]> Pileus Git - ~andy/gtk/log
~andy/gtk
12 years agoFix nicks for the window class enum
Matthias Clasen [Wed, 14 Dec 2011 02:13:31 +0000 (21:13 -0500)]
Fix nicks for the window class enum

As pointed out in bug 665999, these were just not right.
Before this commit, the nicks were 'output' and 'only'.
After this commit, they are 'input-output' and 'input-only'.

12 years agoFix set_focus_child handling during focus changes
Alexander Larsson [Tue, 13 Dec 2011 23:23:57 +0000 (00:23 +0100)]
Fix set_focus_child handling during focus changes

In many cases we used to set focus_child to NULL all the way up
to the top and then to the right value, even if there was
a common ancestor, meaning these see a temporary NULL value for
focus_child. Only when the new focus widgets direct parent was
in the previous ancestor list did we stop early.

This fixes that by always stopping propagation when reaching
the common ancestor.

12 years agoGtkFrame: Fix a casting in the compute_child_allocation() method
Claudio Saavedra [Tue, 13 Dec 2011 17:25:15 +0000 (19:25 +0200)]
GtkFrame: Fix a casting in the compute_child_allocation() method

The casting used to calculate the child allocation is confusing MAX().
As a result, width and height end up with negative values, which makes no sense.

https://bugzilla.gnome.org/show_bug.cgi?id=666109

12 years agothemingengine: split a GtkThemingBackground object
Cosimo Cecchi [Sat, 10 Dec 2011 00:14:12 +0000 (19:14 -0500)]
themingengine: split a GtkThemingBackground object

Allows easier modification and general sanity of the CSS background
rendering code.

12 years agoimage: take misc padding into account when calculating x/y draw coords
Cosimo Cecchi [Mon, 12 Dec 2011 11:30:27 +0000 (12:30 +0100)]
image: take misc padding into account when calculating x/y draw coords

This fixes the misc-alignment.ui reftest.

12 years agoimage: add gtk_image_get_preferred_size()
Cosimo Cecchi [Mon, 12 Dec 2011 11:29:21 +0000 (12:29 +0100)]
image: add gtk_image_get_preferred_size()

Add a function to query the preferred size of the GtkImage, and use it
in the get_preferred_width/height implementations.

12 years agotreeview: Ensure selection's changed signal
Benjamin Otte [Sat, 10 Dec 2011 06:54:34 +0000 (07:54 +0100)]
treeview: Ensure selection's changed signal

There was a corner case where the changed signal was not emitted.
If rows were built like this:
  1    (not selected)
  + 2  (selected)
  + 3  (not selected)
And row 1 was removed, no signal would be emitted.

I like it when writing tests actually finds bugs that have been around
since 2003 - introduced by 4a03ea2334bde7d2e15d0933c5a78f5a2338a854
actually. :)

12 years agotreeview: Don't try to optimize away selection_changed signal
Benjamin Otte [Sat, 10 Dec 2011 05:49:29 +0000 (06:49 +0100)]
treeview: Don't try to optimize away selection_changed signal

We get certain cases, in particular with SELECTION_MULTIPLE, where we
cannot figure out in advance of real_set_cursor() if the selection will
actually change.

12 years agotests: Add select/unselect calls
Benjamin Otte [Sat, 10 Dec 2011 05:24:40 +0000 (06:24 +0100)]
tests: Add select/unselect calls

12 years agotreeview: Add a CURSOR_INVALID flag to set_cursor_row()
Benjamin Otte [Sat, 10 Dec 2011 05:10:05 +0000 (06:10 +0100)]
treeview: Add a CURSOR_INVALID flag to set_cursor_row()

This is necessary so the code doesn't attempt to do things with the
previous cursor, which currently cause crashes.

12 years agotreeview: Replace lots of boolean args with flags
Benjamin Otte [Sat, 10 Dec 2011 05:06:06 +0000 (06:06 +0100)]
treeview: Replace lots of boolean args with flags

I want to add more and 3 boolean args is essentially unreadable code.

12 years agotests: Add sanity checks for selection
Benjamin Otte [Sat, 10 Dec 2011 04:45:02 +0000 (05:45 +0100)]
tests: Add sanity checks for selection

We want to know that the selection emits "changed" when it changes.

12 years agodocs: Update code demo for newer glib
Benjamin Otte [Sat, 10 Dec 2011 04:26:06 +0000 (05:26 +0100)]
docs: Update code demo for newer glib

12 years agotests: Dump the operation that is performed to stderr
Benjamin Otte [Sat, 10 Dec 2011 03:52:48 +0000 (04:52 +0100)]
tests: Dump the operation that is performed to stderr

12 years agotests: Make treechanging test run as fast as possible
Benjamin Otte [Sat, 10 Dec 2011 03:52:33 +0000 (04:52 +0100)]
tests: Make treechanging test run as fast as possible

12 years agotreeview: Handle the case where the cursor row gets deleted
Benjamin Otte [Sat, 10 Dec 2011 03:48:12 +0000 (04:48 +0100)]
treeview: Handle the case where the cursor row gets deleted

Previously, the cursor would just become invalid, which used to
reselect the first row in the treeview later on (without a
cursor-changed signal). This leads to a crash now with the recent
refactorings.

The patch is longer than I'd like it to be, but the situation is kinda
complicated, because we want to make sure to move the cursor to a good
row. It works like this:
1) From the deleted row, search forward for the first row that is not
   going to be deleted and is not a separator.
2) If no such row exists, search backwards for a row that is not a
   separator.
3) If no such node exists, clear the cursor.

12 years agotreeview: Replace cursor handling by keeping the node
Benjamin Otte [Sat, 10 Dec 2011 01:06:33 +0000 (02:06 +0100)]
treeview: Replace cursor handling by keeping the node

Previously the code used a GtkTreeRowReference, which was (a) less
performant and more importantly (b) hiding errors.
The errors being hidden were the referenced row becoming invalid or
collapsed, because such rows would not be valid cursor rows and it would
be necesary to select a new row and emit a "cursor-changed" signal.

So if a crash bisects down to this commit, it is very likely that the
cursor has not correctly been updated and the cursor row is invalid.

12 years agotests: Put random number into local variable
Benjamin Otte [Sat, 10 Dec 2011 01:04:27 +0000 (02:04 +0100)]
tests: Put random number into local variable

This way, we can see what function had previously been called when the
checks fail.

12 years agotreeview: Use _gtk_rbtree_contains() in a11y code
Benjamin Otte [Sat, 10 Dec 2011 01:03:46 +0000 (02:03 +0100)]
treeview: Use _gtk_rbtree_contains() in a11y code

12 years agorbtree: Add _gtk_rbtree_contains()
Benjamin Otte [Sat, 10 Dec 2011 01:02:29 +0000 (02:02 +0100)]
rbtree: Add _gtk_rbtree_contains()

It's useful in a bunch of places, so split it out

12 years agoFix move-resize emulation with XI2
Matthias Clasen [Sat, 10 Dec 2011 02:21:09 +0000 (21:21 -0500)]
Fix move-resize emulation with XI2

https://bugzilla.gnome.org/show_bug.cgi?id=650943

12 years agowayland: Load cursors from wayland directory under XDG_DATA_DIRS
Rob Bradford [Mon, 5 Dec 2011 17:31:50 +0000 (17:31 +0000)]
wayland: Load cursors from wayland directory under XDG_DATA_DIRS

This means we don't have to have hardcoded "/usr/share/wayland" to find the
cursors.

This change also fixes up the warning messages for when loading fails.

12 years agoUpdated Breton translation
Denis Arnaud [Fri, 9 Dec 2011 03:31:33 +0000 (04:31 +0100)]
Updated Breton translation

12 years agoUpdated Breton translation
Denis Arnaud [Fri, 9 Dec 2011 03:20:51 +0000 (04:20 +0100)]
Updated Breton translation

12 years agoRevert 3fd7dbc94074826755da7fc3237e439ba6e81ada
Matthias Clasen [Thu, 8 Dec 2011 17:44:19 +0000 (12:44 -0500)]
Revert 3fd7dbc94074826755da7fc3237e439ba6e81ada

Calling gdk_keymap_add_virtual_modifiers causes _all_ virtual
modifiers to be added, which causes problem when they are co-located
on the same real modifier (as Super and Hyper often are). Effectively,
this made it impossible to enter key combinations involving Super,
since they all turn into Super+Hyper.

12 years agoTest some virtual modifiers
Matthias Clasen [Thu, 8 Dec 2011 13:02:18 +0000 (08:02 -0500)]
Test some virtual modifiers

This adds some accelerators involving Super and Hyper to testmerge.
This helps in testing our accel matching for those.

12 years agoprint to file: fixed lost filename in gui
Timothy Arceri [Thu, 8 Dec 2011 01:03:07 +0000 (20:03 -0500)]
print to file: fixed lost filename in gui

Stop widget creating an event before updating both directory and
filename values.

https://bugzilla.gnome.org/show_bug.cgi?id=603823

12 years agooverlay: Properly allocate the child upon addition
Benjamin Otte [Wed, 7 Dec 2011 23:20:49 +0000 (00:20 +0100)]
overlay: Properly allocate the child upon addition

This ensures that after addition the child gets assigned the proper
allocation and the child window is properly shown/hidden.

12 years agooverlay: Move code
Benjamin Otte [Wed, 7 Dec 2011 22:51:19 +0000 (23:51 +0100)]
overlay: Move code

Move code for child allocation into the child allocation function. Don't
keep it in the overlay allocation code.

See the next commit for why this is useful.

12 years agooverlay: Create child window before setting parent
Benjamin Otte [Wed, 7 Dec 2011 23:20:06 +0000 (00:20 +0100)]
overlay: Create child window before setting parent

Otherwise, the window will not be used when set_parent() maps the child.

12 years agoprogressbar: Set state flags on draw
Rui Matos [Wed, 7 Dec 2011 22:36:53 +0000 (22:36 +0000)]
progressbar: Set state flags on draw

We were missing setting state flags on the style context before drawing.

12 years agowidget: Unset window-unfocused in gtk_widget_unparent()
Rui Matos [Fri, 2 Dec 2011 15:21:20 +0000 (15:21 +0000)]
widget: Unset window-unfocused in gtk_widget_unparent()

Widgets without a parent aren't inside a toplevel window so we must remove
window-unfocused as it doesn't make sense outside a toplevel.

https://bugzilla.gnome.org/show_bug.cgi?id=661428

12 years agotreeview: Fix crash when scrolling to end
Benjamin Otte [Wed, 7 Dec 2011 19:09:16 +0000 (20:09 +0100)]
treeview: Fix crash when scrolling to end

This broke in a4630d0e7b233479825b059e0df0e6d65b0e6734

https://bugzilla.gnome.org/show_bug.cgi?id=665741

12 years agowindow: copy back window surface data when the implicit paint is flushed
Cosimo Cecchi [Tue, 6 Dec 2011 23:41:33 +0000 (18:41 -0500)]
window: copy back window surface data when the implicit paint is flushed

When an implicit paint is flushed during expose, e.g. because a
non-double buffered widget is painting, make sure to copy the existing
data from the window surface we rendered before flushing back to the
paint surface, instead of using an empty base.
Code was already handling that (and said so in the comment), but only
when no implicit paint was used at all, and not in the case when it's
flushed mid-expose.

12 years agowindow: trivial cleanup
Cosimo Cecchi [Tue, 6 Dec 2011 23:27:59 +0000 (18:27 -0500)]
window: trivial cleanup

12 years agowindow: remove unused arguments and variables
Cosimo Cecchi [Tue, 6 Dec 2011 22:48:24 +0000 (17:48 -0500)]
window: remove unused arguments and variables

Trivial cleanup, no functional change.

12 years agomodifierstyle: Fix compile errors
Benjamin Otte [Tue, 6 Dec 2011 17:01:57 +0000 (18:01 +0100)]
modifierstyle: Fix compile errors

Introduced in 7bd2446801ffd0fc2ac5cc98fcd675ac4f3935d6

12 years agoFix demo to use a "normal" transparent value now that it works.
Alexander Larsson [Tue, 6 Dec 2011 13:11:44 +0000 (14:11 +0100)]
Fix demo to use a "normal" transparent value now that it works.

The last commit fixed this, so lets have a sane value in the demo code.

12 years agoRemove broken optimization in style modifier overrides
Alexander Larsson [Tue, 6 Dec 2011 13:09:45 +0000 (14:09 +0100)]
Remove broken optimization in style modifier overrides

We used to only set the override color or font if it was different
than the existing value. However, that means you can't change it to
an override that is the same as the default value for the property.

With this fixed you can e.g. override with a color of 0,0,0,0 which
you couldn't before.

12 years agoTry to fix linking
Matthias Clasen [Tue, 6 Dec 2011 01:56:02 +0000 (20:56 -0500)]
Try to fix linking

We need to link against gdk against pangoft2 on X11.
Also add Requires.private to gdk-3.0.pc, and try to clean
things up some.

https://bugzilla.gnome.org/show_bug.cgi?id=665326

12 years agogtk-demo: Add test of transparent GdkWindows
Alexander Larsson [Mon, 5 Dec 2011 23:23:18 +0000 (00:23 +0100)]
gtk-demo: Add test of transparent GdkWindows

This shows a semi-transparent shadow from a GtkOverlay window

12 years agoRemove an unused variable
Matthias Clasen [Mon, 5 Dec 2011 23:07:41 +0000 (18:07 -0500)]
Remove an unused variable

12 years agogtk,notebook: Invalidate the gap side when reordering tabs
Carlos Garnacho [Mon, 5 Dec 2011 16:21:01 +0000 (17:21 +0100)]
gtk,notebook: Invalidate the gap side when reordering tabs

This narrow area is invalidated so the gap follows the tab
being reordered.

12 years agonotebook: Make the tab reordering dnd window transparent
Alexander Larsson [Mon, 5 Dec 2011 20:13:50 +0000 (21:13 +0100)]
notebook: Make the tab reordering dnd window transparent

12 years agostatusicon: fix the build on !X11
Cosimo Cecchi [Mon, 5 Dec 2011 18:21:11 +0000 (13:21 -0500)]
statusicon: fix the build on !X11

Uninevitable fallout from my last patch to GtkStatusIcon. Thanks to Rob
Bradford for testing this.

12 years agowidget: Update Pango context on state change
Benjamin Otte [Mon, 5 Dec 2011 17:26:17 +0000 (18:26 +0100)]
widget: Update Pango context on state change

The layout depends on the state flags, so it must be updated when they
change.

12 years agolabel: Clear layout on state change
Benjamin Otte [Mon, 5 Dec 2011 17:25:39 +0000 (18:25 +0100)]
label: Clear layout on state change

Selected or active text might be rendered differently.

12 years agolabel: Move callback to new stuff
Benjamin Otte [Mon, 5 Dec 2011 17:24:44 +0000 (18:24 +0100)]
label: Move callback to new stuff

We want to react to state flags changing, not to the (deprecated) state
type changing.

12 years agowayland: Port from wl_shell to wl_surface_shell
Rob Bradford [Mon, 5 Dec 2011 15:55:37 +0000 (15:55 +0000)]
wayland: Port from wl_shell to wl_surface_shell

Previously all the commands that acted on the shell took the surface that was
to be acted on as parameter. Now we retrieve an object from the shell that
represents its state for the surface. With that wl_shell_surface object we can
then call methods on that.

Signed-off-by: Kristian Høgsberg <krh@bitplanet.net>
12 years agostylecontext: always add the ARROW style class when rendering an arrow
Cosimo Cecchi [Fri, 2 Dec 2011 16:58:14 +0000 (11:58 -0500)]
stylecontext: always add the ARROW style class when rendering an arrow

When gtk_render_arrow() is called, always add an ARROW style class to
the GtkStyleContext before rendering, so themes can specify a different
color for it.

https://bugzilla.gnome.org/show_bug.cgi?id=665420

12 years agostylecontext: add GTK_STYLE_CLASS_ARROW
Cosimo Cecchi [Fri, 2 Dec 2011 16:57:44 +0000 (11:57 -0500)]
stylecontext: add GTK_STYLE_CLASS_ARROW

For e.g. submenu items arrow in GtkMenuItems.

https://bugzilla.gnome.org/show_bug.cgi?id=665420

12 years agostatusicon: port to GtkIconHelper
Cosimo Cecchi [Wed, 30 Nov 2011 22:50:51 +0000 (17:50 -0500)]
statusicon: port to GtkIconHelper

Just always render the pixbuf ourselves and set it on the GtkImage in
the X11 case.
Code for other backends was already rendering the pixbuf manually before
translating it into a native type, so this greatly simplifies the code.

https://bugzilla.gnome.org/show_bug.cgi?id=665409

12 years agoMerge branch 'transparent-windows'
Alexander Larsson [Mon, 5 Dec 2011 13:24:28 +0000 (14:24 +0100)]
Merge branch 'transparent-windows'

12 years agoMake window handle transparent
Alexander Larsson [Mon, 5 Dec 2011 11:51:55 +0000 (12:51 +0100)]
Make window handle transparent

Instead of painting the window background on the grip_window we now
only paint it on the GtkWindow->window, and we make the grip_window
have a transparent background.

We can't really make transparent window handle background optional
via css atm, because the handle color is actually based on the
background color, so if that is set to transparent we won't draw
anything.

12 years agogdk: gdk_window_get_update_area don't remove alpha covered areas
Alexander Larsson [Mon, 5 Dec 2011 11:18:22 +0000 (12:18 +0100)]
gdk: gdk_window_get_update_area don't remove alpha covered areas

gdk_window_get_update_area is supposed to get the area where things
need painting, and remove them from the update areas. However, if
some area is covered by other windows with an alpha background we
can't just expect whatever the app choses to render in the update
area as correct, so we don't actually remove these areas, meaning
they will get correctly rendered when we get to the expose handlers.

12 years agogdk: gdk_window_move_region
Alexander Larsson [Mon, 5 Dec 2011 11:16:55 +0000 (12:16 +0100)]
gdk: gdk_window_move_region

gdk_window_move_region doesn't move children, so we can't copy
transparent child window regions with copyarea, so we remove these
from the copy region.

12 years agogdk: Fix alpha window background paintings on non-implicit paints
Alexander Larsson [Mon, 5 Dec 2011 10:40:30 +0000 (11:40 +0100)]
gdk: Fix alpha window background paintings on non-implicit paints

12 years agogdk: Remove unused _gdk_window_calculate_full_clip_region
Alexander Larsson [Mon, 5 Dec 2011 09:59:07 +0000 (10:59 +0100)]
gdk: Remove unused _gdk_window_calculate_full_clip_region

12 years agoHandle has_alpha_background for parent-relative backgrounds
Alexander Larsson [Mon, 5 Dec 2011 09:55:55 +0000 (10:55 +0100)]
Handle has_alpha_background for parent-relative backgrounds

12 years agoentry: avoid a mismatched cairo_save/restore
Cosimo Cecchi [Mon, 5 Dec 2011 00:57:25 +0000 (19:57 -0500)]
entry: avoid a mismatched cairo_save/restore

Leftover from GtkIconHelper migration.

12 years agoRemove unused alignment from GtkTooltip.
Paolo Borelli [Sun, 4 Dec 2011 18:20:23 +0000 (19:20 +0100)]
Remove unused alignment from GtkTooltip.

12 years agoPlug tiny leak in the css parser
Paolo Borelli [Sat, 3 Dec 2011 16:51:36 +0000 (17:51 +0100)]
Plug tiny leak in the css parser

12 years agowayland: Update to recent API changes
Kristian Høgsberg [Wed, 30 Nov 2011 15:21:27 +0000 (10:21 -0500)]
wayland: Update to recent API changes

12 years agoGtkIconView: fix a memory corruption in the drag-n-drop code
Claudio Saavedra [Thu, 1 Dec 2011 21:47:36 +0000 (22:47 +0100)]
GtkIconView: fix a memory corruption in the drag-n-drop code

You shall free with g_slice_free() what you allocate with
g_slice_new().

https://bugzilla.gnome.org/show_bug.cgi?id=665338

12 years agogdk: Fix repaint of layered region during move_region
Alexander Larsson [Thu, 1 Dec 2011 15:45:36 +0000 (16:45 +0100)]
gdk: Fix repaint of layered region during move_region

12 years agogdk: Fix repaint of layered region during scroll
Alexander Larsson [Thu, 1 Dec 2011 15:45:18 +0000 (16:45 +0100)]
gdk: Fix repaint of layered region during scroll

12 years agogdk: Fix repaint of layered region during raise
Alexander Larsson [Thu, 1 Dec 2011 15:44:51 +0000 (16:44 +0100)]
gdk: Fix repaint of layered region during raise

12 years agogdk: Make sure we don't copy to/from layered regions when moving a window
Alexander Larsson [Thu, 1 Dec 2011 15:27:37 +0000 (16:27 +0100)]
gdk: Make sure we don't copy to/from layered regions when moving a window

12 years agogdk: Track the layered area
Alexander Larsson [Thu, 1 Dec 2011 12:57:47 +0000 (13:57 +0100)]
gdk: Track the layered area

We track the areas that have alpha coverage so that we can
avoid using these as sources when copying window contents.
We also don't remove such areas from the clipping regions so
that they are painted both by parent and child.

12 years agogdk: Track wether windows have alpha in the background
Alexander Larsson [Thu, 1 Dec 2011 12:57:27 +0000 (13:57 +0100)]
gdk: Track wether windows have alpha in the background

This will let us handle such windows differently in the
drawing machinery

12 years agotestwindows: Add a button to cause a repaint
Alexander Larsson [Thu, 1 Dec 2011 15:21:45 +0000 (16:21 +0100)]
testwindows: Add a button to cause a repaint

This makes it easy to find various kinds of repaint bugs

12 years agotestwindows: Test transparent backgrounds
Alexander Larsson [Thu, 1 Dec 2011 13:09:52 +0000 (14:09 +0100)]
testwindows: Test transparent backgrounds

12 years agogdk: Expose bottommost windows first
Alexander Larsson [Thu, 1 Dec 2011 12:42:09 +0000 (13:42 +0100)]
gdk: Expose bottommost windows first

This cleans up the expose handling a bit by using the existing
clip regions, and it allows us later to use painters algorithm
to do transparent windows.

12 years agogdk: Remove now unused region tags completely
Alexander Larsson [Thu, 1 Dec 2011 12:38:04 +0000 (13:38 +0100)]
gdk: Remove now unused region tags completely

12 years agoimage: use the default icon size if no icon size is provided
Cosimo Cecchi [Thu, 1 Dec 2011 00:46:31 +0000 (19:46 -0500)]
image: use the default icon size if no icon size is provided

If no icon size is provided (e.g. when setting icon-name or stock-id
using the GObject property directly), use the default icon size. This
matches the previous GtkImage behavior.

12 years agoimage: minor cleanup
Cosimo Cecchi [Thu, 1 Dec 2011 00:46:16 +0000 (19:46 -0500)]
image: minor cleanup

12 years agoIntroduce gtk_render_insertion_cursor
Paolo Borelli [Sun, 27 Nov 2011 15:38:04 +0000 (16:38 +0100)]
Introduce gtk_render_insertion_cursor

The new function provides an API that takes the PangoLayout and index
as input params, this way it handles strong and weak cursors internally
factoring out all code duplicated in the widgets that need to render
cursors.

https://bugzilla.gnome.org/show_bug.cgi?id=640317

12 years agoUse the split-cursor setting for textview cursors
Paolo Borelli [Sun, 27 Nov 2011 14:12:36 +0000 (15:12 +0100)]
Use the split-cursor setting for textview cursors

Explicitely check the split-cursor setting when drawing the textview
insertion cursor instead of relying on the cursor_direction set in the
textlayout. This makes the cursor drawin code more uniform with other
widgets in preparation to refactoring the cursor drawing code in a
shared function.

https://bugzilla.gnome.org/show_bug.cgi?id=640317

12 years agoRework GtkTextView cursor code.
Paolo Borelli [Sat, 26 Nov 2011 22:48:10 +0000 (23:48 +0100)]
Rework GtkTextView cursor code.

Move the handling of primary/secondary cursors to gtktextdisplay, which
makes code simpler and more consistent to how GtkLabel and GtkEntry
draw cursors, which is useful in preparation to further refactoring.

https://bugzilla.gnome.org/show_bug.cgi?id=640317

12 years agoTiny cleanup.
Paolo Borelli [Sat, 26 Nov 2011 20:09:25 +0000 (21:09 +0100)]
Tiny cleanup.

https://bugzilla.gnome.org/show_bug.cgi?id=640317

12 years agoRemove draw_insertion_cursor in label and entry
Paolo Borelli [Sat, 26 Nov 2011 22:11:08 +0000 (23:11 +0100)]
Remove draw_insertion_cursor in label and entry

Remove the the draw_insertion_cursor wrapper which just converts from
GtkTextDir to PangoDirection

https://bugzilla.gnome.org/show_bug.cgi?id=640317

12 years agoimage: simplify code
Cosimo Cecchi [Thu, 1 Dec 2011 00:06:21 +0000 (19:06 -0500)]
image: simplify code

We can unconditionally call into the icon helper methods now that we
removed the g_return_val_if_fail() checks.

12 years agoentry: simplify code
Cosimo Cecchi [Thu, 1 Dec 2011 00:05:34 +0000 (19:05 -0500)]
entry: simplify code

We can unconditionally call into the icon helper methods now that we
removed the g_return_val_if_fail() checks.

12 years agoicon-helper: remove useless checks
Cosimo Cecchi [Thu, 1 Dec 2011 00:04:17 +0000 (19:04 -0500)]
icon-helper: remove useless checks

When we change storage type, we always clear the rest of the object, so
these checks do not add any safety.

12 years agoicon-helper: don't warn out when the icon size is GTK_ICON_SIZE_INVALID
Cosimo Cecchi [Wed, 30 Nov 2011 23:32:01 +0000 (18:32 -0500)]
icon-helper: don't warn out when the icon size is GTK_ICON_SIZE_INVALID

We can call ensure_icon_size() for an empty image; don't warn out in
that case, but just return a zero width.

12 years agoicon-helper: don't reset pixel_size when clearing the helper
Cosimo Cecchi [Wed, 30 Nov 2011 23:15:05 +0000 (18:15 -0500)]
icon-helper: don't reset pixel_size when clearing the helper

Don't reset the pixel size when clearing the helper; pixel-size is e.g a
property of GtkImage, which is stable unless explicitly set.

12 years agoentry: always prepare the style context when using it for entry icons
Cosimo Cecchi [Wed, 30 Nov 2011 16:40:32 +0000 (11:40 -0500)]
entry: always prepare the style context when using it for entry icons

Make sure to always prepare the entry's style context for the entry icon
style when passing it to GtkIconHelper.

12 years agocellrendererpix: make sure to select the default helper as a fallback
Cosimo Cecchi [Wed, 30 Nov 2011 16:27:24 +0000 (11:27 -0500)]
cellrendererpix: make sure to select the default helper as a fallback

If we don't have expander pixbufs, select the default helper to render
the icon, even if the cell is an expander. This mimics what the old code
did.

12 years agocellrendererpix: port to GtkIconHelper
Cosimo Cecchi [Wed, 30 Nov 2011 15:30:32 +0000 (10:30 -0500)]
cellrendererpix: port to GtkIconHelper

12 years agoentry: port to GtkIconHelper
Cosimo Cecchi [Wed, 30 Nov 2011 14:43:09 +0000 (09:43 -0500)]
entry: port to GtkIconHelper

12 years agodnd: port to GtkIconHelper
Cosimo Cecchi [Wed, 30 Nov 2011 05:06:13 +0000 (00:06 -0500)]
dnd: port to GtkIconHelper

12 years agoimage: port to GtkIconHelper
Cosimo Cecchi [Wed, 30 Nov 2011 04:06:37 +0000 (23:06 -0500)]
image: port to GtkIconHelper

Use newly introduced GtkIconHelper to render pixbufs in GtkImage.

12 years agoicon-helper: add GtkIconHelper private object
Cosimo Cecchi [Wed, 30 Nov 2011 03:26:19 +0000 (22:26 -0500)]
icon-helper: add GtkIconHelper private object

GtkIconHelper is a helper object to easily obtain a pixbuf from
different icon sources (e.g. a GIcon, an icon name, a stock id, ...).

Code is ported from GtkImage, which will be adapted in the next commit.

12 years agoGtkCssProvider: don't segfault when CSS file is not found
Stef Walter [Tue, 22 Nov 2011 07:13:04 +0000 (08:13 +0100)]
GtkCssProvider: don't segfault when CSS file is not found

 * Fix null dereference in gtk_css_provider_take_error()
 * And another in gtk_css_provider_emit_error()

https://bugzilla.gnome.org/show_bug.cgi?id=664537

12 years agoXI2: Push error traps around XIQueryDevice calls
Matthias Clasen [Wed, 30 Nov 2011 05:16:43 +0000 (00:16 -0500)]
XI2: Push error traps around XIQueryDevice calls

This may help against the crashes at resume time that some
people have been hitting when input devices mysteriously
disappear.

12 years ago[GtkFileChooserDefault] Don't destroy and re-create the filename entry
Federico Mena Quintero [Tue, 29 Nov 2011 21:08:25 +0000 (15:08 -0600)]
[GtkFileChooserDefault] Don't destroy and re-create the filename entry

This is a leftover from the big rework to remove the expanded/collapsed mode
in Save mode.

Signed-off-by: Federico Mena Quintero <federico@gnome.org>
12 years ago[GtkPathBar] Remove unused error argument
Federico Mena Quintero [Tue, 29 Nov 2011 21:07:01 +0000 (15:07 -0600)]
[GtkPathBar] Remove unused error argument

It used to be that _gtk_path_bar_set_file() would return an error if
it wasn't able to switch to the specified file, but that hasn't been
the case for a long while now, since the file chooser became async.

Signed-off-by: Federico Mena Quintero <federico@gnome.org>
12 years agoAvoid shadowing a variable
Paolo Borelli [Tue, 29 Nov 2011 20:14:24 +0000 (21:14 +0100)]
Avoid shadowing a variable

Move bg_color variable to an inner scope.

12 years ago[path-bar] Add some comments to explain the async recursion on an I/O callback
Federico Mena Quintero [Mon, 28 Nov 2011 18:03:02 +0000 (12:03 -0600)]
[path-bar] Add some comments to explain the async recursion on an I/O callback

Signed-off-by: Federico Mena Quintero <federico@gnome.org>
12 years ago[path-bar] Only reset the children's styles if the children actually changed
Federico Mena Quintero [Mon, 28 Nov 2011 17:49:35 +0000 (11:49 -0600)]
[path-bar] Only reset the children's styles if the children actually changed

Signed-off-by: Federico Mena Quintero <federico@gnome.org>