Break out a setter that manages the weak pointer, and
use it in finalize. This also fixes a bug where we were
forgetting to disconnect the right signal handler in
some cases.
The other Radio* widgets have this convenience method that removes the
memory management of the opaque GSList used to handle the group from the
API usable from language bindings (especially the ones not based on
introspection).
This commit adds gtk_radio_menu_item_join_group().
https://bugzilla.gnome.org/show_bug.cgi?id=671362
Some arguments, like the group and the label of a RadioMenuItem, can be
NULL: the RadioMenuItem has all the code to deal with them. The argument
validation is too strict, though, for instance doing:
return_if_fail (IS_RADIO_MENU_ITEM (foo))
if (foo != NULL)
set_foo (foo)
Which is obviously incorrect.
This commit also modifies the annotations of the API, to ensure that
language bindings do the right thing.
https://bugzilla.gnome.org/show_bug.cgi?id=671362
If we create a RadioMenuItem without a group, and then set a group, the
menu item will still be set as active, which means an inconsistently
drawn radio menu item - as the RadioMenuItem will set the active flag on
itself, but then it won't reset it when it gets a new group.
https://bugzilla.gnome.org/show_bug.cgi?id=671362
It turned out that using mwm hints to instruct wms to
create border-only decorations is not really working
universally. So, instead of doing this, render a solid
frame without shadow on the client-side to handle this
case.
https://bugzilla.gnome.org/show_bug.cgi?id=746222
Signed-off-by: Olivier Fourdan <fourdan@xfce.org>
If the widget has been destroyed since a DBus message had been sent,
we could be in a condition that the widget pointer exists but it does
not have a window.
This bails as if the widget didn't exist if there is no available
GdkWindow.
We also set the extents to 0 to be defensive since this is a vfunc
implementation.
https://bugzilla.gnome.org/show_bug.cgi?id=746586
Since these part really are the same in all of the x or y direction
and we don't blur in that direction we can just blur one line and
repeat it during drawing.
https://bugzilla.gnome.org/show_bug.cgi?id=746468
There is no need to e.g. blur in the x-direction for the top part
of a box shadow. Also, there is no need to extend the mask in the
non-blurred direction.
https://bugzilla.gnome.org/show_bug.cgi?id=746468
For radius 1px the current implementation rounds down to a 1 px box
filter which is a no-op. Rather than creating useless shadow masks
in this case we bail out blurring early.
Another alternative would be to make radius 1px round up to a 2 px box
filter, but that would change the rendering of Adwaita which is probably
not a great idea this late in the cycle.
https://bugzilla.gnome.org/show_bug.cgi?id=746468
Sometimes path nodes can survive longer than the style context that
created them. Don't crash in those cases.
Fixes startup of mutter.
Testcase included.
https://bugzilla.gnome.org/show_bug.cgi?id=746407
This allows monitoring the CSS tree. For now, moving a child to a
different position relative to its siblings while keeping the same
parent will cause a child-added + child-removed emission.
We need to properly track if a node needs to propagate invalidation
state information to its children. We didn't do this properly before and
that could lead to us forgetting to invalidate nodes in corner cases.
Do not propagate the TIMESTAMP change through the node tree, as that
causes lots of uneeded markings of nodes as invalid.
Instead, walk the node tree and find the nodes that have a non-static
style and only invalidate timestamps on those.
Only invalidate timestamps if the node is marked as invalid. We overload
the meaning of "invalid" as "tracks timestamps".
While I don't like the way this is written, it is an important
optimization because 95+% of nodes don't animate so timestamps don't
matter to them. But timestamps are invalidated 60x per second.
We don't return a NULL style to mean "no changes" anymore, instead
we check new_style == old_style to mean that.
Make sure the code reflects this, otherwise we'll send
GTK_CSS_CHANGE_PARENT_STYLE invalidations everywhere and screw up
performance.
Now that the widget node recomputes styles on update_style() we can just
call it during validate(). That way, we don't need the widget node to
manually compute its style.
If CSS values are queried from a widget, recompute them if necessary. Do
not emit style-updated until the validation phase however.
This way, we don't run into performance traps when style-update causes
invalidations that cause new style-updated to be emitted.
... and pass it to the API that computes new styles.
A special timestamp of 0 means "please don't animate" and is used when
no frame clock is available for a node.
This is mainly an attempt to merge the update_style() and validte()
vfuncs. Code is not there yet, but that's the idea.
Also, gtk_css_node_set_style() should not be public. And this gets
closer to that goal, too.
After measuring it, I realized the optimization never triggers for
Adwaita and rarely ever triggers for simple themes. So it is not
useful to keep it around.
This allows adding more API for it.
It also includes code that tracks modifications and invalidates siblings
and their positions whenever nodes get added or removed.
If we know the parent's get_path_for_child() implementation is safe to
be used with GtkCssNode because it doesn't do anything special, we do
that. Unfortunately that requires whitelisting vfuncs because the vfunc
is public API so anyone can override it.
Instead, use gtk_widget_get_path() which makes GtkWidget cache the path.
This is a temporary solution until we can get rid of widget paths.
This increases memory usage quite noticably.
The node declaration has the same functionality as
gtk_css_node_declaration_add_to_widget_path(). So instead of using that
function on a path, you can use the original path and the declaration in
a matcher.
So far the vfunc is kinda quirky (the path argument is an out argument
for something you have to free when you're done with the matcher), but
I'm about to change that.
We don't want to add the current classes to the widget path - which
might potentially be different after a gtk_style_context_save() - but
the root node's ones. So what better thing to do than actually using the
root node?
Instead of directly requesting the layout phase, register a tick
callback. This is what the docs suggest for animations and it's what we
need for the next commit.
... to a bunch of functions.
This requires a tiny change to the heuristics for the style cache - we
now cache styles when they have the same style provider as their parent
instead of when they have the default provider - but that change doesn't
have any effect in practice.
The parent refs the child, so gtk_css_node_set_parent() adds/removes a
reference.
We should probably refactor this so that we name the function
parent.add(node) instead of node.set_parent(parent) - makes the
refcounting more clear.
Instead of passing the style declaration, take the node's style
declaration. Add arguments to allow overriding the state for the one
case where this didn't work.
The code is confusing the stylecontext=>node transition, so I'll remove
it.
WIth the recent introduction of the parent style cachen, I'm not sure
it's worth it at all. Some crude benchmarking suggests a slight speedup
when removing the cache.
So no guarantees about adding it back later.
- GtkCssWidgetNode
for style contexts owned by a widget
- GtkCssPathNode
for style contexts using a GtkWidgetPath
- GtkCssTransientNode
for nodes created with gtk_style_context_save()/restore()
The functionality of it is supposed to grow, so better put it in a
custom file early.
This is just a naive split so far, the next patches will split things
further.
This commit adds the GtkTextTag:underline-rgba and :strikethrough-rgba
properties and the necessary plumbing to apply these colors in GtkTextLayout.
With this change, you can alter the color of underlines including those
of type PANGO_UNDERLINE_ERROR.
You might want to alter the underline color to differentiate between
spelling and grammer mistakes. In code editors, it is convenient to
differentiate between errors and warnings.
Note that the GtkTextAppearance struct is public ABI and has no spare
room for new fields, so we are resorting to some tricky packing to store
the colors in the unused pixel field of the fg_color and bg_color structs.
This packing is accomplished by the macros in gtktextattributesprivate.h.
Signed-off-by: Christian Hergert <christian@hergert.me>
https://bugzilla.gnome.org/show_bug.cgi?id=402168
It has been a long-standing request to interpret scroll events
with Shift held down as horizontal instead of vertical, and
some applications are already doing this on their own.
https://bugzilla.gnome.org/show_bug.cgi?id=132197
This causes pixman warnings, or worse.
Clearly, something is not quite right here, if we end up
redrawing tabs at a time when the allocation is set to (1, 1).
For now, avoid straining the error handling in the lower layers.
https://bugzilla.gnome.org/show_bug.cgi?id=746301
On popover_unmap(), perform the gtk_widget_unmap() call last, so the
GtkWindowPopover data is ensured to be alive throughout the function
if the popover widget is destroyed right on ::unmap.
https://bugzilla.gnome.org/show_bug.cgi?id=745829
This ensures the widget data stays live long enough to perform invariants
check after emission if the widget happens to be destroyed on a callback.
https://bugzilla.gnome.org/show_bug.cgi?id=745829
`bundle install` command on fedora 21 was failing
with following message:
Your Gemfile has no gem server sources. If you need
gems that are not already on your machine, add a line
like this to your Gemfile:
source 'https://rubygems.org'
After adding that line it worked again.
https://bugzilla.gnome.org/show_bug.cgi?id=746205
With the 3.0 transition, this code went from just querying the entry's
height request to doing a full size request.
Then it got code to revert the features that a full size request does.
And then it grew code that manually computed the baseline.
Avoid this and just do what happened back in the days: Do a regular
height request.
This changes the semantics of the get_frame_size() vfunc wrt its
behavior towards subclasses overwriting the get_height() vfuncs, but I'm
happy to live with that.
Applications forget to set the follow-state property and then some
themes used a style where normal symbolic icons have the same color as
prelit or selected backgrounds and that suddenly made icons invisible
and who wants that?
https://bugzilla.gnome.org/show_bug.cgi?id=746201
Rework the way we assign an accessible name to menu buttons,
to make sure we pick up a label, should the button contain
one, and only override the name with "Menu" as a fallback.
This reverts commit 63f59dde3a.
It turns out, the state was not just necessary for style computation,
but also for tracking RTL and LTR. And so it broke the reftests.
gtk_tree_selection_real_modify_range() has a g_return_if_fail() if the
start or end paths passed to it do not correspond to real tree nodes.
However, GtkTreePaths inherently do not have to be valid, so it should
be acceptable to call gtk_tree_selection_select_range() with
non-existent paths. Replace the g_return_if_fail() by a silent return,
and add a unit test.
https://bugzilla.gnome.org/show_bug.cgi?id=712760
We need to filter out platform-specific sources from the files that we
use with the introspection scanner.
In this specific case I could have moved the gtkclipboard-quartz.c and
the gtkdnd-quartz.c files out of the $(gtk_base_c_sources) variable, but
doing a filter-out on the variable itself is more resilient in case we
eventually add files and we forget about the result.
Now that this is split in two separate gestures, both must be reset
when the WM grabs the pointer. Also, do on resize drags like on move
drags, and claim the gesture before resetting, so the ownership is
properly transferred across any other widgets.
https://bugzilla.gnome.org/show_bug.cgi?id=745969
gtk_entry_set_completion() (and _gtk_entry_completion_connect underneath)
may happen when the entry has no toplevel yet. This window acts as the
"parent" for subsurface windows on wayland, and is indispensable to create
its backing wl_subsurface, so this resulted in non-visible popups there.
Fix this by ensuring transient_for is set before showing the popup.
https://bugzilla.gnome.org/show_bug.cgi?id=745463
This reverts commit fb9a6bb6d8.
In a recent test, I've found that Xfce and Mate now support
this, so they will not be affected by this requirement. And
adding the check back will solve the 'client-side shadow'
problem in KDE.
The page_setup of a GtkPrintContext or GtkPrintUnixDialog is nullable,
so all reference count changes to it have to be guarded against NULL
values.
Found by scan-build.
https://bugzilla.gnome.org/show_bug.cgi?id=712760
_gtk_rbtree_first() can potentially return NULL if the RB tree is empty,
which would result in NULL pointer dereferences in the GtkTreeSelection
code. Gracefully handle them.
Found by scan-build.
https://bugzilla.gnome.org/show_bug.cgi?id=712760
These vfuncs cannot be NULL: implementations are provided by GtkEntry,
and subclasses should not set them to NULL. Instead of conditionalising
the calls to the vfuncs, assert that they’re set and call them
unconditionally.
This prevents the possibility of a subclass setting the vfunc to NULL
and then a gtk_entry_get_*_size() call returning undefined values in its
out variables.
https://bugzilla.gnome.org/show_bug.cgi?id=712760
They are a bit terse, and they should point to the preferred way of
overriding the theme, i.e. using a GtkStyleProvider for your
application, and custom style classes.
When the scrollbars are overlayed, the size requisition of the
scrolled window should not depend on whether the scrollbars are
visible or not. This was not quite the case, because we forgot
one case where scrollbar size was still added to the requisition.
Postpone until the last moment whether the target widget still
potentially uses updates from this sequence, or window dragging
actually applies because all gestures on the target went to denied
state.
This fixes window dragging on empty space in a headerbar that is
contained in a paned (as in e.g. gedit).
https://bugzilla.gnome.org/show_bug.cgi?id=745562
These types are not covered by g_test_register_all_types,
and having the types registered when a print dialog is used
helps in cases where GTK+ is dlopened.
https://bugzilla.gnome.org/show_bug.cgi?id=745065
GtkPlacesSidebar applies a sorting function on
the tree model that does not consider the case
of bookmarks, which are sorted by their indexes.
By adding the bookmarks corner case and comparing
then by their indexes, GtkPlacesSidebar can sort
the bookmarks properly in the order they're saved.
https://bugzilla.gnome.org/show_bug.cgi?id=744589
This signal can be emitted by GtkSearchEntry after search has been
cancelled, and other operation mode is set. It doesn't make sense to
populate the search model in that state anymore, so just avoid doing it.
https://bugzilla.gnome.org/show_bug.cgi?id=745479
The object might be destroyed when mid operation, causing crashes as the
query callback still expects the object pointer to be valid. Also, remove
the gdk_threads_enter/leave pairs, the callback will be executed on the
caller (UI) thread, so this is not necessary.
https://bugzilla.gnome.org/show_bug.cgi?id=745479
The search engine might stay alive longer due to extra temporary refs, so
the signal handlers should be removed for the filechooser to ignore these
properly.
https://bugzilla.gnome.org/show_bug.cgi?id=745479
This file now uses round(), which was not available until C99, so include
fallback-c89.c instead of math.h, which includes math.h and does a fallback
implementation of round().
Avoid scrolling animation when scrolling to where we want
to insert the new folder, since adding the entry doesn't
do the right thing if the treeview is still scrolling.
An alternative would be to wait for the animated scrolling
to reach its target before starting the editing, but this
is easier.
https://bugzilla.gnome.org/show_bug.cgi?id=729366
This is not specified specifically by the attached controllers, so let
the scrolledwindow set the mask, as motion events with no buttons pressed
are interesting to it.
https://bugzilla.gnome.org/show_bug.cgi?id=745344
The gesture is hooked to the capture phase, so it works for buttons in
header bars and whatnot. In order to be friendly to the widget it is
capturing events from, an ugly hack is in place to avoid capturing
events when the target widget has a gesture that would consume motion
events.
This happens on button release, which is more convenient if the gesture
can be consumed by something else (eg. window dragging), and already behaves
correctly wrt cancelled gestures, broken grabs, etc.
This also allows us to unify pointer and keyboard behavior, popping up the
menu widget in a single place.
