same as buffer_age==0, i.e. we
returned the full damage for the surface.
[1] https://www.khronos.org/registry/EGL/extensions/EXT/EGL_EXT_buffer_age.txt
Append a variation selector to the Emoji sequences,
to force Emoji presentation. Without this, some
Emoji come out with text presentation by default.
Closes: Pango #334
- step back on toning down the borders. Flatness !> legibility.
- darker active state for light
- draw gradinets from bottom up, to keep px sized shading regardless
of button size.
Instead of getting the translation x/y everytime we use the modelview,
get it once, when extracting the metadata. Do the same with the scale.
And save if the matrix is "simple" at all, i.e. if it only consists of a
translation and/or scale. This will be helpful later when we start
drawing transformed nodes on textures.
Increase the visibility of the box-shadow for menus
Introduce a border-radius variable for menus
Use this variable for all corners of menus except top for the top menus
We wrap SVG data from icons within another SVG with extra styling
information. The wrapped SVG may contain characters that cannot be
part of a data: URL (https://fetch.spec.whatwg.org/#data-urls).
Librsvg 2.45 got more strict in its parsing of data: URLs; whereas
previously it ignored '#' characters in them, now it considers them to
be the start of a fragment identifier, which is not allowed in data:
URLs anyway.
To avoid unallowed characters, we now create a data: URL with a
base-64 encoded SVG.
Fixes https://gitlab.gnome.org/GNOME/gtk/issues/1471
This is important when the target widget of an event is not the one that
would otherwise receive the gesture. For example, the GtkSwitch
implementation currently attaches a pan gesture to the switch itself,
but the target widget below the pointer might be the switch slider or
label.
See #1465
It is permissable to remove a widget using gtk_container_remove from the
gtk_container_foreach callback handler. Document this fact to make it
more discoverable.
Fixes#1461
Currently, gtk_event_controller_scroll_handle_event() always returns
TRUE if it is handled, which stops the propagation of the event. If
there’s a single GtkEventControllerScroll in the widget hierarchy, that
means that no others will run, depending on the propagation phase. In
Nautilus, this can be observed when adding a scroll controller to the
GtkScrolledWindow (ctrl-scrolling controls the zoom level) - either the
scrolling or the zooming breaks.
Fixes https://gitlab.gnome.org/GNOME/gtk/issues/45
Gives the same background color to all separators descending from a
title bar than to its direct childrens.
This prevents separators which are in a titlebar but not direct children
from the widget with the titlebar style class from being almost
transparent and hence it prevent them from revealing the clear color of
the window's titlebar (black).
https://gitlab.gnome.org/GNOME/gtk/issues/1231
This is better than nothing at all. The wording is taken from Carlos's
commit message when he added this shortly before 3.12 (but skip Since).
Skip the bit from his commit message explaining what this replaced; we
don't need to say all the less good things our convenience API replaces.
as per efd3758f6a strcasecmp() is not a C
standard thing (not that we bothered including any header for it anyway)
and so this test failed to build on Windows with Microsoft Visual C.
This is the excellent explanation from Emmanuele at
https://gitlab.gnome.org/GNOME/gtk/merge_requests/402#note_361210:
"
Every time you instantiate a type, the instance_init() function is called for each
parent type T_p of your type T; to preserve invariants, the class pointer inside
the instance data is set to the parent type before each invocation, until you hit
your type T. This means that calling GET_CLASS() inside an instance_init() function
will give you a pointer to the class vtable for the parent type T_p while you're
iterating over parent types. What if you want to access the actual class vtable of
the type T, though? Well, you can because the actual signature of instance_init() is:
void (* GInstanceInitFunc) (GTypeInstance *instance, gpointer g_class);
i.e. all instance_init() functions get passed the instance they are initialising
and the class vtable of the real type you're instantiating.
This is how GtkToolButton works: it "peeks ahead" at instance initialisation time,
to use the button_type class field of the actual type you're instantiating,
and calls g_object_new() with it to store the resulting object in its own private
data structure.
This whole contrived mechanism is needed to allow out-of-tree tool buttons to just
set the button type on their class init, and have their parent class create the
button they want, instead of asking all tool buttons to do this themselves and have
a virtual function called get_button() for GtkToolButton to call whenever it needs
to operate on the button instance.
Now we're coming to a close: we cannot use the G_DEFINE_TYPE macro because the
instance_init() function it creates internally will not pass the class pointer
to your custom instance_init(). Since we cannot use G_DEFINE_TYPE, we also cannot use
G_ADD_PRIVATE either.
This is the reason why, when I ported GTK 3 to the new private instance data structure
macros, I left GtkToolButton alone. I should have left a comment there, because @matthiasc
tried doing that as well, and then had to revert it in commit 1c4a7bd5. So: my bad,
sorry about that.
If we want to drop the G_TYPE_INSTANCE_GET_PRIVATE and the g_type_class_add_private() calls,
we cannot use G_DEFINE_TYPE, but what we can do is unrolling what the macros do themselves:
- add a global GtkToolButton_private_offset variable
- add a static inline gtk_tool_button_get_instance_private() that does return
(G_STRUCT_MEMBER_P (self, GtkToolButton_private_offset));
- call g_type_add_instance_private (g_define_type_id, sizeof (GtkToolButtonPrivate)) inside
gtk_tool_button_get_type() and store the result in GtkToolButton_private_offset
- replace g_type_class_add_private() inside gtk_tool_button_class_init() with
g_type_class_adjust_private_offset (klass, &GtkToolButton_private_offset)
"
When we decide to fall back because the settings portal
is not present, adhere to that decision elsewhere. And
treat the fontconfig-timestamp like the other special-cased
settings, with G_TYPE_NONE.
Under Wayland, we are currently directly using GSettings
for desktop settings. But in a sandbox, we may not have
access to dconf, so this may fail. Use the new settings
portal instead.
As GSettings now supports session-specific defaults, GNOME Classic
no longer uses a separate schema and the decoration layout is always
determined by the regular schema.
This essentially reverts commit add67b516c (although the code was
moved since then).
https://gitlab.gnome.org/GNOME/gtk/merge_requests/400
By returning a default surface. The situation where there's no
currentContext arises when GtkCSS is trying to determine the
layout sizes so no actual display is necessary.
Closes: #1411
Do not lie to W32 about the formats that we provide or accept.
Originally the logic behind such lies was that GdkPixbuf allows us to
convert any supported image to BMP or PNG, and therefore we should
announce that we always provide/accept BMP and PNG along with other
formats.
But that's not how it works. GDK has built-in serializers and
deserializers for all pixbuf formats (where it just invokes GdkPixbuf
API) and will use them automatically to read or write GdkTexture
objects (internally wrapping GdkPixbuf objects where necessary). The
encoding and decoding of images is handled
by GdkContent(De)Serializers, backend has nothing to do with it.
Therefore W32 GDK backend should only offer formats that it can
actually do conversion for by itself (such as image/bmp <-> CF_DIB,
or text/uri-list <-> CFSTR_SHELLIDLIST).
and use 150 as natural-width.
Currently there's no way for a GtkEntry to be less
than 150px wide (apart from using "width-chars" property),
this is too much for a default minimum-width, an app
developer may need to have a shorter GtkEntry, for example
when the UI it's been shrunk by the user (see [1]) or when
you want to match the size of another widget (which is less
than 150px) see [2] for Evince bug on using
gtk_combo_box_new_with_model_and_entry() for PDF forms where
GtkEntry of ComboBox is too wide and doesn't match the combo
list width.
Using "width-chars" property may be a workaround to obtain
a short minimum-width for the entry, but is not a proper
solution for the mentioned cases as you may not know how
short your GtkEntry will be, or the fact that using "chars"
as a width unit is not pixel accurate.
Curious note: the commit that introduced the GtkEntry
minimum-width to be 150px is from 20 years ago, see
https://bit.ly/2ySEfK4
[1] This change was already suggested by Benjamin Otte
in a blog comment https://bit.ly/2J96wRo
[2] Fixes issue evince#1002
- Selection mode does not get the special devel styling.
- removed teh last-child() selector for it doesn't work anymore.
Better style all section of the headerbar than none. Proper fix pending.
https://source.puri.sm/Librem5/libhandy/issues/57
Instead we just cache the monitor number and get
out of it the nsscreen when it is needed. This is
a requirement since it nsscreen it is not supposed
to be cached.
Fixes: https://gitlab.gnome.org/GNOME/gtk/issues/1312
This was noticed in Firefox and demonstrated using a GtkBuilder ui file.
buildable_add_child() calls set_tab_label(), but the latter did nothing
to update the menu_label corresponding to that tab with the new text.
Using Builder to populate the tab child, only tabs other than last got
the right non-default labels, and even that was mostly coincidental, as
adding the main child called update_labels() via real_insert_page(), so
it took effect when the 2nd last main child is added, updating the rest
but leaving the last with the default label, not that given in Builder.
Fix by factoring out the code from child_reordered() to a new helper
menu_item_recreate() and calling that in set_tab_label(), so that
whenever the tab_label is updated, so is its corresponding menu_label.
This fixes the reported case and presumably others that we could write.
fixes https://gitlab.gnome.org/GNOME/gtk/issues/1397
Comments matched to reassure the compiler that fallthrough is
intentional are supposed to precede the case or default keywords, at
least in GCC, so the one here did not suppress the warning with GCC. We
can just the if condition and put the comment at the end to solve that.
https://developers.redhat.com/blog/2017/03/10/wimplicit-fallthrough-in-gcc-7/
This meson port is not upstream yet, so a wrap file is not included.
Upstream has expressed interest but the port hasn't been tested on all
platforms yet. Will be added when it gets upstreamed.
Link to WIP port: https://github.com/centricular/harfbuzz
We do this for every single node, which is a little costly, especially
since the common case for the modelview matrix these days is a simple
translation. So, check whether the new modelview matrix is only a
translation matrix and if so, don't do a full matrix multiplication per
node.
Commit 64a489ad inadvertently introduced a regression that broke Korean
text input because the changes there resulted that only the last input
string that we have from ImmGetCompositionStringW() for each time the
commit signal is emitted is kept, and also as a result the final Korean
character that is input by hitting space is also lost as a result, as we
didn't check for whether we are done with preediting.
Fix these issues by doing the following when we receive the
WM_IME_COMPOSITION message with GCS_RESULTSTR from Windows:
-Do not emit the commit signal during WM_IME_ENDCOMPOSITION, and...
-Emit the commit signal anyways, as we did before c255ba68, however...
-We still save up the string to commit, because we need to re-compute
the cursor position when we do ->get_preedit_string(), which needs to
take the GCS_RESULTSTR string we get from WM_IME_COMPOSITION into
account as well, so that we avoid getting the Pango criticals that
occur during Chinese (and most likely Japanese) input as the cursor
position is out-of-range.
Fixes issue #1350.
The previous type was a pointer to a pointer, which seems to be a copy-paste
error from GtkBuildable.custom_tag_start which is an out parameter. It was
always cast in use so this is an API break, but not an ABI one.
This leverages the normal input context switching mechanism in GTK
by making it think that the gtk-im-module setting changed.
The backend returns gtk-im-module value as "ime" if W32
IME API says that an IME is in use. Otherwise it returns
and empty string - this still triggers an input context
switching code, which, not being able to create the desired context
(which is and empty string), falls back to looking at current
keyboard layout (currently that code is still a FIXME).
Paired with the code that signals gtk-im-module change on keyboard layout
switches, this is sufficient to make GTK capable of switching to
the appropriate IM context at runtime. At least, the kinds of context
that specify languages for which they are used automatically by default
(once locale matching is implemented), and the IME context.
Loading other kinds of IM context might still work via specifying
the gtk-im-module setting in gtk ini file, but doing so will likely
make GTK incapable of using the IME context that is used
for Korean, Chinese and Japanese (and some other languages).
Until someone figures out a way to actually change gtk-im-module
setting on Windows at runtime with meaningful values, the behaviour
introduced by this commit seems like a sufficient workaround.
GNOME Shell 3.32 will remove support for the app menu
so we need to move its contents to the primary (hamburger)
menu.
widget-factory already had a primary menu.
The only item in the app menu was About.
https://gitlab.gnome.org/GNOME/Initiatives/issues/4
The gtk_stack_snapshot_slide() function dereferences the
last_visible_child pointer without proper != NULL ckeck. This might
result in NULL pointer dereference and crash if last_visible_child is
invalid.
Add a != NULL check before dereferencing the pointer.
There’s a short-path done for focus rectangles, but it can be taken in other conditions, and then fail occasionally to render a dashed line if the border-width is too big.
Variable, added, would be a garbage value if model is NULL and
the following code, if condition, use the uninitialized variable.
A side effect could be occurred by that.
To avoid, the variable is initialized to zero.
The executable is called autotestkeywords, so we shouldn't try to run
an executable named keywords. Also rename the metadata file to match.
Signed-off-by: Simon McVittie <smcv@debian.org>
The installed-tests are now namespaced as gtk-4.0 to avoid colliding
with GTK+ 3, but these files weren't updated.
Signed-off-by: Simon McVittie <smcv@debian.org>
After removing elements, there were a few cases where the tree wasn't
properly balanced which could further down violate assumptions about the
layout.
Attached is the original testcase that triggered it. I didn't bother
simplifying it.
Commit 359df028be changed the
code to send GDK_SCROLL_SMOOTH with deltas instead of
GDK_SCROLL_(UP|DOWN|LEFT|RIGHT).
Windows defines deltas inversed for vertical direction
(positive values mean the wheel was turned forward)
but not for horizontal direction
(positive values mean the wheel was turned towards the right).
This commit fixes behavior as both axes were inverted previously.
Commit d64467b334 changed the
code to send GDK_SCROLL_SMOOTH with deltas instead of
GDK_SCROLL_(UP|DOWN|LEFT|RIGHT). Change it again, to send
both the GDK_SCROLL_SMOOTH and the GDK_SCROLL_(UP|DOWN|LEFT|RIGHT)
event separately (with the discrete event marked as emulated),
as this is what other backends (such as wayland) do.
Up until now when allocating the child it only used the natural size
while the measuring also used the minimum size, resulting in a clipped
child when animating if the child had different minimum size and
natural size. This was an obvious case when using labels that had
ellipsization.
This commit gives full allocation to the child by inverting the size
the revealer reduces from its animation progress.
Code done by Benjamin Otte.
Closes: https://gitlab.gnome.org/GNOME/gtk/issues/635
The complexity with model items vs row items is really confusing. Add to
that treelistmodel position vs child model position vs parent position,
and you're so confused, even the best naming can't help.
And once you're there, consider passthrough vs non-passthrough...
When passthrough is enabled, it should return the GType
of the child GListModels; when disabled, it should be
GTK_TYPE_TREE_LIST_ROW.
The conditions are inverted however, causing a few
warnings to trigger.
Fix that by returning the correct GType.
Most of the code creating the two types of dialogs (open file,
choose folder) is the same. This refactors the common code into a
helper method. This also makes it easier to add other chooser types
for this test (e.g. save file).
This is way more complicated than it should be, because it requires
manually limiting the number of open file enumerators.
On the other hand, it exhaustively tests the items-changed emission of
all involved listmodels because those signals come in pretty much
randomly.
It's also 50% slower than the sync version, with the caeat that the sync
version only shows the UI after it's done loading, while this version
shows it right away.
This model just takes an object and a property name and recursively
looks it up. In particular, I want it for:
widget, widget.parent, widget.parent.parent, ...
The code gets rid of the GtkTreeView and replaces it with a GtkListBox.
Most of the logic is now done via GListModel subclasses.
A big change is that this new list is now tracking updates itself and
doesn't need to be manually updated. All code that used to cause rescans
or add forgotten objects to the tree has been removed.
If objects are missing from the object tree, the logic for tracking them
needs to be added.
This patch does multiple things:
1. Add a custom persistent per-row object.
2. Move all per-row API to that object. This means notifications are now
possible.
3. Add a "passthrough" construct-only property to the TreeListModel that
influences if the model returns these new object or passes through
the ones from the model.
This greatly simplifies the code needed to be written for widgetry,
because one can just connect the per-row object to the expanders that
expand and collapse rows.
As an added power feature, these objects can also be passed through
further models (like filter models).
It also adds kind of a hack to Adwaita to make the test look neat.
Let separators be declared as sidebars to have the same style as those
drawn by GtkStackSidebar. This also let them handle the selection-mode
class, whether they are assigned it or they descend from something in
selection mode.
Also drop setting the selection mode color for non-sidebar separators.
This is convenient when building a custom sidebar using a GtkSeparator
and to extend a sidebar to the title bar.
This is needed to work around headerbar sliding animation issues without
refactoring Adwaita's support of titlebars and headerbars as it may
break applications.
https://gitlab.gnome.org/GNOME/gtk/issues/1264
This step was missed before, again.
SASS 3.6 emits rgba(0, 0, 0, 0) instead of transparent, so it wants to
change those too, but that patch was only committed in March and isn't
being backported to the previous stable, so I don't know if others'
versions will do the same - so until it's shown that anyone else (A) is
regenerating CSS and (B) also has 3.6, I'm skipping those changes. See:
c287f312ac
A number of applications want to track the state of the screensaver.
Make this information available as a boolean property. We only listen
for state changes when ::register-session is set to TRUE.
This is implemented for unsandboxed D-Bus access by talking
directly to org.gnome.ScreenSaver or org.freedesktop.ScreenSaver,
and for sandboxed D-Bus by using a (new) portal API.
A Quartz implementation is missing.
Currently, GtkRevealer clips the child if the transition type is
sliding, regardless of whether the animation had already ended. An
example where that is a problem would be in Nautilus: the file
operations popover button is animated on reveal to draw attention, but,
given that the button is in turn stashed inside a revealer with a
sliding animation, things suddenly fall apart.
Instead, use a popup and gdk_surface_move_to_rect.
I have not tried to reproduce all details of the old
positioning logic, but moving the popup above/below
the entry works as before.
In order to make tooltip positioning portable, make use of the
move_to_rect API. Some semantical changes are made, as identical
semantics cannot be implemented using the move-to-rect API.
Primarily the implemented semantics are:
Position the tooltip in the center pixels slightly below (defaults to 4
units below) the tooltipped widget. This is always the case for keyboard
driven tooltips; the case where it tries to avoid the pointer cursor is
not implemented.
For pointer position triggered tooltips, implement the following
additional semantics:
Use the current cursor size to determine the padding used to enlarge the
anchor rectangle. This is to try to avoid the cursor overlapping the
tooltip.
If the anchor rectangle is too tall (meaning if we'd be constrained
and flip on the Y axis, it'd flip too far away from the originally
intended position), rely only on the pointer position to position the
tooltip. The approximate pointer cursor rectangle is used as a anchor
rectangle. Ideally we should use the actual pointer cursor rectangle
(image used as well as hotspot coordinate), but we don't have API to
get that information.
If the anchor rectangle isn't to tall, just make sure the tooltip isn't
too far away from the pointer position on the X axis.
Closes: #134Closes: #432Closes: #574Closes: #579Closes: #878
Let's just use the fact that a window was mapped as a subsurface to
remap it above another transient parent instead of relying on the more
complicated 'should-map-as-subsurface' helper function.
Set delta_x or delta_y for GdkScrollEvent.
HIWORD (wParam) in WM_MOUSE(H)WHEEL is the scroll delta.
A delta value of WHEEL_DELTA (which is 120) means scrolling
one full unit of something (for example, a line).
The delta should also be multiplied by the value that the
SystemParametersInfo (SPI_GETWHEELSCROLL(LINES|CHARS), 0, &value, 0)
call gives back, unless it gives back 0xffffffff, in which case
it indicates that scrolling is page- or screen-based, not line-based
(GDK doesn't support that at the moment).
Also, all deltas should be inverted, since MS sends negative deltas
when scrolling down (rotating the wheel back, in the direction of
the user).
With deltas set the mode should be set to GDK_SCROLL_SMOOTH.
Fixes issue 1263.
The intention of this check was to skip the keyword
test if no c++ compiler is found. But the meson
docs say that add_languages() will abort unless we
pass required: false.
because filesystem readdir order is indeterministic.
Without this patch, building openSUSE's balsa package
had variations between builds in /usr/share/balsa/icon-theme.cache
When calling PickColor on org.gnome.Shell, we get back an "a{sv}", which
GDBus provides to us as "(a{sv})".
At the minute we're not unpacking this tuple, and so picking fails with
messages like:
GLib-CRITICAL **: 13:38:19.439: g_variant_lookup_value: assertion 'g_variant_is_of_type (dictionary, G_VARIANT_TYPE ("a{s*}")) || g_variant_is_of_type (dictionary, G_VARIANT_TYPE ("a{o*}"))' failed
Gtk-WARNING **: 13:38:19.439: Picking color failed: No color received
Let's unpack it.
The additional assignment to the old result variable just adds an
indirection even though we know the point where we assign it in all
cases. Just pass the values out and return in those cases instead.
Previously, GtkBin was only snapshot'ing its one and only child, but
nowadays it doesn't implement snapshot at all and the default
implementation in GtkWidget just snapshots all child widgets, which is
exactly what the implementation in gtkmodelbutton.c was doing.
Since the original implementation was likely based on GTK+ 3, the change
in default visibility might have not been considered, which results in
all rows suddenly sporting a visible spinner when opening a fresh file
chooser.
Unparenting a GtkListBoxRow can drop its last reference, which
will free its memory. Right after unparenting, though, we were
accessing the row's iter - which assumes that the row is still
alive. This causes a crash when, for example, binding two or
more models to the listbox.
Fix that by storing the iter in a variable, and not trying to
access it after unparenting. After unparenting, the variables
that are potentially garbage were explicitly assigned NULL for
clarity.
Fixes https://gitlab.gnome.org/GNOME/gtk/issues/1258
bindings now treat identifiers and strings the same way.
The only difference was that one allowed lookup of enum/flags by name
while the other didn't and g_warning()ed. Now both work.
Perform scrollbar visibility checks through a motion controller,
always based on GtkScrolledView-relative coordinates. The captured
event handler remains though, for a tiny bit of GDK_SCROLL event
handling.
Use a distinct key controller so we correctly handle navigation
across matches and search cancellation. As the events are forwarded
to the search_window, those need to be pushed down the entry manually.
CSD titlebar are included in the focus-chain. The logic used makes sure that the
initial focus avoids the titlebar, but tabbing around will eventually get there.
This logic fails in case the window has no other focusable widgets apart from
the ones in the header-bar. If this happens keynav focus will be lost. To handle
the above scenario, we need to fallback to focus the header-bar (if any).
Fixes: https://gitlab.gnome.org/GNOME/gnome-software/issues/404
Drop the drag-highlight and drag surfaces. The highlighting
is broken anyway, so just drop it for now. And for dragging
the header button, we can just position it properly, that
works just as well as this reparenting approach.
This fixes a potential leak of a PangoAttrList that is set when chaining
up to the parent get_preedit_string(). We check to see if the attr list
was created and reuse it instead of leaking the previous value.
Remove gtk_menu_popup_for_device() and gtk_menu_popup(), as they cannot
be implemented in a portable manner by all backends. They have been
deprecated for proper alternative APIs for some time, so lets remove
them now before its too late.
While at it, fix the example documentation for mapping a menu.
This is a temporary measure to make the check-icon-names
test not fail in ci. We still have to figure out the best
way to include a core icontheme with GTK+.
GLib master propagates argument types in g_clear_pointer(), which causes
the usual function pointer casts to GDestroyNotify to trip compiler
warnings. Additionally, this commit changes some cleanup functions where
appropriate (wl_data_source_destroy ->
gtk_primary_selection_source_destroy for struct
gtk_primary_selection_source).
GtkEntryCompletion can rapidly release and claim ownership of the
primary selection. This generates multiple XFixesSelectionNotify events,
first stating that no one owns the selection, then another stating that
we own the selection. The notification that no one owns the selection
causes GtkEntryCompletion to deselect the text, breaking inline
autocompletion.
This fixes it by ignoring any XFixesSelectionNotify with a timestamp
earlier than our clipboard timestamp.
Fixes#14
file_uri_deserializer splices a memory stream, as opposed to
string_deserializer, which uses a converter and filter stream. This
commit fixes erroneous use of GMemoryOutputStream as
GFilterOutputStream.
Since the function is usually called from GtkWidget::drag-{begin,end} handlers,
taking a GdkDrop does not work, especially given that
::drag-action-requested is emitted without checking the type.
Fixes https://gitlab.gnome.org/GNOME/gtk/issues/1220
The previous attempt at removing configure events entirely
was causing some dialogs not to show up under Wayland.
Presumably due to ordering issues with emitting ::size-change
out of the backend.
Instead, keep configure events in the event queue, but handle
them on the gdk side. This keeps the ordering intact, while
still removing configure events from the api. The dialogs
show up now.
The opaque region is only set when the background color is opaque. So
we need to do something about it when the background color changes.
However, in the case where a size allocation is going to happen, we
already do this update in size_allocate(), so in that case avoid doing
it twice.
Instead of instantly invalidating, we now cache the old render node and
do the update in an idle handler.
While that gives us a 1 frame delay, it avoids all the tricky things
like queueing resizes while resizing or queueing draws while drawing.
The only remaining issue (and a *big* one at that) is that a nested
widget paintable will now cause the widget to snapshot its previous
render node when creating a new one. And that one will snapshot its
previous render node, and that one will...
And nothing so far breaks this recursion.
Change GdkDrag::action to GdkDrag::selected-action, which is
more clearly different from actions, and follows the existing
name of the struct field and getter.
This lets us drop the ::action-changed signal for the
property change notification. But, can just as well move
the signal class handers which just update the cursor
to the ::action setter. No need to do this in the backends.
Some of the _diff implementations did a whole bunch of work just to
throw it away afterwards and invalidate the entire union of the two
render nodes, most notably the two clip nodes. Fix this to only call
gsk_render_node_diff_impossible if the previous if-condition is FALSE
and not always.
This is still fallout from the bin_window removal. We aren't moving the
GdkWindow/GdkSurface anymore so we have to account for the scrolling
ourselves.
Since those are widgets and widgets need to be size-allocate'd properly,
we need to queue an allocate, as well as actually add the hadjustment's
value to the column x position.
Fixes#1202
The purpose of a searchbar is to start a search on visible widgets when
a key is pressed. Starting a search on e.g. a stack page that is not
visible at all is not very useful.
... if none of the debug displays have any debug flags set. This way, we
can ignore the first parameter to e.g. GTK_DISPLAY_NOTE, which is
usually a call to gtk_widget_get_display.
Before this patch, gtk_widget_get_display was the slowest part of
gtk_widget_query_size_for_orientation.
We were leaking the GBytes for the image memory, which is a
noticeable memleak to anyone who's casually running a memory monitor.
Go KDE users!
Closes#1200
The min size on the oriented axis used to come from style props with
default values in the source file, used if the theme did not provide a
min size in CSS. When the style props were removed, so was any notion of
a minimal size for proressbars' main axis, meaning that now progressbars
without expand or any other source of min size were just tiny specks.
The right place to do that was always the theme, so in our themes now,
fix that by copying the old default values for the style properties; see:
https://gitlab.gnome.org/GNOME/gtk/issues/1191#note_259393https://gitlab.gnome.org/GNOME/gtk/blob/gtk-3-24/gtk/gtkprogressbar.c#L92
The result should be the same in that (A) the min size is now what it is
in GTK+ 3 & (B) an app/user can override the theme exactly the same way.
Close https://gitlab.gnome.org/GNOME/gtk/issues/1192
7733f646d6 renamed GdkDragContext to
GdkDrag, which broke the docs, as a reference to
gdk_drag_context_get_type() still exists. This commit renames the type
accordingly and adds GdkDrop.
(A) Put a space in "scrolled window" like the other doc comments
(B) Say "i.e." rather than "ie."
(C) Fix grammar from "makes [...] exactly reaches" to "exactly reach"
Rename gdkdnd.h to gdkdrag.h, to go along with gdkdrop.h
This commit includes the necessary updates to the X11, Wayland
and Broadway backends. Other backends have to be updated separately.
This is to go along with the newly introduced GdkDrop.
This commit includes the necessary updates to the X11, Wayland
and Broadway backends. Other backends have to be updated separately.
It looks like this got dropped during the move from autotools and never
restored. I can see why, since making it work wasn't a hugely fun task!
Notes on some less then obvious details:
* PlacesSidebar is private now and didn't seem to be to be particularly
easy to adapt to, so this moves to checking for it by name, not TYPE.
I couldn't find a (fast) better way; if you know how, please clean up
* added 2 casts to avoid warnings from the new type-propagating ref()
* GdkClipboard and GdkContentProvider need some properties dodged
* GtkToolItemGroup is gone
* fixed indentation and used TypeName:property-name syntax in a print()
Also update the cursor surfaces of every seat when an output changes
scale. This could for example happen when a monitor scale is changed via
Settings.
If we check it too early, we will not unset priv->draw_neeeded, which
will then cause queue_draw() calls to not have an effect later. And that
causes changes in opacity to not register.
Closes#1180
The comment above explains neatly why subclassing GtkButton for
GtkColorButton was a bad idea. Nowadays it's a GtkWidget subclass
containing a GtkButton so let's remove the special case here.
Binds this property to the button's label, allowing a model button to
have text with markup.
This will be convenient for buttons like 'Online Accounts <sup>↗</sup>'.
When deciding whether or not to emulate a press event, we're translating
the last event coordinates and mutating the given event structure
unconditionally.
We should modify the newly created GdkEvent copy, since it's what we're
going to use when emitting the press event.
This avoids mutating a constant GdkEvent and global state, and also
avoids a compiler warning.
The idea is that GTK+ 4 will be an epoch, API-wise.
Everything that was around for 4.0 has been there
since the beginning of the epoch and doesn't need
markers.
If the parent get_preedit_string implementation returns a nonnull
zero-length string, then we ignore it, which is almost fine. We have to
free it, though.
Fixes#1174
Expanders used to be 16px high. With the move from the gtk2 rendering
to gtk3 rendering they shrunk to 12px, making them hard to see, because
it's now the icon which is 16px high and the icon contains transparent
borders.
This makes the HighContrast theme use 24px icons instead, to restore
16px expanders. This may expander some containers a bit.
Closes#1046
GtkTextHandle was neglected by whoever removed the ::draw signal,
leaving it entirely broken. Update to using GtkGizmo so we can
implement snapshot of text handles.
Input has received a revamp too, handling is done through a
GtkGestureDrag and coordinate calculations simplified by storing
the delta to the hotspot on ::begin instead of ::update, as this
value is constant throughout the gesture. Widget state management
on crossing events happens implicitly, so no longer needs to be
done here.
Last but not least, CSS has also been updated so handles are
rendered at the correct size and proportion, and with the padding
that code expects of it.
Set up a gesture on the sidebar rows to detect pointer clicks on
it. The row DnD management has been moved to the row widget itself,
it makes more sense even if the drag is began from the sidebar widget.
We still need a drag gesture both on front (capture) and back (bubble)
to handle dragging from both the GtkWindow widget and chrome in the
headerbar. But we can do it through 2 drag gestures, instead of special
event handling code.
This has been broken since we switched key event delivery to follow
the same semantics than pointer/touch. There, GTK+ grabs will influence
the topmost widget during event delivery, rendering the toplevel
unable to handle key navigation. The toplevel must handle those key
events in an explicit manner then.
We don't render the keyboard focus rectangle yet, but I assume that's
something else.
Use an event controller on GtkFontChooserDialog, a nice side effect
is that we can use gtk_event_controller_key_forward() and
gtk_search_entry_set_key_capture_widget() instead of passing events
around for dialog search.
Instead of doing all handling manually in the ::event vfunc,
set up drag/multipress gestures on icon images, and implement
emission of ::icon-press/release and DnD there.
As a side effect, the GdkEvent field in ::icon-press/release
signals has been dropped. Callers that might be interested on it
may still use gtk_get_current_event*().
This isn't really necessary, if keyboard focus forcibly goes somewhere
else we will get ::grab-notify, which is sufficient to deactivate the
button again.
Selected rows in tree views in HighContrast have a background colour the
same or nearly as the normal text colour, so we cannot let entries in
such rows have transparent backgrounds, or the text inside the entry
becomes nearly or totally impossible to see.
Dodge this by giving entry.flat inside treeview and with :focus the
$base_color, which is different from the text & so lets that be seen.
https://gitlab.gnome.org/GNOME/gtk/merge_requests/125
This reverts commit 8e74eb382f.
This code is not necessary. It worked around a bug in graphene where
graphene was requiring stricter alignment than glib allocators could
guarantee.
The else case was wrongly resetting the accessible description on the
primary icon, which might not exist and can therefore cause a crash.
https://gitlab.gnome.org/GNOME/gtk/issues/1160
This functionality is similar to Linux's memfd. It creates anonymous shared memory without touching the filesystem, which allows it to work in Capsicum capability mode (sandbox).
The claimed status check should happen after ::end is emitted,
as the gesture may deny the sequence that much late. In this
case the event should keep propagating.
https://gitlab.gnome.org/GNOME/gtk/issues/1159Closes: #1159
We are poking again into the event propagation machinery, which
expects events in toplevel coordinates. Since we can't fetch the
original event back at this point, translate the coordinates
back to the toplevel so the emulated press ends up in the right
place.
https://gitlab.gnome.org/GNOME/gtk/issues/1159Closes: #1159
* There's no GdkDragContext->dest_surface anymore.
Add dest_window field to GdkWin32DragContext,
and use that instead.
* Remove unused function prototypes
* Add more comments
* Rename variables and fields from 'window' to 'surface'
where appropriate
* Fix header indentation a bit
* Try to ensure that uninitialized/unknown handle variables
and fields are set to INVALID_HANDLE_VALUE instead of NULL,
as there may be cases where NULL is a valid handle value.
This might be foreign Windows and we don't want to create surfaces for
those.
Also, stop using GdkDragContext.dest_surface, that variable is meant to
go away.
In particular, this patch removes:
gdk_surface_get_events()
gdk_surface_set_events()
gdk_surface_get_device_events()
gdk_surface_set_device_events()
Event masks so far still exist for grabs.
GdkDragContext => GdkDrop
This is all in preparation of separation of the drag and drop.
Also, don't check for GDK_DRAG_PROTO_XDND anymore - it's the only
possible value for the protocol on the target side.
Use the new method of connecting to the xevent signal instead.
Also, don't consume the xevent, there might be other code listening for
it. And we don't use PropertyNotify in the generic code path anymore, so
it'll just be ignored there.
* Remove clipdrop->dnd_target_state, it's not used anymore
* Remove non-functioning _gdk_dropfiles_store(), store dropfiles
list in GdkWin32Drop instead
* Fix multiple comment typos
* Fix _gdk_win32_get_clipboard_format_name_as_interned_mimetype() to
leave names that look like mime/types alone
* Refactor _gdk_win32_add_w32format_to_pairs() to populate
GdkContentFormatsBuilder directly, instead of making a GList
* Rename context -> drag (still using GdkDragContext type,
but [almost?] all variables and comments say "drag" now)
* Rename GdkDropContext -> GdkDrop
* Rename some parameter names for clarity
* Rewrite local protocol to look more like OLE2 protocol
instead of mirroring the structure of the X11 API.
* Add handle_events field to GdkWin32DragContext,
to shut off event handling (temporary fix until GTK is patched up)
* Remove _gdk_win32_drag_context_find() - the drag object is stored
in GdkDrop instead. Use _gdk_win32_find_drag_for_dest_surface()
to get it initially.
* Remove target_ctx_for_window, droptarget context is stored
in the surface instead.
* Call gdk_drag_context_set_cursor() just like wayland backend does
(slightly broken for now)
* Clean up the action choosing code (filter source actions by using
keyboard state, pass that to GTK, get all actions supported by GTK in
response, match them up with filtered source actions, return the
result, falling back to COPY in case of multiple actions)
* Check drag_win32->protocol instead of the use_ole2_dnd variable where
possible
* Remove protocol checks from functions that are only used by the local
protocol
* Use event state to manufacture the keyboard state for WM_MOUSEMOVE
* Change function names printed by GDK_NOTE to name the actual
functions, not their theoretical generic GDK stack ancestors
* Consistently use drag_win32 and drop_win32 variables instead of a mix
of that and win32_drag/win32_drop
* Return FALSE from button handler to ensure that GTK gets the button
event to break implicit grab
* Emit leave event on failed idroptarget_drop() calls
Instead of looking at the list of contexts, just look at the current
drop context. There is only one, after all.
Then remove the is_source argument from gdk_drag_context_find().
The filters now return TRUE/FALSE and no longer a GdkFilterReturn. They
also don't conform to the GdkFilterFunc typedef anymore but instead take
the arguments that they need.
This is not needed because GTK must be run in the main thread these days,
which is the same one that runs the main loop. So when this function is
called, the main loop is awake.
https://bugzilla.gnome.org/show_bug.cgi?id=550989
This uses the new method without GDK_ACTION_ASK:
Either it is a single action (queryable via gdk_drag_action_is_unique())
or it is not and then the drop target has to make a decision
(potentially by asking someone).
The ultimate goal of this patch series is to split GdkDragContext into
GdkDrop + GdkDrag classes for the destination and source side of a dnd
operation.
The refactoring is meant to work something like this:
1. Introduce GdkDrop as a base class
2. Make all drop related code (like GdkEvent) use GdkDrop instead of
GdkDragContext. Move/duplicate APIs to allow that.
3. Port all drop contexts in the backends from GdkDragContext to GdkDrop
4. Delete all APIs in GdkDragContext that aren't needed anymore.
5. Make GdkDragContext no longer a GdkDrop subclass
6. Rename GdkDragContext to GdkDrag
In 01455399e8 ("gdk: do not deactivate surface on keyboard grabs"), we
made gdk avoid deactivating surfaces when another application takes a
keyboard grab, by using has_focus_window instead of has_focus. That however
broke activating surfaces when the gdk application acquired a grab itself,
in which case has_focus_window is false but has_focus is true.
We thus actually need to use both: surfaces should be activated either
because we have normal keyboard focus, or because we grabbed the keyboard.
This also renames HAS_FOCUS to APPEARS_FOCUSED to better reflect its
role.
Fixes#85
There is no reason why we shouldn't pass this flag every time
Z-order changes. We have separate routines that are used to
maintain relative Z-order, so it should be completely OK to
pass SWP_NOOWNERZORDER to let the OS know that it shouldn't try
to maintain relative Z-order of the windows when raising them.
Pass SWP_NOOWNERZORDER when rising TEMP surfaces to the top. This ensures that
they don't drag anything else to the top with them. The use-case for this is
a tooltip appearing for a non-foreground surface, causing said surface to rise
above other surfaces, some of which maybe foreground at the moment.
https://bugzilla.gnome.org/show_bug.cgi?id=784766
According to the old new thing[0], we should use the instance handle
of the GDK/GTK DLL when registering GDK-specific types in the system.
Using the instance handle for the whole application in these circumstances
is not an error, but can potentially clash with the types registered
by the application itself.
Also, extract window class icons from the GDK/GTK DLL, not from the
application executable.
[0]: https://blogs.msdn.microsoft.com/oldnewthing/20050418-59/?p=35873
The argument is eventually passed to g_conv(), so it should
be the charset, not the mime/type. Without this change the
contentype converter will fail to convert UTF-8 strings to, say,
CP-1251 later on.
Unconditionally putting 'gdkwayland_inc' in src_dir argument of gtkdoc
call tells gtkdoc-scan to scan source files in a non-existent build
directory, gdk/wayland. To avoid causing build failure when a specific
backend is disabled, we should include directories conditionally.
Otherwise gcc complains when we use these as arguments to g_new() on
32bit architectures with:
../gtk/gtkcomposetable.c: In function ‘gtk_compose_table_list_add_array’:
/usr/include/glib-2.0/glib/gmem.h:217:10: warning: argument 1 range [2147483648, 4294967295] exceeds maximum object size 2147483647 [-Walloc-size-larger-than=]
__p = g_##func##_n (__n, __s); \
~~~~^~~~~~~~~~~~~~~~~~~~~~~~~
/usr/include/glib-2.0/glib/gmem.h:279:42: note: in expansion of macro ‘_G_NEW’
#define g_new0(struct_type, n_structs) _G_NEW (struct_type, n_structs, malloc0)
^~~~~~
../gtk/gtkcomposetable.c:851:22: note: in expansion of macro ‘g_new0’
gtk_compose_seqs = g_new0 (guint16, length);
^~~~~~
/usr/include/glib-2.0/glib/gmem.h:96:10: note: in a call to allocation function ‘g_malloc0_n’ declared here
gpointer g_malloc0_n (gsize n_blocks,
^~~~~~~~~~~
When a remote instance of a GTK application implementing the Startup
Notification protocol gets spawned it will pass the startup sequence
ID as "platform data" to the main instance. Thus, we need to make sure
that the startup sequence gets completed in that case, since the remote
instance won't do it by itself, since it won't map any top level window.
Checking for this "platform data" in the implementation of the after_emit()
virtual method in the primary instance should be a good place to do so, since
the existence of such data proves that a remote instance has been spawned.
https://gitlab.gnome.org/GNOME/gtk/issues/1084
The DESKTOP_STARTUP_ID gets cleared early after the process is spawned,
meaning that it's too late at add_platform_data() to pick it up and send
it over to the primary instance, as it will be always unset at that point.
To solve this, we use the new gdk_display_get_startup_notification_id()
method to pull the startup notification ID for the application, if present,
out of the display and pass it over to that primary instance.
https://gitlab.gnome.org/GNOME/gtk/issues/1084
Includes implementation for Wayland and X11, which are the only backends
implementing the Startup Notification Protocol, returns NULL otherwise.
https://gitlab.gnome.org/GNOME/gtk/issues/1084
Similar to what has been done recently for DESKTOP_AUTOSTART_ID [1],
we need to get rid of this call to g_unsetenv() in the displays'
backends for X11 and Wayland, so that it's guarantee to happen any
thread is created, while still being accessible when needed.
Let's stash the value of this environment variable when loading the
GDK library, and provide a private method so that it can be retrieved
from the displays' backend when implementing gdk_display_make_default().
[1] https://gitlab.gnome.org/GNOME/gtk/commit/22269902
Closes: https://gitlab.gnome.org/GNOME/gtk/issues/979
check_dir_mtime() is called by ftw() and is given
the real stat struct, not its glib version (which may
or may not be the same as "struct stat").
This is irrelevant for MSVC (it has no ftw()) and
works correctly for MinGW-w64 (which declares stat
structures correctly). If mingw.org complains, add
a special ifdef for it later.
It's quite old, but mostly harmless (both "message == WM_KEYUP"
and "message = WM_KEYUP" evaluate to not-FALSE, and message
value is not used after that line).
Any data that is later fed to graphene must be
allocated with proper alignment, if graphene
uses SSE2 or GCC vector instructions.
This adds custom array code (a streamlined copy
of GArray with all unnecessary bells and whistles removed),
which is then used for the state_stack instead of GArray.
There's also a runtime check for the size of GtkSnapshotState
itself being a multiple of 16. If that is not so, any array
elements past the 0th element will lose alignment.
There are probably struct attributes that can
make GtkSnapshotState always have size that is a multiple
of 16, but we'll burn that bridge if we cross it.
The check survived from GTK2 when that function could still return
GdkPixmap and GdkFont objects and was accompanied by this comment:
/* We may receive events such as NoExpose/GraphicsExpose
* and ShmCompletion for pixmaps
*/
and gtk_image_set_can_shrink().
Images are meant to always be icon-sized, they can never shrink below
that.
And images are icons, so they are meant to be square. If they are
not, we pretned that's by accident and keep aspect ratio.
This commit introduces GtkPicture, which is supposed to complement
GtkImage.
GtkImage will be adapted to always display an icon, while
GtkPicture displays regular imagery.
Instead declare a priv local. We should do this even if we don't remove
the priv pointer from GtkWidget entirely, just to stay consistent with
new code we introduce.
The code is mostly stolen from graphene.
Allocators support any alignment, but their implementation
only calls system aligned allocator functions if malloc()
is not aligned to 16-byte boundaries. If it is aligned,
the implementation just calls malloc() regardless of which
alignment is requested by the caller.
This can be fixed by saving the result of meson malloc()
alignment check and adding a few conditions to the implementation,
but right now GSK and GTK only need 16-byte alignment either way.
* A bunch of new variables for config.h.meson
* A check for aligned allocation being necessary at all
(graphene must use GCC vector instructions or SSE2)
* A check for C malloc() being aligned at 16-byte boundaries
* A check for a few special aligned allocator functions being
present and not being built-ins (posix_memalign is a builtin
in GCC, even on platforms where there is no posix_memalign
system function)
* Added -mstackrealign flag on Windows, since otherwise
stack variables may become unaligned when the stack briefly
passes through OS code (such as in various callbacks and
handlers)
Otherwise, requesting a min size in em where the equivalent in px had a
fractional part would lead to the widget getting allocated 1 too few px.
You could see this in the CSS property vs. allocation in the Inspector.
Note that margin/border/padding are left alone: the rationale is that we
do as browsers do, and Benjamin said we already do that for those,
whereas his tests on min-(width|height) showed otherwise. My subsequent
analysis indicated it to be far less clear-cut than that, but he remains
unconvinced that we should ceil() all the things! So just do these ones.
https://gitlab.gnome.org/GNOME/gtk/issues/1088
This is the API used by GtkMenu to properly position menus on the screen
without requiring GTK to query the menu window's position or the work
area of where the window is positioned. It makes it possible to position
popup windows properly when using Wayland.
Make this API available to external users so custom popup windows can be
positioned properly as well.
Related: https://gitlab.gnome.org/GNOME/gtk/issues/997
Causing a grab in the handler for ::pressed by, e.g., popping up a
context menu will cause the gesture to be canceled and, subsequently,
::end and ::released to be fired, all while the button is still
physically pressed. That results in no event being available to the
::released handler and garbage coordinates, given that
gtk_gesture_get_point() returns FALSE.
Emitting ::released can be avoided by checking the return value
gtk_gesture_get_point().
Querying the event sequence of a gesture will always yield NULL for
non-touch events, but passing NULL in to calls to
gtk_gesture_get_last_event() is a perfectly valid use case.
That code branch is meant to check for key events, seems obvious we want
GDK_KEY_PRESS, not GDK_BUTTON_PRESS (which also broke the branch right
below).
Makes us all able to dismiss popovers again.
We can just as well use notify::has-focus for the purpose of
focus tracking, and we can at the same time avoid emitting the
deprecated AtkObject::focus-event signal.
A recent dependency change in MSYS2 made it pull in vulkan, which made
meson think it's available but it somehow links against the system vulkan dll
instead.
Disable vulkan for now.
:climb-rate is not about what you get when you single-click on a button,
as this implied: it's what happens if you hold down a button or a key.
Fix the description of @climb_rate to new(), and while here, mention the
key in the blurb of :climb-rate itself.
The last round of patches to get the desired direction of value move in
response to scrolls/keypresses on scales had the inadvertent side effect
of giving the opposite direction on scrollbars. Seeing as gtkrange.c is
already a collection of hacks, add another so that fix only holds if the
instance is a GtkScale, since that is what those patches were aimed at.
Close https://gitlab.gnome.org/GNOME/gtk/issues/1065
We were mutating the list while iterating over it. This was not a
problem before since remove_controller just set the controller pointer
to NULL instead of actually removing it from the list of controllers.
We can avoid a signal connection per event controller (and the
EventControllerData struct) since every event controller knows the
widget it's attached to.
GtkTextView scrolls to the insertion point when the text
buffer signals a paste is done. This is wrong when there
are multiple views on the same buffer, and the paste
happened in another view.
To fix this, flip the handling of the scroll_after_paste
boolean to only be TRUE if we know that we want to scroll.
The gtk_app_chooser_dialog_set_heading() function do emit
notify::heading. Since the setter simply calls the function,
the setter itself shouldn't emit a notify signal by itself.
The bubble_timeout_id was reset only on some special case.
And so warnings were shown when the source is being tried
to be removed with the already removed id.
Fix this by unconditionally resetting the id on start of the function.
When an animated cursor was set and the previous cursor animation delay
happened to be the same, we wouldn't restart the animation timeout and
just return G_SOURCE_CONTINUE assuming the timer would continue. This
assumption is however only valid if the function was called from the
timeout, which is not the case.
Instead also arm the timer also if there is no previous timer active.
gdk_wayland_*_grab()/ungrab() would emit crossing events which translate
as focus_in/focus_out events for keyboard.
However, the ungrab() functions compare the native toplevel as this is
what gets the Wayland pointer enter/leave events with the grab surface,
so if the grab is issued on a child gdk surface, those won't match and
we would emit more focus_out events than focus_in.
This means that a widget such as spice-gtk which issues a keyboard grab
whenever the pointer enters the surface and releases the grab when it
leaves the surface would get uneven numbers of focus_in/focus_out
events.
Also, gdk_wayland_seat_ungrab() would not emit crossing events for
keyboard devices, whereas gdk_wayland_device_ungrab() does, which adds
even more potential discrepancies between focus_in/focus_out events.
To solve this problem, introduce two new helper functions which check
the relevant native surfaces to emit crossing events when needed that
get called evenly from both gdk_wayland_seat_grab()/ungrab() and gdk
_wayland_device_grab()/ungrab() APIs.
Fixes: https://bugzilla.gnome.org/show_bug.cgi?id=780422
Closes https://gitlab.gnome.org/GNOME/gtk/issues/792
The last parameter of the signal callback from .ui
is the template's object from which the class is
derived.
And so, we already have access to the window object.
Let's just use it.
Our flatpak-builder manifests include building Graphene from Git; since
we're building the GTK demos, it's pointless to build the Graphene tests
as well. Disabling tests and benchmarks avoids pointless installations
inside the Flatpak build repo that will just be removed by the time we
bundle the demo.
1. Include the broadway renderer (so we can test it properly fails on
Wayland or X11)
2. List all potential renderers, print useful information when Vulkan
is not compiled in instea dof omitting it
3. Improve docs
Instead of tracking offers in GdkWaylandSelection objects, track the
pending offer in the GdkWaylandSeat and pass it to the GdkDragContext
once we get an enter event.
A container node inside another container node doesn't make a lot of
sense, we can instead just use the parent container node and add the
child container node's children to it directly.
The header linux/input.h used by GDK is specific to Linux. It is
possible to get a few Linux headers on FreeBSD by installing v4l_compat,
but it is usually better to use the one shipped with FreeBSD.
We prefer dev/evdev/input.h to linux/input.h here, so it will always use
dev/evdev/input.h on FreeBSD regardless of v4l_compat.
https://svnweb.freebsd.org/changeset/ports/465644
Broke up a long line, added an empty one, indented another one, and re-aligned
a large amount of function parameter names that got misaligned in some past
refactoring.
https://bugzilla.gnome.org/show_bug.cgi?id=791939
This commit adds support the stable version of the xdg-shell protocol.
Support for the last version of the unstable series is left intact, but
will not receive new features.
The stable version is prioritized above the older version.
https://bugzilla.gnome.org/show_bug.cgi?id=791939
When pressing e.g. a window manager shortcut, which acquires keyboard grab,
Xorg would send FocusOut NotifyGrab then FocusIn NotifyUngrab. Currently
gdk would then deactivate the current surface, which makes accessibility
screen readers think that we have switched to a non-accessible application
and came back again, and thus reannounce the application frame etc. which we
don't want when e.g. just raising volume.
And actually, receiving FocusOut NotifyGrab does not mean losing the
X focus, it only means an application aqcuired a grab, i.e. it is
temporarily stealing keyboard events. On Wayland, this isn't even
notified actually.
This commit makes gdk only deactivate surfaces when there was an actual
focus switch to another window, as determined by has_focus_window (instead
of just has_focus), which happens either normally through FocusOut with
NotifyNormal, or during grabs through FocusOut with NotifyWhileGrabbed.
Fixes#85
gdk_win32_keymap_check_compose() shouldn't be called for
non-W32 displays (i.e. when using broadway or other backends
that could be made to run on Windows).
We no longer set the widget on construction, but instead require an
explicit call to gtk_widget_add_controller().
This way, the reference handling becomes explicit and bindable.
Because gtk_widget_add_controller() is (transfer: full), we don't
even need to unref the controller after adding it.
And we don't need to keep track of it, because controllers get cleaned
up by GtkWidget.
This is the first step towards refactoring how widgets deal with event
controllers.
In the future, the widget will treat controllers the same way it treats
child widgets:
1. The controllers will be created without a widget.
2. There will be gtk_widget_add/remove_controller() functions to add
or remove controllers.
3. The widget will hold a reference to all its controllers.
This way we will ultimately be able to automate controllers with ui
files.
gtk_get_current_event() returns a new reference to the event, it should
be freed across various return branches to avoid the event leak, or we
just fetch the little stuff we're interested in.
Now that all Cairo contexts are ported to managing cairo surfaces
themselves, the old fallback code that didi the managing is no longer
needed.
Also clarify the behavior of gdk_cairo_context_cairo_create() wrt the
vfunc by doing the early exit and the clipping outside of it.
* Remove DC refcounting (we trust GDK to always do
begin_frame/end_frame calls in pairs)
* Now that there's no GDK-provided double-buffer up the stack,
double-buffering is implemented here
(though it's disabled by default - in my tests it didn't provide
any visual improvements, but did decrease performance).
* For some reason delaying window resizes until the point where
we need to blit the double-buffer into the window leads
to visual glitches, so doulbe-buffered windows are resized
in begin_frame, same as non-double-buffered ones.
* New code to clear the paint region, for all drawing modes.
Hopefully, it isn't duplicated anywhere up the stack.
* GL has its own context now, so remove any GL-related comments.
* Layered windows are still used (because cairo actually works
better with them)
* A bit more code re-use for layered windows
* Some functions that were local to gdksurface-win32.c are made
usable for the whole backend
* Drag-indicator drawing is temporarily commented out to match
a similar change in X11 backend
We used to pass 2 regions to GdkDrawCotnext.end_frame() but code was
confusing what they meant. So we now don't do that anymore and only pass
the region that matters: The frame region.
This makes the previous gdk_draw_context_is_drawing() function public
under a new name.
I decided against the old name because we use the term "frame" for a
drawing operation, so I wanted to have this boolean flag reuse the term.
As they require a draw context and the draw context is already bound to
the surface, it makes much more sense and reduces abiguity by moving
these APIs to the draw context.
As a side effect, we simplify GdkSurface APIs to a point where
GdkSurface now does not concern itself with drawing anymore at all,
apart from being the object that creates draw contexts.
Previously, we got the damage, then computed the changed area, then
started a frame with that changed area.
But starting a frame computes the damage for us.
So now we start a frame, then get the damage area from that, then
compute the change area.
Also, split it into its own file - which was the original reason for
looking at this code, the rewrite was an unintentional side effect.
This changes the context to create surfaces on demand.
So whenever the compositor holds onto a surface while GDK wants to
render, it just creates a new surface. If the compositor releases
surfaces, we will retain one for the next frame to be rendered, but free
all extra ones.
This way, we should get to a stage where we have exactly as many
surfaces as needed and never allocate/free any.
Also, don't implement SurfaceClass.ref_cairo_surface() anymore. This
means calls to it will crash now. But as they only happen in the generic
GdkCairoContext implementation, we shouldn't be affected by that.
Plus, once all backends have been ported, that call is going away
anyway.
And make the GdkCairoContext as abstract.
The idea of this and thje following commits is to get rid of all
Cairo code in gdksurface.c (and $backend/gdksurface-$backend.c)
by moving that code into the Cairo context files.
In particular, the GdkSurfaceClass.begin_frame/end_frame()
functions (which are currently exclusively used by the Cairo code
should end up being moved to GdkDrawContextClass.begin/end_frame().
This has multiple benefits:
1. It unifies code between the different drawing contexts.
GL lives in GLContext, Vulkan in VulkanContext and Cairo in
CairoContext. In turn, this makes it way easier to reason about
what's going on in surface-specific code. Currently pretty much
all backends do things wrong when they want to sync to drawing
or to the frame clock.
2. It makes the API of GdkSurface smaller. No drawing code (apart
from creating the contexts) needs to remain.
3. It confines Cairo to the Drawcontext, thereby making it way
more obvious when backends are still using it in situations
where it may now conflict with OpenGL (like when doing the dnd
failed animation or in the APIs that I'm removing in this
branch).
4. We have 2 very different types of Cairo contexts: The X/win32
model, where we have a natively supported Cairo backend but do
double buffering ourselves and use similar surfaces and the
Wayland/Broadway model where we use image surfaces without any
Cairo backend support and have to submit the buffers manually.
By not sharing code between those 2 versions, we can make the
actual code way smaller. We also get around the need to create
1x1 image surfaces in the Wayland backend where we pretend
there's a native Cairo surface.
This does nothing but disallow passing NULL to gdk_surface_begin_paint()
and instead require this context.
The ultimate goal is to split out Cairo drawing into its own source file
so it doesn't clutter up the generic rendering path.
1. Remove set_icons property
2. Make it a GtkWidget subclass
3. Add gtk_fishbowl_set_creation_func()
4. Make the widgetbowl use the new benchmarking infrastructure of the
fishbowl
And of course, gsk_render_node_get_name() is gone, too.
The replacement is of course debug nodes.
As a side effect, GskRenderNode is now *really* immutable.
The shortcuts inhibitors hash table is created when we create a
GdkWaylandWindow implementation for a GdkWindow, and it's destroyed once
we finalize the instance. The fake "root" window we create for the
Wayland display does not have a backing native window, so the shortcuts
inhibitors hash table is set to NULL; this causes a critical error
message when calling g_hash_table_destroy() on it. The finalization of
the root window happens when we close a display connection.
We should use g_clear_pointer(), instead, as it's NULL safe.
Without this change, the displayclose test fails, as all warnings are
considered fatal.
Instead of hard-coding linker flags for a specific operating system and
a specific compiler, we can should cc.links to test them, so they can be
used on more operating systems and compilers.
All of the four platform-dependent backends are enabled by default. It
is usually a good default because it requires users to explicitly choose
backends they want to use. Rules in meson.build also automatically
disable unavailable backends for macOS, Windows, Linux, so users on
these 3 major platforms don't have to manually disable things when
running meson commands.
However, meson.build doesn't do the same thing for other Unix-like
systems, which is acceptable but not ideal. To make it easier to build
GTK+ on these systems, the Linux case, which enables X11 and Wayland and
disables Win32 and Quartz, is made the default for all operating systems
that are not Windows or macOS.
This commit also changes most 'host_machine.system()' calls to os_*
variables, which are easier to read and less likely to be used wrongly.
The shortcuts inhibitors hash table is created when we create a
GdkWaylandWindow implementation for a GdkWindow, and it's destroyed once
we finalize the instance. The fake "root" window we create for the
Wayland display does not have a backing native window, so the shortcuts
inhibitors hash table is set to NULL; this causes a critical error
message when calling g_hash_table_destroy() on it. The finalization of
the root window happens when we close a display connection.
We should use g_clear_pointer(), instead, as it's NULL safe.
Without this change, the displayclose test fails, as all warnings are
considered fatal.
.linked assumes the container is a GtkBox, which is documented as never
flipping children in RTL, so :first-child is always the left child, etc.
GtkBox does that by reordering its CSS nodes when the direction changes.
But most widgets don’t do that, so :first|last-child are 1st/last ADDED
and swap sides in RTL. GtkPathBar is so, and ignoring that in our themes
meant that in RTL, its left/right buttons got each other’s borders. Yuk!
This patch adds the groundwork for supporting widgets like that, via the
%linked_flippable placeholder, and applies that to override buttons in
filechooser .path-bar.linked > button
so that the correct borders get applied to those buttons when using RTL.
Note that I select only PathBars within a FileChooser because we also
have NautilusPathBar, which also uses widget.path-bar – but *does* flip
its nodes for RTL already, so letting that get affected broke it again!
https://bugzilla.gnome.org/show_bug.cgi?id=772817
See the previous commit.
There may be other cases of these being swapped by Gadget conversions,
but hopefully someone else will find and fix those before I have to…
Close https://gitlab.gnome.org/GNOME/gtk/issues/200
Otherwise, if the Popover is destroyed before the MenuButton, the latter
still had a non-NULL but invalid instance and tried to use it in dispose
Close https://gitlab.gnome.org/GNOME/gtk/issues/199
Use g_signal_connect_data() instead of g_signal_connect_object()
to make sure the callback gets disconnected when the data object
is destroyed. This avoids problems in garbage-collected bindings.
https://bugzilla.gnome.org/show_bug.cgi?id=789215
Instead of hard-coding linker flags for a specific operating system and
a specific compiler, we can should cc.links to test them, so they can be
used on more operating systems and compilers.
All of the four platform-dependent backends are enabled by default. It
is usually a good default because it requires users to explicitly choose
backends they want to use. Rules in meson.build also automatically
disable unavailable backends for macOS, Windows, Linux, so users on
these 3 major platforms don't have to manually disable things when
running meson commands.
However, meson.build doesn't do the same thing for other Unix-like
systems, which is acceptable but not ideal. To make it easier to build
GTK+ on these systems, the Linux case, which enables X11 and Wayland and
disables Win32 and Quartz, is made the default for all operating systems
that are not Windows or macOS.
This commit also changes most 'host_machine.system()' calls to os_*
variables, which are easier to read and less likely to be used wrongly.
Well, they don't require a redraw of the widget, because the widget
itself didn't change.
They require a redraw of the parent, because that now displays the
widget in a different position.
And this means we can keep the cache of the widget's render node.
My fishbowl numbers are through the roof^W water surface. Vulkan gets
4000 now.
Due to the few type checks in gtk_widget_get_display(), it was the
slowest part of a call to gtk_widget_query_size_for_orientation if the
in case of a cache hit.
We pulled out the bounds calculation for performance reasons, but the
caller can't know how to properly compute them. Inside gtk+, we can do
that but it's not good enough for public API.
The GVariant we are getting here might not be coming
from GTK+, but rather from some other source. Best to
be forgiving and deal with missing data without crashing.
This was causing the GTK+ portal backends to crash on
print requests from Qt.
It might happen otherwise that a change is recorded in between the
widget dispose and finalization, causing a crash when setting
the visible name for the GtkStack (as that will be NULL at that point)
This will be used in subsequent commits to fix the sign by which the
value is changed in response to directional scroll or keypress events.
The idea is: you have a movement to make – in the form of a delta that
follows widget directions, i.e. −1 means left or up, +1 means right or
down – and you want to know whether that delta needs to be inverted in
order to produce the intuitively expected directional change of :value.
The existing should_invert() is not sufficient: it just determines
whether to invert visually, but we need more nuance than that for input.
To answer that – while not doubling up the work for scrolls and keys – I
add a helper should_invert_move(), which considers other relevant state:
• A parallel movement on priv->orientation should just use the existing
should_invert(), which already worked OK for this case (not others).
• Movements on the other orientation now depend on priv->orientation:
◦ For a horizontal Range, always invert, so up (i.e. −ve in terms of
widget coords) always means increase value & vice-versa. This was
done in get_wheel_delta(), but move it here for use with keys too.
◦ For a vertical Range, ignore :invert as it’s only relevant to the
parallel orientation. Do not care about text direction here either
as RTL locales do not invert number lines, Cartesian plots, etc.
This returns TRUE if the delta should be inverted before applying to the
value, and we can now use this function in both scroll and key handlers.
https://bugzilla.gnome.org/show_bug.cgi?id=407242https://bugzilla.gnome.org/show_bug.cgi?id=791802
If widgets want to clip things, they now need to do it themselves.
By not taking care of clip, we avoid the need to track clip. And by not
tracking clip, we can avoid all unnecessary cache invalidations that we
were doing for render nodes whenever the clip changed.
And when you are scrolling, the clip changes *a lot*.
priv->button is a guint, but we assigned it to a local gint.
gtk/gtkmenushell.c:734:37: warning: comparison between signed and
unsigned integer expressions [-Wsign-compare]
if (button && (new_button != button) && priv->parent_menu_shell)
^
...from CellRenderer::start-editing, to point people in the direction of
info about the lifecycle of the Editable and how to do generic setup.
https://gitlab.gnome.org/GNOME/gtk/issues/154
Drop the line copied from .activate(), replace it with a description of
what this method actually does, and explain what a NULL result means.
https://gitlab.gnome.org/GNOME/gtk/issues/154
* Note in the intro that we're really thinking about temporary widgets
* Mention a gotcha regarding GtkEntry and how ::focus-out stops editing
* Give some examples of what you'd want to do in ::editing-done
* Be a bit more precise about what ::remove-widget represents
* Summarise the lifecycle between Renderer/Editable in .start_editing()
* Emphasise again there that this should be viewed as a temporary widget
https://gitlab.gnome.org/GNOME/gtk/issues/154
When filing a new merge request it's often hard to know who to ask for a
review; using the Git log doesn't always help — the person that touched
a file last may just be fixing the build or a compiler warning.
The `CODE-OWNERS` file format is something that GitHub uses in order to
pre-fill the list of reviewers:
https://help.github.com/articles/about-codeowners/
Ideally, in the future, we'll be able to use this file with a bot like
homu to automatically go through newly filed merge requests and
automatically ask the relevant people for reviews, instead of doing this
manually.
Keep in sync with the current tree.
The changes are mostly caused by updates in the internal hierarchy of
composite widgets, and the fact that the order in which the widget tree
is traversed is not exactly stable.
Instead of having a single massive test running through the a11y
directory, we can split off each individual file into its own unit.
Having individual units has several advantages:
- units are executed in parallel
- it's easier to identify the failing units
- logs for failed units are easier to read
We can't freeze the frame clock on commit, but only after-paint,
otherwise the frameclock will resume in the paint stage.
So freeze the frame clock at the end of the frame if we are waiting for
a frame callback.
Note; The diff is only lage because of indentation changes due to
avoiding early returns in favor of a branch.
GTK does use libintl directly (in gtkmain.c, for example) and thus
needs to be linked to it (if found and/or needed).
Previously we most likely were getting libintl from glib, but
that stopped for some reason. Either way, explicit linking is
the right thing to do here.
Instead of going through an ancillary script to strip away the
`WL_EXPORT` annotation from the generated code, we should bump up the
required version of Wayland, and use the `private-code` argument for
wayland-scanner, which does the right thing for us.
There where some problems (??) with ccache not detecting changes during meson
checks. Setting CCACHE_DISABLE during the meson execution makes ccache not use
the cache and pass things directly to the compiler.
Instead of connecting to / disconnecting from the frame clock, do it
inside the vfuncs next to changing the priv->realized boolean.
This removes a race between those 2 cases that could cause child
widgets' unrealize handlers to reconnect this widget to the frame clock
because it was still marked as realize when the widget had already
disconnected from the frame clock.
Fixes#168
But in turn, also allow it to work on widgets with their own surface.
This way, we can chain up from everywhere and won't have to export
gtk_widget_set_realized().
When using strncpy() with a buffer we need to account for the
terminating NUL character. GCC 8 started warning when using PPD_MAX_NAME
as the buffer length for strncpy() because the buffer we're copying into
has the same length — which means that the terminating NUL may be
skipped if the source string has a length of PPD_MAX_NAME.
The appropriate way to handle the case where we're copying a source with
a length bigger than of PPD_MAX_NAME is, as reported in the strncpy()
documentation, to copy `PPD_MAX_NAME - 1` bytes, and explicitly NUL
terminate the destination buffer. This has the additional benefit of
avoiding the compiler warning.
* Previous commit had misleading info. The code was
added to begin_paint() instead of end_paint(). Though
that did not affect its performance in any visible way.
* Company advised to move the code to an "after_paint" signal
handler, so that it works on all renderers, not just Cairo.
This change caused high fluctuation in FPS values in fishbowl
when it is put in a situation where it cannot achieve 60fps
(such as using Cairo renderer at ultra-high resolution).
This seems to be deliberate and not a bug.
There is no easily apparent way of being notified when frame updates
happene exactly, so we just query frame info at the end of each paint.
If we query too often (faster than DWM refresh rate), we just get
the same values twice in a row, but that is, hopefully, highly unlikely.
Now that queue_draw() isn't restricted to clip anymore, we don't need to
care about clip in the CSS engine either.
We do keep GTK_CSS_AFFECTS_CLIP around though because GtkWindow does
care for the window's size.
When asked for a nonexistent (positive) monitor number,
gdk_x11_display_get_monitor would (at best) return an uninitialized pointer,
instead of returning NULL.
That way, we can store the right region there: The actual painted area
instead of the exposed area (which is way too small).
Also, the GL context is the only user of this data, so storing it there
seems way smarter.
Even widgets with an empty allocation may still want to draw stuff.
Examples include shadows or child widgets with negative margins.
Fixes GtkEntry's progressbar not showing up anymore.
Deferring a bit further making those a standalone controller, make
binding activation happen on run_controllers(), so it happens by
default on widgets (unless the key event was consumed earlier)
without the need of a legacy event controller.
Non gesture controllers have no means to collaborate with other
controllers, thus should be considered standalone entities. It makes
no sense to propagate any further if scroll/key controllers handled
the event.
Use a key controller instead. Note that this currently
breaks the handling of Enter, since we activate key bindings
twice, causing us to switch to the label and back.
This is a GtkGesture done to deal with stylus events from drawing tablets.
Those have a special number of characteristics that extend a regular
pointer, so it makes sense to wrap that.
This may result on the later emission of crossing events, with one of the
sides being already unmapped/unrealized. The widget being unmapped will
result on repick and emission of a set of crossing events anyway.
This event controller is meant to replace usage from key-press/release-event
handlers all through. Optionally it can be set a GtkIMContext, so interaction
is carried by the controller.
1. Pass clip rectangles to gtk_snapshot_push_state() that point into
the state array.
2. g_array_set_size(len+1) the state array
3. Make that function realloc() the state array.
4. The clip rectangle now points into invalid memory
5. Use the clip array
This patch fixes things by moving step 5 to before step 2.
1b9aa1b708 ('a11y: drop the focus tracker') removed a bit too much. We
still have to emit window:activate/deactivate events. They are easy to
emit anyway.
Fixes#127
Overlays are drawings that get rendered on top of the inspected window.
The only overlay in existence so far is the highlight overlay, which is
used to highlight widgets and replaces the "draw" signal handler used
previously.
Instead of just notifying the inspector of what is going to be rendered,
allow the inspector to modify it.
This way, the inspector can overlay information it deems relevant over
the render node while still having access to what the actual widget
(without the inspector) would paint.
If you want to draw a widget to cairo today, you create a widget
paintable, snapshot it to a render node and then draw the render node to
cairo.
And yes, this is that complicated on purpose. Don't draw widgets to
Cairo.
This allows being more specific about the size.
It's useful in particular when the resulting render nodes might be
too small for the size, not only when they are too large. For the
latter case, using a clip node would be enough.
It also requires adding a clip node when rendering the resulting
paintable, but that should be optimized out by GtkSnapshot when not
necessary.
This is actually not just a mechnaism to protect against too many
signals, but it's also a method to getting those signals at the wrong
time.
For every size/content change, a widget needs to invalidate twice:
Once when it queues a resize/redraw (going valid => invalid) and once
when the new size/content is actually assigned (going invalid => valid).
However, one of those invalidations might be inconvenient for the
listener. GtkImage for example does not like receiving
invalidate-contents signals when new contents are assigned, but is fine
with them when the old ones go invalid. And it will not try to draw the
paintable in between anyway.
So by bypassing the 2nd emission if nothing was changed, we can make
GtkImage happy.
This error message is printed into the journal if a GTK app can't
connect to eithre Wayland or X11. Make it at least mention who is not
capable of connecting to a server.
Unrelated, we might want to improve our error reporting when a GTK app
can't start, so that debugging issues with system startup / login get
easier to resolve.
When the clip changes that is passed to a snapshot function, we need to
create eventual cached render nodes because they might not have drawn
their whole area before.
Fixes issues with redrawing when scrolling.
When the max cost for finding a path gets to high, the diff can now be
aborted.
Because render nodes have a fallback method (by just marking the whole
bounds of the nodes as different), we use this to improve performance
of diffs.
This brings fishbowl (which is basically a container node with N images
that change every frame) back to close to previous performance.
Now that we have the full render nodes available, there is not much
benefit in fine-grained control over multiple rectangles.
In particular, it's causing pain with complex regions.
There might be a benefit in clipping to the region's rectangles in cases
like widget-factory where the whole diff is made up of the 2 rectangles
of spinner and the pulsing progress bar, but it needs a good heuristic
for where this is useful.
Instead of calling gdk_surface_invalidate_region(), just
gdk_surface_queue_expose() and rely on the renderer computing the diff
from the previous rendering.
... and diff the previous node with the current one to determine the
clip region.
This doubles the work necessary to track clip regions, but the following
commits will clean that up.
It doesn't need to be exported anymore.
As a side effect, the inspector no longer has any information about the
render region, so remove the code that was taking care of that.
Now that we don't clip the created render nodes anymore, we don't have
to compute the clip region beforehand.
So snapshot the render nodes before initializing the renderer.
This includes a copy of the diff(1) algorithm used by git diff by Davide
Libenzi.
It's used for the common case ofcontainer nodes having only very few
changes for the few nodes of child widgets that changed (like a button
lighting up when hilighted or a spinning spinner).
This requires a bunch of refactorings:
1. Don't pass the current clip region to gtk_widget_snapshot()
so we don't create full widget contents
3. Have a widget->priv->draw_needed that we invalidate on every
queue_draw() call and set on every snapshot()
2. In queue_draw(), walk the widget chain to invalidate the
render nodes of all parents
... and gsk_render_node_can_diff(). Those are vfuncs to compute a region
containing all the pixels that differ between the two nodes.
This is just the plumbing that chains into node classes. No node
implements it yet.
delete_range_cb is set to be called before the text suppression done by
the gtktextlayout (otherwise it does not work properly). But at that
point the cursor position is not yet up to date. We thus need to move
the accessibility cursor notification to after the actual text
suppression, by using another callback.
This fixes cursor position in brltty screen reading.
(cherry picked from commit fa6994d033)
The second parameter of the text-changed::delete event is to be the length,
not the end position. This fixes spurious text removals in brltty
screen reading.
(cherry picked from commit 209f908a03)
Like other widgets, this returns a floating reference, so
(transfer full) is wrong. Just omit the annotation as others do,
thus implying (transfer none).
Close https://gitlab.gnome.org/GNOME/gtk/issues/156
This is meant as an input to the font chooser.
We don't want the user to select a language, but
rather have fonts presented as they would work for
the current language. Therefore, do away with the
lang/script combo on the tweak page.
Really exclude the portions in the gtkfontchooserwidget.c that are built
when HarfBuzz and PangoFT2 are built, and update the Meson files to
exclude such sources as well from the main GTK SO/DLL and from the
gtk4-demo program.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
Remove g_auto*() usage from these sources and use the traditional
g_free(), as g_auto*() are GCCisms (or CLangisms).
Also, don't include unistd.h unconditionally and stop including
langinfo.h and dirent.h, since they seem to be unused.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
The demos are now built as GUI programs, which will require the presence
of WinMain() on Visual Studio builds, unless we specify the entry point.
Pass the /entry:mainCRTStartup linker flag on Visual Studio builds for
the demo programs so that they can link properly.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
For some font features, we can figure out affected
glyphs, and show before/after. For some others, we
hardcode typical sequences.
Still to do: figure out how to find ligatures and
show them.
We now properly use GdkFrameTimings and can therefor check for dropping
even a single frame in the history (of 16 frames or 1/4 of a second).
Once that happens, we immediately stop adding new items to the bowl.
A side effect is that the number of icons is now a *lot* smaller.
This ensures that the frame clock gets updated with correct presentation
times even if nothing was drawn.
This is necessary for benchmarking but would also be relevant for videos
that want to sync to the frame clock but draw frames a lot less.
This commit ensures that each GdkSurface impl remembers the
cursor that GDK sets for it, and that this cursor is set
each time WM_SETCURSOR is called for that sufrace's HWND.
This is needed because W32, unlike X, has no per-window cursors -
the cursor on W32 is a global resource, and we need to keep track
of which cursor should be set when pointer is over which surface
ourselves (WM_SETCURSOR exists exactly for this reason).
This commit also makes GDK remember the surface that has an implicit
grab (since implicit grabs are gone from the upper levels of the toolkit),
and ensures that crossing events are correctly synthesized and the grab
is broken when surface focus changes. This fixes a bug where opening
a new window (by clicking something in some other, pre-existing window)
will make that new window not get any mouse input due to the fact
that the mouse-button-down event from that click caused an implicit
grab on the pre-existing window, and that grab was not released afterward.
Instead of now-unused GdkWin32Cursor class (a subclass of GdkCursor),
add a stand-alone GdkWin32HCursor class that is a wrapper around
HCURSOR handle.
On creation it's given a display instance, a HCURSOR handle and a boolean
that indicates whether the HCURSOR handle can or cannot be destroyed
(this depends on how the handle was obtained).
That information is stored in a hash table inside the GdkWin32Display
singleton, each entry of that table has reference count.
When the GdkWin32HCursor object is finalized, it reduces the reference
count on the table entry in the GdkWin32Display. When it's created,
it either adds such an entry or refs an existing one.
This way two pieces of code (or the same piece of code called
multiple times) that independently obtain the same HCURSOR from the OS
will get to different GdkWin32HCursor instances, but GdkWin32Display
will know that both use the same handle.
Once the reference count reaches 0 on the table entry, it is freed
and the handle (if destroyable) is put on the destruction list,
and an idle destruction function is queued.
If the same handle is once again registered for use before the
idle destructior is invoked (this happens, for example, when
an old cursor is destroyed and then replaced with a new one),
the handle gets removed from the destruction list.
The destructor just calls DestroyCursor() on each handle, calling
SetCursor(NULL) before doing that when the handle is in use.
This ensures that SetCursor(NULL) (which will cause cursor to disappear,
which is bad by itself, and which will also cause flickering if the
cursor is set to a non-NULL again shortly afterward)
is almost never called, unless GTK messes up and keeps using a cursor
beyond its lifetime.
This scheme also ensures that non-destructable cursors are not destroyed.
It's also possible to call _gdk_win32_display_hcursor_ref()
and _gdk_win32_display_hcursor_unref() manually instead of creating
GdkWin32HCursor objects, but that is not recommended.
If GtkExpander:sensitive was FALSE, the arrow still got the normal fg
colour, which made it look clickable, in contrast to the adjacent label.
Fix this by adding selectors to catch the applicable :disabled states.
Note: Needing these may indicate an oops in generic styles elsewhere,
but I couldn’t see any, so let’s just get it looking right for now.
Close https://gitlab.gnome.org/GNOME/gtk/issues/146
Add a new W32 backend-specific message filtering mechanism.
Works roughly the same way old event filtering did, but without
events (events are GDK/X11 concept that never really made sense
on W32), so there's no functionality for 'altering' events being
emitted. If an event needs to be emitted in response to a message
do it yourself.
Implemented like this, it should give better performance than
if we were to use GLib signals for this, since W32 sends a LOT
of messages (unlike X11, which doesn't send events as often)
all the time, and invoking the signal machinery on *each* message
would probably be bad.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
Rename GdkWin32Selection to GdkWin32Clipdrop, since GdkSelection
is mostly gone, and the word "selection" does not reflect the
functionality of this object too well.
Clipboard is now handled by a separate thread, most of the code for
it now lives in gdkclipdrop-win32.c, gdkclipboard-win32.c just uses
clipdrop as a backend.
The DnD source part is also put into a thread.
The DnD target part does not spin the main loop, it just
emits a GDK event and returns a default value if it doesn't get a reply
by the time the event is processed.
Both clipboard and DnD use a new GOutputStream subclass to get data
from GTK and put it into a HGLOBAL.
GdkWin32DragContext is split into GdkWin32DragContext and GdkWin32DropContext,
anticipating a similar change that slated to happen to GdkDragContext.
OLE2 DnD protocol is now used by default, set GDK_WIN32_OLE2_DND envvar to 0
to make GDK use the old LOCAL and DROPFILES protocols.
https://bugzilla.gnome.org/show_bug.cgi?id=773299
GtkWidget*w=gtk_label_new("pLorem ipsum dolor sit amet, consetetur sadipscing elitr, sed diam nonumy eirmod tempor invidunt ut labore et dolore magna aliquyam erat, sed diam voluptua.");
<property name="text">Grumpy wizards make toxic brew for the evil Queen and Jack. A quick movement of the enemy will jeopardize six gunboats. The job of waxing linoleum frequently peeves chintzy kids. My girl wove six dozen plaid jackets before she quit. Twelve ziggurats quickly jumped a finch box.
<object class="GtkStackPage">
<property name="name">entry</property>
<property name="child">
<object class="GtkEntry" id="entry">
<property name="text">Grumpy wizards make toxic brew for the evil Queen and Jack. A quick movement of the enemy will jeopardize six gunboats. The job of waxing linoleum frequently peeves chintzy kids. My girl wove six dozen plaid jackets before she quit. Twelve ziggurats quickly jumped a finch box.
Разъяренный чтец эгоистично бьёт пятью жердями шустрого фехтовальщика. Наш банк вчера же выплатил Ф.Я. Эйхгольду комиссию за ценные вещи. Эх, чужак, общий съём цен шляп (юфть) – вдрызг! В чащах юга жил бы цитрус? Да, но фальшивый экземпляр!
/* Create the store and fill it with the contents of '/' */
parent=g_strdup("/");
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.