GitLab

g_logv adds one for us already.

Instead of
  /org/gtk/libgtk/theme/$THEME-$VARIANT.css
look at
  /org/gtk/libgtk/theme/$THEME/gtk-$VARIANT.css
and that way mirror the directory layout of real themes.

Currently, GtkCssProvider can emit ::parsing-error only during
the actual parsing, although the documentation hints that it might
happen at other times.

This commit adds a emit_error method to the GtkStyleProviderPrivate
interface that will let us emit errors from the compute() implementations
as well, which can be useful (e.g. if an image fails to load).

https://bugzilla.gnome.org/show_bug.cgi?id=760238

Split the CSS docs off from the GtkCssProvider docs and
give them their own chapter. Among other things, this commit
introduces more or less complete definitions of the syntax for
the supported selectors, a complete list of all supported
properties, and definitions for their values. This includes
documentation for GTK+-specific properties such as -gtk-icon-source.

Remove references to regions and engines, emphasize
element names over type names, update the list of
pseudo-states.

We know that parent is NULL at this point.

Also, fix indentation.

This is not a standard CSS property, so rename it to
-gtk-icon-shadow. We still support the old name, with a
deprecation warning.

Emit a deprecation warning if a property is not used
by its official name.

This helps people fix up their css.

This makes G_PARAM_DEPRECATED useful for style properties.

For @import, we want to emit the error as part of the @import statement,
not as part of the new file.

When loading a nonexisting CSS file using
gtk_css_provider_load_from_file() or gtk_css_provider_load_from_path()
we would emit the error using a NULL scanner. Don't do that, because
we'll have a NULL section in that case and error handlers don't like
that.

Testcase attached.

https://bugzilla.redhat.com/show_bug.cgi?id=1277959

Signal handlers expect a section to be present, so provide them with
one.

New testcase included.

https://bugzilla.gnome.org/show_bug.cgi?id=757240

This is in preparation for not allocating an array at all if there
are no matches.

Nobody is able to look up those hex values.

GtkCssProvider was using the wrong unref function for GtkCssKeyframe
objects. Funnily, this didn't crash at all, the refcount was at the
same place in the struct - it just didn't free all the memory, causing
valgrind to complain.

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.

This is in preaparation for the big GtkCssComputedValues => GtkCssStyle
rename

Instead of walking the selector tree, we walk the selectors.

Look for themes in $prefix/themes/$name/gtk-$version/ (for
version = 3.16, 3.14, ...), before using the old location
$prefix/themes/$name/gtk-3.0/. This gives theme authors a
way to support multiple versions of GTK+ 3 with separate
css files.

Move finding the gtk.css file into its own function. We will
make this more complicated in the near future.

GTK+ relies on this function silently failing for invalid data.

This is common enough that we should just provide an API for it
- we already had a function for this purpose anyway.

https://bugzilla.gnome.org/show_bug.cgi?id=672946

The font-stretch CSS property is defined in the Level 3 CSS Fonts
module, available at:

  http://dev.w3.org/csswg/css-fonts/#propdef-font-stretch

It allows defining a normal, condensed, or expanded face to the font
description. Pango already supports it, so this is literally just the
CSS parser machinery needed to bridge our CSS to the FontDescription
API.

https://bugzilla.gnome.org/show_bug.cgi?id=735593

Don't take a state when constructing the CSS matcher. Instead, rely on
the newly introduced state in the widget path.

This way, the state can be queried not only on the first element, but on
all elements of the widget path.

Add a private function that lets us decide at runtime whether to
keep css section information around - this will be useful in the
GtkInspector.

We want to have the "-gtk-" prefix for our custom CSS properties. But
we also want to parse names starting with a "-" as style properties.
So make sure that "-gtk-" is treated like a normal property and we emit
errors when somebody uses it wrong.

This is to catch errors with people typing
  -gtk-iconsource: none;
instead of the correct
  -gtk-icon-source: none;

commit 7f6a964c replaced entities, but
escaped the replacement text also inside literals, which resulted in the
escaping '\' to also appear in the documentation.

https://bugzilla.gnome.org/show_bug.cgi?id=727322

  expected a valid semicolon
sounds kinda not so great. Make it say
  expected semicolon

Unless somebody can tell me the difference between valid and invalid
semicolons?

Without sgml mode, we can't escape /* as /* anymore,
so just switch to // for comments in examples.

With all element markup gone, it is time to turn off
sgml mode, and get rid of entities as well.