This will be the future way to connect signals automatically (and be
used for other things, too).
For now, gtk_builder_connect_signals_default() is ported to use it.
This is pretty unused and gets in the way of the next steps.
A potential side effect is that for templates the widget was passed as
the user data argument. If that turns out to be important, we have to
special case that situation.
It's old and busted, and mostly broken in weird ways when it comes to
extended input devices. All that XIM does, these days, is make a mess
when people enable it by mistake.
The "iconified" state is mostly an X11-ism; every other platform calls
this state "minimized" because it may not involve turning a window into
an icon at all.
The new check/radio style increased their specificity, but it was not
synchronized in the menu styling.
This commit increases the specificity to match the updated check/radio
styling.
https://gitlab.gnome.org/GNOME/gtk/issues/2096
- Adjust the emoji picker appearance to be the same as gtk3's.
- Fix button.emoji-section margins in RTL.
- Prevent the .emoji hover effect from being applied to the entire
nested popover.
The signal handler that calls notify_child_cb() is disconnected
from a child widget before the child is removed from the header bar.
gtk_header_bar_dispose() chains up before destroying the start and end
boxes, thus avoiding calls to notify_child_cb() after start_box and
end_box have been cleared.
Fixes#2246
A GtkBox in a titlebar could have unwanted spacing.
This caused a glitch in split-header applications where
parts of the titlebar would be transparent or black.
This commit tweaks Adwaita to make sure no spacing is added for boxes when used as titlebars.
Fixes#2233
Signal names must start with a letter. They can’t start with a hyphen.
It was originally added in commit 74f4c6f6bd with no given reason
for why the signal name was chosen as it was.
Signed-off-by: Philip Withnall <withnall@endlessm.com>
Some of the CSS API has been moved to a public namespace, so we need to
include it into the introspection data we build in order for people to
use it.
Fixes: #2230
This adds support using the GtkTextHistory helper for undo/redo to the
GtkText widget. It is similar in use to GtkTextView, but with a simplified
interface.
You can disable undo support using the GtkText:enable-undo property. By
default, it is enabled.
This changes the semantics of ::deleted-text to perform the removal of
text in the default handler. This means, that if you want the old behavior
where your signal handler is called after the text has been deleted, you
should connect with G_CONNECT_AFTER in your signal handler (or use
g_signal_connect_after).
Without this change, there was never a way to get the deleted-text before
the operation had completed and this is necessary to provide undo support
to the text widgets.
This builds upon the GtkTextHistory helper to provide undo and redo support
for the GtkTextView widget and GtkTextBuffer object.
You can undo/redo using familiar shortcuts such as Primary+Z,
Primary+Shift+Z, ad Primary+Y.
Developers that wish to disable undo, should set the
GtkTextBuffer:enable-undo property to FALSE.
You can wrap irreversible actions
gtk_text_buffer_begin_irreversible_action() and
gtk_text_buffer_end_irreversible_action(). This will cause the undo stack
to drop all undo/redo actions and the changes made between them will be
the "initial state" of the buffer.
Calling gtk_text_buffer_set_text() will do this automatically for you.
The GtkTextHistory helper provides the fundamental undo/redo stack that
can be integrated with other text widgets. It allows coalescing related
actions to reduce both the number of undo actions to the user as well as
the memory overhead.
A new istring helper is used by GtkTextHistory to allow for "inline
strings" that gracefully grow to using allocations with g_realloc(). This
ensure that most undo operations require no additional allocations other
than the struct for the action itself.
A queue of undoable and redoable actions are maintained and the link for
the queue is embedded in the undo action union. This allows again, for
reducing the number of allocations involved for undo operations.
set_active() is meant to set the state of the property, which happens
during setup.
The clicked signal is emitting events from user actions.
It is impossible to use Togglebuttons for MVC applications as long as it
does that.
Unrealize the GSK renderer before destroying children.
This makes the renderer drop any texture caches that
it might have, so that we don't needlessly download
them when releasing the widget-side holder objects.
As a fortunate side effect, this fixes crashes on
exit with GtkGLArea-containing windows under Wayland.
Add the margins to the menubutton and not the button inside the
menubutton. This way popovers properly point to the inner button again
instead to somewhere 6px off.
GtkIconInfo contains too much redundant data,
and it is hard to know which fields are to be
used when.
This commit starts to move towards dropping the
icon_file field - we prefer to work with the
filename directly, since it lets us avoid mime
sniffing and strdups.
This function is not useful. Every icon theme
on my system either does not have that field,
or has it as 'folder'. So, just use 'folder'
when you need an example icon.
When we are loading themed icons, we know if
we deal with an svg or png file, so it is
entirely unnecessarily to have gdk-pixbuf
use g_content_type guess to rediscover that
information.
Change the pixbuf utils apis we have to allow
passing format information down to where we
can use it when creating the pixbuf loader.
... inside the default vfunc.
Instead, walk the children until we find the first widget that can be
focused. If no child can be focused, return FALSE from grab_focus.
The expander icon is renamed from "arrow" to "expander".
The expander widget itself is renamed from "expander" to
"expander-widget" (Better ideas welcome).
This makes it possible to have an "expander" icon in more places then
the GtkExpander widget (in particular in tree lists) and not
confuse it with arrows.
Previously, with selection bounds, we could have a state where a line
display with selections set would eroniously stay in the cache. This
aggresively releases those at the boundary cases fixing deselection
drawing state.
We can avoid recreating a number of text nodes from render_para() on
sub-sequent runs if we cache the rendernode instead of just the
PangoLayout.
When used with GtkSourceMap, this can yield a ~7 FPS improvement during
smooth scrolling at the cost of some more memory.
This creates a new GtkTextViewChild that can manage overlay children at
given x,y offsets in buffer coordinates. This simplifies GtkTextView by
extracting this from GtkTextWindow as well as providing a real widget for
the borders.
With this change, we also rename gtk_text_view_add_child_in_window() to
gtk_text_view_add_overlay(). For those that were using
GTK_TEXT_WINDOW_WIDGET, they can use a GtkOverlay. It does not appear
that anyone was using GTK_TEXT_WINDOW_(LEFT|RIGHT|TOP|BOTTOM) for widgets
in this fashion, but that can be done by setting a gutter widget with
gtk_text_view_set_gutter(). We can make GtkTextViewChild public if
necessary to simplify this should it become necessary.
GtkTextViewChild will setup a CSS node of either "text" or "border"
depending on the GtkTextWindowType.
The old GtkTextViewChild has been renamed to AnchoredChild as it is only
used for widgets with anchors in the GtkTextBuffer. This also removes the
use of allocated GSList and instead embeds a GQueue and GList to save a
few extraneous allocations.
The renaming of this function doesn't make much since because the window
is the GtkTextWindowType, not GdkWindow specifically. So we can keep the
old name which is closer to the proper meaning and less code for consumers
to change when porting to 4.x.
It’s possible for code which uses a `GtkListBox` to reuse a single
header row, and move it around between rows. For example, this might
happen if the code has interactive widgets (like buttons) in the row,
and doesn’t want to continually recreate them and reattach signals to
them whenever the row headers change.
Unfortunately, this was broken, as the old header widget was
unconditionally unparented, even if it had just been set as the header
for a different row in the same `GtkListBox`. This left it assigned as
a child widget in the `GtkListBox` (so it was iterated over by
`forall`), but without its parent widget set.
Fix that by only unparenting the header if it hasn’t already been
assigned as the parent of a different row.
Signed-off-by: Philip Withnall <withnall@endlessm.com>
G_ENABLE_DEBUG is tied to the meson builttype property, so building with "plain"
results in G_ENABLE_DEBUG not being defined and the GTK_DEBUG env var just gets ignored
for that build.
Since it can be confusing that GTK_DEBUG has no effect print a warning message instead.
See #2020. This is a port of !1109 to master
While the ::scroll signal always returns whether it handled the event,
the others do not, for example ::decelerate.
Previously, this caused the event to stop at a scroll controller with
CAPTURE phase, never emitting the ::decelerate signal on later
controllers with BUBBLE phase.
Fixes#2151
This currenly just wraps GMarkupParser, but the plan is to expose this
instead of GMarkup in the GtkBuildable interfaces, allowing us to
replace the parser with something that handles pre-parsed input
instead.
Note that we duplicate some of the features of GMarkup to implement
the APIs rather then call down to GMarkup, as we need to support these
in the pre-parsed case anyway.
We were converting the same colors over and over again to a string, just
to free them again at the end of the function. We know the colors at
compile time however, so don't convert them at all.
load_symbolic_svg was loading the pixbuf just to get its size via
gdk_pixbuf_get_{width,height}. However, this function is called in a
loop in gtk_make_symbolic_pixbuf_from_data.
So, do this only once and pass the icon size along to load_symbolic_svg.
Add a cache of icon infos that we keep around a little longer, to avoid
loading icons from disk that only exist for a short amount of time (e.g.
during one frame of a cell renderer snapshot).
We make sure recently used items are kept alive by just adding them to
the cache on lookup.
We were allocating IconThemeDir instances and then only later assign a
value to has_icons. In the !has_icons case, we were directly throwing
the IconThemeDir away again.
Delay allocating the IconThemeDirs until we know that it has icons.
This avoids allocating and then de-allocating around 1400 IconThemeDir
instances when opening the widget-factory.
The previous code was hiding the fact that the scan_resources function
almost always did nothing and just used g_resources_enumerate_children()
and then returned FALSE, leaving the caller with cleaning up the already
allocated IconThemeDir. By inlining this, we make sure that calling code
does not even need to allocate the IconThemeDir.
It doesn't really make sense to save the applications and groups in
recent infos as linked lists. We get them from glib as arrays, so we can
as well just save them as such.
Creating a recent manager can be fairly expensive and we won't use it if
the widget is not visible or the recent mode has not been entered. Code
other places can already handle a NULL recent manager, so just create it
when entering the recent mode. And shove 25ms of startup time off the
widget-factory this way.
Delaying this by one frame by putting it in an idle just makes the code
more complex for no gain. The actual slow part is reading the
recently-used.xbel, which happens when creating the recent manager.
