]> Pileus Git - ~andy/gtk/log
~andy/gtk
12 years agocss: Move shadow parse/print/compute funcs
Benjamin Otte [Tue, 27 Mar 2012 05:15:23 +0000 (07:15 +0200)]
css: Move shadow parse/print/compute funcs

They don't belong in the style funcs, as those are for custom
properties, and the shadow type is private.

12 years agostylepropertyimpl: Add parse funcs for enums
Benjamin Otte [Tue, 27 Mar 2012 04:29:37 +0000 (06:29 +0200)]
stylepropertyimpl: Add parse funcs for enums

12 years agostylepropertyimpl: Add color parsing function
Benjamin Otte [Tue, 27 Mar 2012 03:11:58 +0000 (05:11 +0200)]
stylepropertyimpl: Add color parsing function

This is in preparation for removing the specified type and computed type
properties from GtkCssStyleProperty, which is in preparation for really
using GtkCssValue classes and not GTypes.

12 years agocustomproperty: Don't rely on querying style property
Benjamin Otte [Tue, 27 Mar 2012 03:00:35 +0000 (05:00 +0200)]
customproperty: Don't rely on querying style property

This way, we can remove type madness from GtkCssStyleProperty later.

12 years agostyleproperty: Make gtk_style_property_register() not be valist
Benjamin Otte [Tue, 27 Mar 2012 02:58:15 +0000 (04:58 +0200)]
styleproperty: Make gtk_style_property_register() not be valist

Instead, make the caller create a GtkCssValue in advance.

12 years agocssvalue: Add _gtk_css_value_equal()
Benjamin Otte [Mon, 26 Mar 2012 23:43:12 +0000 (01:43 +0200)]
cssvalue: Add _gtk_css_value_equal()

For now, we return FALSE for all default css values, so this is not very
useful.

I also think of this as an optimization equal, not a guaranteed equal,
because we don't even have a notion of what "equal" means.

For example, for background-repeat, "repeat, repeat" and "repeat"
are functionally equivalent. But the cssvalue has no idea that it's used
for background-repeat.
As a more complicated example, "repeat, no-repeat" and "repeat" are
equal to what one sees as long as there's only one image listed
background-image-source. But once you start transition'ing to an image
with 2 sources, it's different...

12 years agotests: Change expected errors to conform with output
Benjamin Otte [Mon, 26 Mar 2012 16:58:23 +0000 (18:58 +0200)]
tests: Change expected errors to conform with output

This change isn't strictly correct, but I can't be bothered until we get
a tokenizer that can really differentiate between allowed values and
invalid syntax.

12 years agocssparser: Use _gtk_css_parser_error_full()
Benjamin Otte [Mon, 26 Mar 2012 16:54:30 +0000 (18:54 +0200)]
cssparser: Use _gtk_css_parser_error_full()

12 years agocssparser: Add _gtk_css_parser_error_full()
Benjamin Otte [Mon, 26 Mar 2012 16:36:20 +0000 (18:36 +0200)]
cssparser: Add _gtk_css_parser_error_full()

12 years agostyleproperty: Make _gtk_style_property_parse_value() return a CssValue
Benjamin Otte [Mon, 26 Mar 2012 15:24:02 +0000 (17:24 +0200)]
styleproperty: Make _gtk_style_property_parse_value() return a CssValue

Also split out initial/inherit handling into a custom GtkCssValue class.

12 years agocss: Remove _gtk_css_style_property_is_specified_type()
Benjamin Otte [Mon, 26 Mar 2012 14:22:53 +0000 (16:22 +0200)]
css: Remove _gtk_css_style_property_is_specified_type()

That check is going to be refactored away.

12 years agocssvalue: Add a CSS value for arrays
Benjamin Otte [Mon, 26 Mar 2012 06:30:24 +0000 (08:30 +0200)]
cssvalue: Add a CSS value for arrays

Make the value auto-cycle its child values, so it can be used for
background properties.

12 years agocssvalue: Remove _gtk_css_value_new_take_gvalue()
Benjamin Otte [Mon, 26 Mar 2012 05:46:18 +0000 (07:46 +0200)]
cssvalue: Remove _gtk_css_value_new_take_gvalue()

Having two constructors from GValues complicates refactorings, so I'd
rather not have them.

12 years agocssvalue: Get rid of unused functions
Benjamin Otte [Mon, 26 Mar 2012 05:26:45 +0000 (07:26 +0200)]
cssvalue: Get rid of unused functions

12 years agostyleproperty: Make _gtk_style_property_query() take a GValue
Benjamin Otte [Mon, 26 Mar 2012 05:08:24 +0000 (07:08 +0200)]
styleproperty: Make _gtk_style_property_query() take a GValue

... and don't make it return a GtkCssValue. We want to use this for
compat with the old GValue APIs after all...

12 years agostyleproperty: Make print_func take a GtkCssValue
Benjamin Otte [Mon, 26 Mar 2012 04:46:29 +0000 (06:46 +0200)]
styleproperty: Make print_func take a GtkCssValue

Also, constify GtkCssValue getters, so we can pass a const GtkCssValue
to the print_func.

12 years agocssvalue: Make the structure vtable-based
Benjamin Otte [Mon, 26 Mar 2012 04:13:57 +0000 (06:13 +0200)]
cssvalue: Make the structure vtable-based

Don't use real classes, just a vtable.

12 years agostyleproperty: Add support for equal_func
Benjamin Otte [Sat, 24 Mar 2012 06:02:59 +0000 (07:02 +0100)]
styleproperty: Add support for equal_func

12 years agostylecontext: Don't clear cache when only state changes
Benjamin Otte [Sat, 24 Mar 2012 01:15:41 +0000 (02:15 +0100)]
stylecontext: Don't clear cache when only state changes

This takes more memory, but changes to backdrop or active state are
quite expensive otherwise.

12 years agocssmatcher: Use quarks for classes
Benjamin Otte [Sat, 24 Mar 2012 00:13:50 +0000 (01:13 +0100)]
cssmatcher: Use quarks for classes

12 years agostyleproperty: Add a function to compare values for equality
Benjamin Otte [Fri, 23 Mar 2012 22:52:48 +0000 (23:52 +0100)]
styleproperty: Add a function to compare values for equality

and default to never compare them as equal.

12 years agocssvalue: Constify a bunch of APIs
Benjamin Otte [Fri, 23 Mar 2012 22:23:30 +0000 (23:23 +0100)]
cssvalue: Constify a bunch of APIs

12 years agocssvalue: Remove unused functions
Benjamin Otte [Fri, 23 Mar 2012 15:19:22 +0000 (16:19 +0100)]
cssvalue: Remove unused functions

12 years agostylecontext: Really queue style changes
Benjamin Otte [Fri, 23 Mar 2012 02:01:16 +0000 (03:01 +0100)]
stylecontext: Really queue style changes

Instead of instantly applying a new style, just mark the context as
invalid. Only apply the new style at layout time.

12 years agostylecontext: Optimize the common case of "style didn't change"
Benjamin Otte [Fri, 23 Mar 2012 00:52:38 +0000 (01:52 +0100)]
stylecontext: Optimize the common case of "style didn't change"

12 years agogtk: Make widget only invalidate widget positions
Benjamin Otte [Wed, 21 Mar 2012 08:37:09 +0000 (09:37 +0100)]
gtk: Make widget only invalidate widget positions

Instead of resetting them completely

12 years agowindow: Set parent context instead of just resetting style
Benjamin Otte [Tue, 20 Mar 2012 17:08:08 +0000 (18:08 +0100)]
window: Set parent context instead of just resetting style

Otherwise the inherit properties won't inherit properly.

12 years agowidget: Replace reset_style() calls
Benjamin Otte [Tue, 20 Mar 2012 17:04:37 +0000 (18:04 +0100)]
widget: Replace reset_style() calls

... with finer grained _gtk_widget_invalidate_style_context() calls.

12 years agowidget: Don't go via path when resetting style
Benjamin Otte [Tue, 20 Mar 2012 15:04:13 +0000 (16:04 +0100)]
widget: Don't go via path when resetting style

Instead, invalidate the path when the style context changes. And
invalidate the style context directly.

12 years agowidget: Add _gtk_widget_invalidate_style_context()
Benjamin Otte [Tue, 20 Mar 2012 14:40:53 +0000 (15:40 +0100)]
widget: Add _gtk_widget_invalidate_style_context()

This is supposed to replace gtk_widget_reset_style().

12 years agowidget: Merge function into only caller
Benjamin Otte [Tue, 20 Mar 2012 13:38:44 +0000 (14:38 +0100)]
widget: Merge function into only caller

12 years agowidget: Clear broken widget path in constructor
Benjamin Otte [Sun, 18 Mar 2012 22:00:25 +0000 (23:00 +0100)]
widget: Clear broken widget path in constructor

... where it belongs instead of fiddling with it in get_path().

12 years agostylecontext: On failure, exit the loop, don't try again
Benjamin Otte [Tue, 20 Mar 2012 02:44:26 +0000 (03:44 +0100)]
stylecontext: On failure, exit the loop, don't try again

Because we will fail again. And then we try again. And then we fail
again. Ad infinitum.

12 years agostylecontext: Invalidate when setting classes/state/regions
Benjamin Otte [Tue, 20 Mar 2012 02:35:36 +0000 (03:35 +0100)]
stylecontext: Invalidate when setting classes/state/regions

... and nobody has called style_context_save().

12 years agostylecontext: Use _gtk_style_context_queue_invalidate() on self
Benjamin Otte [Tue, 20 Mar 2012 02:21:10 +0000 (03:21 +0100)]
stylecontext: Use _gtk_style_context_queue_invalidate() on self

12 years agostylecontext: Introduce _gtk_style_context_queue_invalidate()
Benjamin Otte [Tue, 20 Mar 2012 02:15:26 +0000 (03:15 +0100)]
stylecontext: Introduce _gtk_style_context_queue_invalidate()

This is for only queueing invalidations instead of doing a full
invalidation cycle.

12 years agostylecontext: Keep track of children
Benjamin Otte [Tue, 20 Mar 2012 01:06:00 +0000 (02:06 +0100)]
stylecontext: Keep track of children

12 years agowidget: Don't set widget path on style context
Benjamin Otte [Mon, 19 Mar 2012 00:19:59 +0000 (01:19 +0100)]
widget: Don't set widget path on style context

Instead, make the style context use the widget's path. Saves a bunch of
memory.

12 years agostylecontext: Notify widget directly of changes
Benjamin Otte [Sun, 18 Mar 2012 23:25:31 +0000 (00:25 +0100)]
stylecontext: Notify widget directly of changes

This way, we don't need to g_signal_connect(), which saves a bunch of
memory and performance.

12 years agostylecontext: Add setter for widget
Benjamin Otte [Sun, 18 Mar 2012 22:02:33 +0000 (23:02 +0100)]
stylecontext: Add setter for widget

... and actually set the widget on the style context. Note that this
function does not take a reference on the widget, which is a very good
reason to keep it private.

12 years agomatcher: Add a superset matcher
Benjamin Otte [Sun, 18 Mar 2012 15:13:09 +0000 (16:13 +0100)]
matcher: Add a superset matcher

This matcher always matches only on some relvant things and ignores the
rest. This allows you to match only on name and class, but ignore state
and parents/siblings for example.

12 years agomatcher: Add a matcher that matches anything
Benjamin Otte [Sun, 18 Mar 2012 13:42:50 +0000 (14:42 +0100)]
matcher: Add a matcher that matches anything

Lesson learned: Adwaita uses all possible selector types and all of
those for parents, too. But so far no siblings.

12 years agoreftests: Add a reftest for nth-child
Benjamin Otte [Sun, 18 Mar 2012 01:53:44 +0000 (02:53 +0100)]
reftests: Add a reftest for nth-child

Prime numbers are fun.

12 years agotests: Add a parsing test for nth-child()
Benjamin Otte [Sun, 18 Mar 2012 01:19:41 +0000 (02:19 +0100)]
tests: Add a parsing test for nth-child()

Now that we support the an+b notation, make sure we really do.
Maybe my generating script went a bit overboard with the space
possibilities though...

12 years agoselector: Rewrite position tracking
Benjamin Otte [Sun, 18 Mar 2012 01:11:44 +0000 (02:11 +0100)]
selector: Rewrite position tracking

We now track the position as a (type,a,b) tuple where the numbers make
up the an + b formula from CSS3 nth-child.

Also, the get_sibling() and get_sibling_index() vfuncs were replaced by
a has_position() vfunc. This is mostly so that the matcher can always
return TRUE. And I need that for the everything matcher.

12 years agosyleprovider: Add a vfunc to get the changes
Benjamin Otte [Sat, 17 Mar 2012 20:09:51 +0000 (21:09 +0100)]
syleprovider: Add a vfunc to get the changes

This way we can check what changes are even interesting for our matcher.

12 years agoselector: Rename selector class
Benjamin Otte [Sat, 17 Mar 2012 20:02:43 +0000 (21:02 +0100)]
selector: Rename selector class

It's a 'position' pseudoclass, not a 'region' pseudoclass.

12 years agocss: Add selector change types
Benjamin Otte [Sat, 17 Mar 2012 18:23:53 +0000 (19:23 +0100)]
css: Add selector change types

This allows querying selectors for which changes would change their
"matchingness".

12 years agomatcher: Turn GtkCssMatcher into a union
Benjamin Otte [Sat, 17 Mar 2012 13:32:52 +0000 (14:32 +0100)]
matcher: Turn GtkCssMatcher into a union

That way, we can add more matchers as we need them.

12 years agomatcher: Use a vtable
Benjamin Otte [Sat, 17 Mar 2012 13:28:07 +0000 (14:28 +0100)]
matcher: Use a vtable

Now we can do lots of fancy matchers, yay!

12 years agocss: Make the style provider take a matcher as an input argument
Benjamin Otte [Sat, 17 Mar 2012 01:36:11 +0000 (02:36 +0100)]
css: Make the style provider take a matcher as an input argument

12 years agocss: Add GtkCssMatcher
Benjamin Otte [Sat, 17 Mar 2012 01:17:23 +0000 (02:17 +0100)]
css: Add GtkCssMatcher

This is so we can later do matching with other things than
GtkWidgetPath.
In particular, this is a requirement for getting rid of GtkWidgetPath.

12 years agostyleprovider: Add a custom object for a list of style providers
Benjamin Otte [Fri, 16 Mar 2012 04:14:41 +0000 (05:14 +0100)]
styleprovider: Add a custom object for a list of style providers

This way, we don't have to do magic inside GtkStyleContext, but have a
real API.
As a cute bonus, this object implements GtkStyleProvider itself. So we
can just pretend there's only one provider.

12 years agostylecontext: Stop querying style providers for icon factories
Benjamin Otte [Fri, 16 Mar 2012 02:35:13 +0000 (03:35 +0100)]
stylecontext: Stop querying style providers for icon factories

Nobody implements this feature and I don't wanna maintain it. If it
turns out somebody is using it, we gotta invent something.

12 years agogdk: Add deprecation/availability macros for 3.6
Benjamin Otte [Fri, 16 Mar 2012 00:23:26 +0000 (01:23 +0100)]
gdk: Add deprecation/availability macros for 3.6

12 years ago[l10n] Updated German translation
Mario Blättermann [Mon, 16 Apr 2012 19:42:14 +0000 (21:42 +0200)]
[l10n] Updated German translation

12 years agowayland: Prefer X11 backend over Wayland
Kristian Høgsberg [Tue, 7 Feb 2012 14:08:48 +0000 (09:08 -0500)]
wayland: Prefer X11 backend over Wayland

https://bugzilla.gnome.org/show_bug.cgi?id=674102

12 years agoapplication-window: Move desktop file handling code to separate function
Kalev Lember [Sat, 14 Apr 2012 18:30:26 +0000 (21:30 +0300)]
application-window: Move desktop file handling code to separate function

https://bugzilla.gnome.org/show_bug.cgi?id=674118

12 years agobuild: Add an enable flag to allow enabling Cairo GL backend
Rob Bradford [Mon, 16 Apr 2012 14:29:44 +0000 (15:29 +0100)]
build: Add an enable flag to allow enabling Cairo GL backend

This change adds --enable-wayland-cairo-gl which turns on the define used in
the Wayland backend to determine whether to use EGL surfaces with Cairo GL or
whether to use the Cairo image backend with an SHM surface (the default).

Part of the fix for: https://bugzilla.gnome.org/show_bug.cgi?id=672361

12 years agowayland: Use an ARGB buffer for the Cairo and SHM surface
Rob Bradford [Thu, 12 Apr 2012 11:11:04 +0000 (12:11 +0100)]
wayland: Use an ARGB buffer for the Cairo and SHM surface

12 years agowayland: Force an expose for the whole area when scrolling
Rob Bradford [Thu, 19 Jan 2012 11:51:50 +0000 (11:51 +0000)]
wayland: Force an expose for the whole area when scrolling

12 years agowayland: Add support for rendering into an SHM buffer with Cairo image backend
Rob Bradford [Wed, 4 Apr 2012 16:20:13 +0000 (17:20 +0100)]
wayland: Add support for rendering into an SHM buffer with Cairo image backend

The first version of this change included a bug that meant that if you don't
compile for any other backend then it wouldn't search for cairo. Credit for
identifying the bug goes to darxus@chaosreigns.com.

Fixes: https://bugzilla.gnome.org/show_bug.cgi?id=672361
12 years agowayland: Move the buffer creation to the cairo surface creation
Rob Bradford [Wed, 4 Apr 2012 13:05:57 +0000 (14:05 +0100)]
wayland: Move the buffer creation to the cairo surface creation

12 years agowayland: Move the finalize function to allow a clear logical block of code
Rob Bradford [Wed, 4 Apr 2012 13:05:27 +0000 (14:05 +0100)]
wayland: Move the finalize function to allow a clear logical block of code

12 years agowayland: Remove unused member from struct
Rob Bradford [Wed, 4 Apr 2012 10:40:52 +0000 (11:40 +0100)]
wayland: Remove unused member from struct

12 years agoUpdated Crimean Tatar (Crimean Turkish) translation
Reşat SABIQ [Mon, 16 Apr 2012 06:56:55 +0000 (01:56 -0500)]
Updated Crimean Tatar (Crimean Turkish) translation

12 years agoUpdated Crimean Tatar (Crimean Turkish) translation
Reşat SABIQ [Mon, 16 Apr 2012 06:54:47 +0000 (01:54 -0500)]
Updated Crimean Tatar (Crimean Turkish) translation

12 years agoupdate Simplified Chinese (zh_CN) translation
Yinghua Wang [Mon, 16 Apr 2012 06:14:33 +0000 (14:14 +0800)]
update Simplified Chinese (zh_CN) translation

12 years agoGtkImage: Move g_clear_object() to gtk_image_finalize()
Michael Vogt [Mon, 16 Apr 2012 04:51:43 +0000 (06:51 +0200)]
GtkImage: Move g_clear_object() to gtk_image_finalize()

This moves the freeing of the icon_helper from the destory to the finalize
function to avoid segfaults when trying to access a destroyed object before it
is disposed. This often happens in signal handlers which get called
asynchronously after destroy.

https://bugzilla.gnome.org/show_bug.cgi?id=674050

12 years agoUpdated Norwegian bokmål translation
Kjartan Maraas [Sun, 15 Apr 2012 17:38:20 +0000 (19:38 +0200)]
Updated Norwegian bokmål translation

12 years agoBump version
Matthias Clasen [Sat, 14 Apr 2012 01:26:43 +0000 (21:26 -0400)]
Bump version

12 years ago3.4.1
Matthias Clasen [Sat, 14 Apr 2012 01:08:06 +0000 (21:08 -0400)]
3.4.1

12 years agoapplication-window: try to use the desktop name in the fallback menu
Cosimo Cecchi [Wed, 11 Apr 2012 14:24:04 +0000 (10:24 -0400)]
application-window: try to use the desktop name in the fallback menu

Try to fetch the name from the application desktop file for the
fallback menu if possible, instead of forcing applications to use
g_set_application_name or hardcoding "Application".

https://bugzilla.gnome.org/show_bug.cgi?id=673882

12 years agobuild: split out a HAVE_GIO_UNIX config variable
Cosimo Cecchi [Wed, 11 Apr 2012 14:22:57 +0000 (10:22 -0400)]
build: split out a HAVE_GIO_UNIX config variable

We'll use it in GtkApplicationWindow, since we need to conditionally
depend on gio-unix there.

https://bugzilla.gnome.org/show_bug.cgi?id=673882

12 years agoFix malformed doc comments
Matthias Clasen [Fri, 13 Apr 2012 00:07:28 +0000 (20:07 -0400)]
Fix malformed doc comments

Most of these are forgotten :'s and similar details
which gtk-doc now warns about.

12 years agoconfigure: Use AM_PATH_GLIB_2_0() for finding glib-compile-resources
Kalev Lember [Wed, 11 Apr 2012 20:12:15 +0000 (23:12 +0300)]
configure: Use AM_PATH_GLIB_2_0() for finding glib-compile-resources

No need to manually set GLIB_COMPILE_RESOURCES when AM_PATH_GLIB_2_0()
does this for us.

12 years agoFix sourcedir!=builddir .gir build from source tarball
Pavel Holejsovsky [Fri, 23 Mar 2012 06:09:02 +0000 (07:09 +0100)]
Fix sourcedir!=builddir .gir build from source tarball

Setting -I$(top_builddir) before -I$(top_srcdir) causes that
g-ir-scanner picks up boxing definitions generated by glib-mkenums.

https://bugzilla.gnome.org/show_bug.cgi?id=672133

12 years ago[quartz] Fix manual resizing of windows
Kristian Rietveld [Sun, 19 Feb 2012 15:08:26 +0000 (16:08 +0100)]
[quartz] Fix manual resizing of windows

In the Quartz backend, there are two methods by which windows are
resized. The first method is fully handled by Quartz and does not appear
in the event stream the application resizes. The second method is when
we resize windows by ourselves. In OS X this happens when a GTK+ resize
grip is used. This resize grip is larger than the Quartz resize grip.
When the resize is started outside the "Quartz area", we have to handle
it by ourselves.

This patch fixes this manual window resizing by ignoring events while we
are in the process of resizing (such that the events actually arrive at
the sendEvent handler of GdkQuartzWindow where this resize is handled).
When the resize has finished we break all grabs such that GDK is not
stuck thinking the cursor is still in the resize window.

12 years agoExport break_all_grabs() within Quartz backend
Kristian Rietveld [Sun, 19 Feb 2012 15:02:09 +0000 (16:02 +0100)]
Export break_all_grabs() within Quartz backend

Function was renamed to _gdk_quartz_events_break_all_grabs().

12 years ago[l10n]Updated Catalan (Valencian) translation
Carles Ferrando [Tue, 10 Apr 2012 19:22:30 +0000 (21:22 +0200)]
[l10n]Updated Catalan (Valencian) translation

12 years ago[l10n] Fixes on Catalan translation
Jordi Serratosa [Tue, 10 Apr 2012 19:22:22 +0000 (21:22 +0200)]
[l10n] Fixes on Catalan translation

12 years ago[l10n]Updated Catalan (Valencian) translation
Carles Ferrando [Tue, 10 Apr 2012 19:21:06 +0000 (21:21 +0200)]
[l10n]Updated Catalan (Valencian) translation

12 years ago[l10n] Fixes on Catalan translation
Jordi Serratosa [Tue, 10 Apr 2012 19:20:55 +0000 (21:20 +0200)]
[l10n] Fixes on Catalan translation

12 years agostyleproperty: 3rd hotfix for a 3-line patch
Benjamin Otte [Tue, 10 Apr 2012 14:55:05 +0000 (16:55 +0200)]
styleproperty: 3rd hotfix for a 3-line patch

I really should not write code sometimes...

Copy/paste error this time.

12 years agoRevert part of "textview: Remove extra magic for drawing children"
Benjamin Otte [Tue, 10 Apr 2012 14:14:59 +0000 (16:14 +0200)]
Revert part of "textview: Remove extra magic for drawing children"

This partially reverts commit df37446f9b4116887865b7ecfddeb535fbc0eacb.
The commit changed API that was public (or at least semi-public) and
it's not worth changing that.

The original bug should still be fixed.

https://bugzilla.gnome.org/show_bug.cgi?id=673839

12 years ago styleproperty: Fix thinko
Benjamin Otte [Mon, 9 Apr 2012 18:33:46 +0000 (20:33 +0200)]
 styleproperty: Fix thinko

Introduced in 8243b16ad79cb4965ab5314e53a20ef16b2af9c3 - seems I wasn't
really on top of my own capabilities on Saturday.

12 years agocustomproperty: Fix thinko
Benjamin Otte [Mon, 9 Apr 2012 01:27:18 +0000 (03:27 +0200)]
customproperty: Fix thinko

.. in 43c56d70ea247b3317c386e22407ad20769d2762

12 years agostyleproperty: Save some memory
Benjamin Otte [Sat, 7 Apr 2012 19:32:39 +0000 (21:32 +0200)]
styleproperty: Save some memory

Only create a new CssValue when we actually need one.

12 years agocustomproperty: Ensure style properties are registered
Benjamin Otte [Sat, 7 Apr 2012 15:24:50 +0000 (17:24 +0200)]
customproperty: Ensure style properties are registered

... before reigstering our own. Also, use that capability to emit a
warning.

12 years agocsscustomproperty: Redo RGBA/Color parsing
Benjamin Otte [Mon, 9 Apr 2012 01:05:12 +0000 (03:05 +0200)]
csscustomproperty: Redo RGBA/Color parsing

This fixes custom parse functions returning RGBA values where we
expected symbolic colors.

12 years agotests: Add a uint8 test
Benjamin Otte [Mon, 9 Apr 2012 00:54:17 +0000 (02:54 +0200)]
tests: Add a uint8 test

... to ensure the last test works.

12 years agocssvalue: Re-add support for random value types
Benjamin Otte [Mon, 9 Apr 2012 00:51:07 +0000 (02:51 +0200)]
cssvalue: Re-add support for random value types

This got lost in the CssValue transition, and apparently some people use
this.

12 years agoentrycompletion: set_property() should call property setters
Pavel Holejsovsky [Fri, 23 Mar 2012 06:09:02 +0000 (07:09 +0100)]
entrycompletion: set_property() should call property setters

gtk_entry_completion_set_property() was setting many properties by
directly modifying priv values, bypassing notification invocation and
possibly another actions done by gtk_completion_entry_set_xxx ()
functions.  Fix by invoking set_xxx() instead of setting the property
value directly.

The real bug observerd was that setting text-column property using
g_object_set() caused SIGFPE later when entry completion was about to
appear.  gtk_entry_completion_set_text_column () apparently does way
more important things than just setting priv->text_column member.

https://bugzilla.gnome.org/show_bug.cgi?id=673693

12 years agotextview: Remove extra magic for drawing children
Benjamin Otte [Tue, 3 Apr 2012 16:41:25 +0000 (18:41 +0200)]
textview: Remove extra magic for drawing children

Instead, just draw the children. The cairo code will keep track of
things, so there's no need to track things.
Also, the old code was doing it wrong.

https://bugzilla.gnome.org/show_bug.cgi?id=672544

12 years agocsscomputedvalues: Unref old value when setting new one
Benjamin Otte [Sat, 31 Mar 2012 01:35:48 +0000 (03:35 +0200)]
csscomputedvalues: Unref old value when setting new one

12 years agodocs: Add gtkx.h to the list of headers
Benjamin Otte [Mon, 2 Apr 2012 01:20:06 +0000 (03:20 +0200)]
docs: Add gtkx.h to the list of headers

12 years agotreeview: Don't cache expander size
Benjamin Otte [Thu, 29 Mar 2012 11:41:32 +0000 (13:41 +0200)]
treeview: Don't cache expander size

Style properties should not be cached, they should be queried live.

Also, this fixes the case where the expander size wasn't set when
constructing the widget which caused expanders to go missing.

12 years agowidget: Create style context using regular API
Benjamin Otte [Sun, 18 Mar 2012 13:02:58 +0000 (14:02 +0100)]
widget: Create style context using regular API

12 years agomenu: Undo a hack with style contexts
Benjamin Otte [Tue, 20 Mar 2012 03:06:49 +0000 (04:06 +0100)]
menu: Undo a hack with style contexts

Widget implementations aren't allowed to fiddle with the widget's style
context.

Sheesh.

12 years agotray: Add suggested braces around empty body in 'if' statements
Benjamin Otte [Sat, 7 Apr 2012 10:56:39 +0000 (12:56 +0200)]
tray: Add suggested braces around empty body in 'if' statements