We need to clear the cache manually on full revalidates because
_gtk_css_change_for_child() will clear the full revalidation flags.
And then gtk_style_context_update_cache() will not do the right thing
(which is to clear itself).
Because we can switch from animating to non-animating pretty much
anywhere, do the check for animations unconditionally instead of trying
to cram it into the correct if path (and failing).
We now cache the results of lookups on the parent GtkCssStyle. This
allows sharing styles between widgets (recursively). However, this
only works if the styles can't potentially depend on siblings -
neither directly via sibling selectors or via position pseudo-classes
like :first-child.
Unfortunately, Adwaita currently uses first-child a lot, and in
particular for labels, which are the most common widgets.
The big benefits of this change are both less CPU - due to not needing
to compute styles again - and less memory usage - due to sharing of
the styles between widgets. Here's some nonscientific numbers I
collected while pondering the usefulness of this patch:
glade glade widget
demo demo factory
runtime styles styles
Adwaita before 19.1s 5800 1150
Adwaita now 18.9s 3800 970
Adwaita hacked now 14.5s 3100 910
simple before 11.3s 5800 1150
simple now 10.8s 1300 590
Adwaita: Adwaita as provided by GTK
Adwaita hacked: Adwaita with the first/last-child for GtkLabel removed
simple: A 250 lines simple GTK theme I use for testing
before: This patch not applied
now: this patch applied
glade demo runtime: Starting glade opening a large file and closing it
glade demo styles: GtkCssStaticStyle objects after opening glade with
the large file as per inspector
widget factory styles: GtkCssStaticStyle objects after startup as per
inspector
Previously we looked at the save/restore state when determining the
parent. This is wrong in the case where we're updating the cache though.
So we now save the parent in the style info.
After the parent changes in commit
3a337156d1 we need to refresh the cached
styles after the current style. After all, they now depend on the base
style.
After 3a337156d1 style lookups still used
the parent context's style as the parent style, even though after a
gtk_style_context_save() the root style of the style context is the
proper parent.
Testcase attached.
Instead of keeping an animated style everywhere, we replace it with the
static style when nothing gets animated.
Apart from making the code cleaner, this gets rid of a bunch of animated
style values that do nothing but wrap a static style.
The only style that is animated is the style of the unsaved primary
node. So there's no need to create animated style objects for the other
ones.
There is a bunch of ugliness in the code currently. Further commits are
expected to fix them.
For now, this is only an implementation detail of the animated style.
The idea is to use GtkCssStaticStyle as the result of CSS queries and
then put a GtkCssAnimatedStyle on top that manages the animations. The
neat thing about this is that you can cache the static values.
GtkCssStyle is the base class to be used for all types of styles that do
exist.
GtkCssAnimatedStyle is the only implementation so far, that is exactly a
copy/paste of the old GtkCssStyle code.
I removed it in 14f5ce7108 because I
thought it was unnecessary, but it wasn't. When we build a tree like
this:
.matches ─┬─ .doesntmatch
└─ .alsodoesntmatch
We would get the changes for the .matches part returned. This is however
only right if that node of the tree contains results. If results only
exist with the child nodes (all of which don't match), then this part
should not match either.
GtkCalendar can have an invalid date — mostly at initialization. This
means that GDateTime construction may fail. We need to handle that case
gracefully, like the old code did.
This fixes the `notify` test suite, which started failing with:
/Notification/GtkCalendar:
GLib-CRITICAL **: g_date_time_get_day_of_week: assertion 'datetime != NULL' failed
inside the Continuous builder.
This is so we can detect crossing events on the scrolledwindow widget,
which can be useful for toggling certain behaviors or not depending on
whether the mouse is on the widget.
If the first of the month was falling on a Sunday, we would not
render any days of the previous month, and instead show two weeks
of the next month at the bottom. Improve this by showing one week
of each.
https://bugzilla.gnome.org/show_bug.cgi?id=301835
Changing font size does not cause widgets to update their size
properly anymore, since we forgot an invalidation flag. The
problem can be observed e.g. by running gtk3-demo and calling
gsettings set org.gnome.desktop.interface font-name "Cantarell 22"
On the wayland backend, set up GDK_WINDOW_SUBSURFACE windows
for popovers. In the popover code, the popover-relative-to-parent
calculation had to be tweaked, and it's been made to always prefer
the given popover position, since there's no sizing limitations.
https://bugzilla.gnome.org/show_bug.cgi?id=738891
We were asserting priv->filename != NULL in various places,
which leads to apps refusing to work when HOME is set to a
nonexisting or nonwritable value. Since it isn't hard, just
make GtkRecentManager survive without a filename. We won't
save or read any recently used files in this state, but
thats ok.
https://bugzilla.gnome.org/show_bug.cgi?id=739038
Provide a mechanism for hiding the "Quit", "About" and "Preferences"
menu items from the normal places in a traditional menubar layout (in
the File and Edit menus) when the menu is being rendered in the Mac OS
menubar.
These items can already be found in the application menu.
With this feature, applications can now define a single menu to use in
all 'traditional' scenarios.
Use this new attribute in Bloatpad.
https://bugzilla.gnome.org/show_bug.cgi?id=741610
In addition to the fixed-size mode that is used by the text view,
add a mode in which the magnifier requests enough size to render
the entire inspected widget at the current magnification. In this
mode, the magnifier will update its size when the size of the
inspected widget changes. Also, make the magnifier redraw on its
own whenever the inspected widget draws.
There is no good reason to assign the value directly.
Also, this fixes d23f3254b7
where widgets that chained up instead of calling
gtk_widget_set_allocation() would not draw becaues of empty clip.
(1) Get rid of supports_clip flag. All widgets (implicitly) support
clip.
(2) Don't reset the clip to { 0, 0, 0, 0 } before the "size-allocate"
signal.
(3) Make gtk_widget_set_allocation() set the clip (to the allocation).
This ensures that eveyr widget has a clip set.
Note: It overrides previous calls to gtk_widget_set_clip(), while in
3.14 this didn't happen.
(4) As the clip is set by gtk_widget_set_allocation() now, don't set
it after the "size-allocate" signal anymore.
This fixes calls to gtk_widget_queue_draw() from inside the
size_allocate vfunc.
Override the gtk-shell-shows-app-menu and gtk-shell-shows-menubar
settings to FALSE, if we can detect that we are not on the same
session bus as the xsettings provider that we got these settings
from.
We determine this by comparing the bus ID of 'our' session
bus with the one reported in the Gtk/SessionBusId xsetting.
If they are different, then it very likely that we are in an ssh
situation where we see the forwarded X display, but not the
session bus. The D-Bus based menu exporting will not work
in this situation.
https://bugzilla.gnome.org/show_bug.cgi?id=671802
The magnifier is now set enough height to show the line being currently
manipulated, which makes it just big enough to show the layout height at
that size and magnification.
- It is not possible anymore to trigger text DnD through touch, pressing
and dragging from within the selection will instead extend it. Text
shrinking is still available through the handles
- The selection mode for touch is per-word, char-level manipulation is
still available through the handles.
- Tapping within the selection will bring in text handles, and toggle
text selection popover.
This mode could also shrink the selection, plus the starting point would
seem somewhat arbitrary (actually dependent on the dragging direction of
the last selection).
Made this mode more consistent by only allowing it to extend the selection,
only in one direction for each operation, and so it keeps the current
selection as a minimum.
Instead of passing a GdkEvent and let the function figure out whether the
selection should be extended, let that to the caller and just pass a
boolean here.
Using the parent widget context is a leftover of the pre-popover
implementation, which used GdkWindows directly. This will make the context
reflect widget state, at the expense of changing the selector paths
that used to match the handles.
Checking the return value was valid for most gestures, but
GtkGestureLongPress, where the first press triggers internally an action,
but does nothing for the sequence to be claimed/denied, FALSE was eventually
returned, and the button/sequence functions would be incorrect when
::pressed is emitted.
So check that the sequence is being handled by the gesture, this is more
desirable than the return value as it's independent of sequence state,
and still will be FALSE for the cases we want to catch here.
When a RTL paragraph is not set to wrap, the right margin is not respected
because of the margins counted twice so we replace
display->width by PIXEL_BOUND (extents.width), the same quantity
without the margins.
https://bugzilla.gnome.org/show_bug.cgi?id=741702
GtkIconTheme doesn't scale icons beyond the size specified in the theme
anymore. This can result in arbitrarily large menu items when a theme
only provides large icons.
Force icons to always be 16x16 to ensure that menu items always have the
same height.
https://bugzilla.gnome.org/show_bug.cgi?id=741259
This drops the maybe_allocate_buffers that re-allocates buffers
at any point. Instead we just set have_buffers to FALSE and have
the buffers re-created when needed.
This also makes the buffer creation code imdeponent and makes it
clean up no longer needed buffers in order to handle being called
multiple times due to the above.
We also ensure we re-allocate the buffers when we're resizing
and the buffers are already created.
On creation, we call action_removed() in case the action was missing
from the start. Because we just created the action, 'can_activate' will
always be FALSE here and this function will therefore always do nothing.
We do want the visibility state to be updated though, for the case where
the action is missing but the item should still be visible from the
start.
Update the visibility directly instead of trying to call
action_removed().
https://bugzilla.gnome.org/show_bug.cgi?id=735122
We were only properly setting the "is-visible" flag to TRUE for menu
items with associated actions and not (for example) on submenus.
This was fine because the code for building GtkMenus from models
(correctly) assumed that submenus should always be visible and never
checked the property.
This is not true for the Mac OS code, which actually checked the
property and found it to be false for submenus.
Initialise the property to TRUE so that we get the correct value
reported for items that don't have actions.
https://bugzilla.gnome.org/show_bug.cgi?id=735122
All popover sides have extra margins to possibly hold the tail, this is
accounted for in gtk_popover_get_rect_coords(), and should be accounted
for too in the tail position calculation.
This fixes the gtk_render_frame_gap() warnings seen when a popover is
pushed far too close to window sides.
After b49c7c3421 we were no longer doing a
full revalidate after GTK_CSS_CHANGE_SOURCE changes.
This fixes spurious failures of widgets not properly updating when
changing the theme.
These functions, while added for use by the GTK inspector, are generally
useful to applications that need to resolve what action groups are
available to a particular GtkWidget.
https://bugzilla.gnome.org/show_bug.cgi?id=741386
Without this, the focus will not be inside the top_stack when
we switch to the statistics page and it shows the excuse. This
in turn will make the next page not take focus as expected when
switching away from the statistics page again.
Add notebook-like focus handling: Keep track of the last focused
descendent of each page, and focus it again when switching back
to the page. If there is no last focused child, we move the focus
into the page as if the user had hit Tab.
When using a completion with some custom cells in the cell layout,
if would often size wrong when first presented on screen.
The entry completion is the only place in the entire gtk code base
that calls gtk_tree_view_column_cell_get_size outside of gtktreeview
itself. It calls into the function before the tree view has done some
important validation on its cell state, the net result of which is
only the first element in the gtkcellareabox the entry completion uses
well actually have its size respected.
We now call gtk_widget_get_preferred_size on the tree view before
calling into the individual cell size routines, to guarantee that the
tree view has run its validate_rows routine and cell state is valid.
https://bugzilla.gnome.org/show_bug.cgi?id=741130
Use the new scrollable API for getting non-scrollable borders
and draw over/undershoot at the right place. In practice, this
means that they now appear below treeview headers.
Add a vfunc to return a non-scrollable border around scrollables.
This would be nicer as a property, but we can't add properties
to an interface without breaking 3rd party implementations, so
make this an optional vfunc, and handle it not being set.
The new region selector tracks more than just one soimple selector, so
it requires some more advanced specificity tracking.
Technically, this is still not correct, because it will report the same
specificity for
tab:sorted
and
tab:sorted:sorted
(and the second selector will be printed as the first) but this is
regions, so meh.
This is just a way to handle regions more conveniently. What this does
is to change the descendant matcher into a maybe-descendant matcher
whenever the current object we're parsing might be a region. Because "*"
might also refer to a region and not just a new element.
See testsuite/reftests/css-match-region-matches-star.ui for a testcase.
When parsing a selector like
tab:active
We used to create 2 selector objects, one for the region, and one for
the pseudoclass. Now, for convenience, we special-case region
pseudoclass parsing so that the active flag is added to the region
selector.
A side effect is that ":nth-child(1)" no longer works for regions.
Hopefully people didn't use this but used the saner ":first-child"
instead.
