* gtk/gtkdnd.c: (gtk_drag_begin_internal),
(gtk_drag_source_info_destroy), (gtk_drag_end),
(gtk_drag_grab_notify_cb):
Cancel drag when the grab is shadowed. (#122688)
add arrow-size property to control the minimum size of the arrow, have the
2006-11-01 Kristian Rietveld <kris@imendio.com>
* gtk/gtkcombobox.c (gtk_combo_box_class_init),
(gtk_combo_box_size_request): add arrow-size property to control
the minimum size of the arrow, have the arrow scale up with the
font by default. (#357950).
don't forget to queue a draw for the new cursor node. (Fixes #366548,
2006-10-30 Kristian Rietveld <kris@gtk.org>
* gtk/gtktreeview.c (gtk_tree_view_move_cursor_page_up_down): don't
forget to queue a draw for the new cursor node. (Fixes #366548,
reported by Sven Herzberg).
change left/right keybindings to collapse/expand rows instead of moving
2006-10-30 Kristian Rietveld <kris@gtk.org>
* gtk/gtktreeview.c (gtk_tree_view_class_init): change left/right
keybindings to collapse/expand rows instead of moving the focus
cursor. (#105895, Brian Bober and others).
Tor Lillqvist [Sat, 28 Oct 2006 23:37:18 +0000 (23:37 +0000)]
pixbufloader_*.def Remove pixbufloader_*.def, they are not used. (#365388)
2006-10-29 Tor Lillqvist <tml@novell.com>
* pixbufloader_*.def
* Makefile.am (EXTRA_DIST): Remove pixbufloader_*.def, they are
not used. (#365388) They were all identical, and exported just two
symbols, so if somebody needs them for a MSVC build that doesn't
use built-in modules it's trivial to hack makefile.msc to produce
a common .def file for the loaders. The two symbols that need to
be exported are fill_info and fill_vtable.
rework the area_above == 0 case to not overwrite the variables in the
2006-10-27 Kristian Rietveld <kris@gtk.org>
* gtk/gtktreeview.c (validate_visible_area): rework the area_above == 0
case to not overwrite the variables in the lower scope (which are
used later on ...), get the path correctly and bail out when there
isn't a node above us. (#359231).
Tor Lillqvist [Wed, 11 Oct 2006 12:41:14 +0000 (12:41 +0000)]
Enable having some gdk-pixbuf loaders built-in even if loading the others
2006-10-11 Tor Lillqvist <tml@novell.com>
* configure.in: Enable having some gdk-pixbuf loaders built-in
even if loading the others dynamically. Define Automake
conditional INCLUDE_FOO for each loader as TRUE if that loader is
built-in. See also gdk-pixbuf/ChangeLog.
2006-10-11 Tor Lillqvist <tml@novell.com>
Enable having some loaders built-in even if loading the others
dynamically. Have loaders in the same order as in configure.in in
all places where they are handled/listed.
* Makefile.am: Define the STATIC_FOO_LIB and FOO_LIB macros
conditionally depending on whether said loader is built-in or
not. Use the names libstatic-pixbufloader-foo.la instead of
libpixbufloader-static-foo.la for the built-in ones so that the
"echo libpixbufloader-*.la" won't match them.
* gdk-pixbuf-io.c: Reorganize code as to plumb the built-in
loaders also if USE_GMODULE. Don't warn about a missing
gdk-pixbuf.loaders file if we have at least one included loader.
* gdk-pixbuf-private.h
* io-*.c: Don't define a generic MODULE_ENTRY, as we can't use it
in the loaders anyway. Each loader needs to check if INCLUDE_foo
is defined.
Tor Lillqvist [Tue, 10 Oct 2006 21:51:11 +0000 (21:51 +0000)]
Merge from 2.10 branch:
2006-10-11 Tor Lillqvist <tml@novell.com>
Merge from 2.10 branch:
* gtk/gtkmain.c (do_pre_parse_initialization): On Win32, if
environment variable LC_ALL or LANG is set, set the Win32 thread
locale to the corresponding locale. Then call the C library
setlocale() to set the C library locale accordingly. The
inconsistency mentioned below is gone. (#339756) Do some special
casing for Serbia and Montenegro. Handle the Latin and Cyrillic
scripts for Azeri, Uzbek and Serbian.
(enum_locale_proc): Helper function for the above functionality.
* gtk/gtkcalendar.c (gtk_calendar_init): No longer need to check
if the environment variables are set here, as they have already
been taken into account and the Win32 thread locale has been
set.
Tim Janik [Tue, 10 Oct 2006 14:40:50 +0000 (14:40 +0000)]
applied patch from Michael Natterer to move to inline docs. applied
Tue Oct 10 16:38:23 2006 Tim Janik <timj@imendio.com>
* gtk/tmpl/gtkbindings.sgml:
* gtk/gtkbindings.c: applied patch from Michael Natterer to move to
inline docs. applied wording fixes suggested by Martyn Russell.
Matthias Clasen [Mon, 9 Oct 2006 14:36:16 +0000 (14:36 +0000)]
Update the docs to state which out parameters may be NULL. (#360870,
2006-10-09 Matthias Clasen <mclasen@redhat.com>
* gdk/x11/gdkkeys-x11.c (gdk_keymap_translate_keyboard_state):
Update the docs to state which out parameters may be NULL.
(#360870, Attilio Fiandrotti)
add getter/setter for show-expanders property, queue a redraw when this
2006-10-08 Kristian Rietveld <kris@gtk.org>
* gtk/gtktreeview.[ch] (gtk_tree_view_set_property),
(gtk_tree_view_set_show_expanders),
(gtk_tree_view_get_show_expanders): add getter/setter for
show-expanders property, queue a redraw when this property
is toggled (#351167, Martin Ejdestig),
(gtk_tree_view_set_level_indentation),
(gtk_tree_view_get_level_indentation): also add getter/setter
and docs for level-indentation property.
Tor Lillqvist [Sat, 7 Oct 2006 21:32:53 +0000 (21:32 +0000)]
Use the Win32 thread locale's first day of week setting on Win32.
2006-10-07 Tor Lillqvist <tml@novell.com>
* gtk/gtkcalendar.c (gtk_calendar_init): Use the Win32 thread
locale's first day of week setting on Win32. (#339752, Bogdan
Nicula) Don't do this if one of the environment variables that
affect gettext is set, though. In that case use the week start day
from the corresponding message catalog.
Unfortunately the same logic isn't possible in the weekday and
month name lookup; there even if you have set one of the
aforementioned environment variables, you still get the weekday
and month names from the Win32 thread locale. Yes, this is
inconsistent.
Use only wide-character API in the Win32 code in this file, too.
(gtk_binding_entry_skip): new API which marks the entry as unbound.
Changed code so it returns FALSE when "marks_unbound == TRUE" is
encountered while activating bindings, effectively letting the
binding appear unbound (regardless of still existing bindings in
lower binding priority levels). Fixes bug #358329.
(gtk_binding_entry_add)
(gtk_binding_entry_clear)
(gtk_binding_entry_add_signall)
(gtk_binding_parse_binding): deprected these functions.
(_gtk_binding_parse_binding)
(_gtk_binding_entry_add_signall): new internal API.
Matthias Clasen [Mon, 2 Oct 2006 03:50:38 +0000 (03:50 +0000)]
Make the event_window large enough to cover the full height of the
2006-10-01 Matthias Clasen <mclasen@redhat.com>
* gtk/gtkexpander.c (gtk_expander_realize)
(gtk_expander_size_allocate): Make the event_window large
enough to cover the full height of the label_widget. (#358351,
Scott Horowitz)
bring zero ref count loops in sync with each other and the sort model.
2006-09-26 Kristian Rietveld <kris@imendio.com>
* gtk/gtktreemodelfilter.c (gtk_tree_model_filter_free_level),
(gtk_tree_model_filter_ref_node),
(gtk_tree_model_filter_real_unref_node): bring zero ref count loops
in sync with each other and the sort model.
unref the virtual root path once that node (or one if its ancestors) is
2006-09-26 Kristian Rietveld <kris@imendio.com>
* gtk/gtktreemodelfilter.c (gtk_tree_model_filter_init),
(gtk_tree_model_filter_finalize), (gtk_tree_model_filter_row_deleted),
(gtk_tree_model_filter_new): unref the virtual root path once that
node (or one if its ancestors) is deleted, set virtual_root_deleted
so we only unref it once.
ref count nodes the proper way, (gtk_tree_model_sort_ref_node),
2006-09-26 Kristian Rietveld <kris@imendio.com>
* gtk/gtktreemodelsort.c (gtk_tree_model_sort_row_changed),
(gtk_tree_model_sort_sort_level): ref count nodes the proper way,
(gtk_tree_model_sort_ref_node), (gtk_tree_model_sort_real_unref_node),
(gtk_tree_model_sort_free_level): bring zero ref count loops in
sync,
(gtk_tree_model_sort_free_level): free child levels before
decreasing the zero ref count of the current level,
(gtk_tree_model_sort_clear_cache): only clear cache if zero_ref_count
is > 0.
if the tree view has already been realized; scroll to background area
2006-09-25 Kristian Rietveld <kris@gtk.org>
* gtk/gtktreeview.c (gtk_tree_view_scroll_to_cell): if the tree view
has already been realized; scroll to background area instead of
cell area. (Fixes #330683, Johan Dahlin, Tommi Komulainen).
actually add align-set property to the API (left-over patch from #157439).
2006-09-24 Kristian Rietveld <kris@gtk.org>
* gtk/gtkcellrenderertext.c (gtk_cell_renderer_text_class_init),
(gtk_cell_renderer_text_[sg]et_property): actually add align-set
property to the API (left-over patch from #157439).
draw the indicator if the column has a sort column id (and the model is
2006-09-24 Kristian Rietveld <kris@gtk.org>
* gtk/gtktreeviewcolumn.c (gtk_tree_view_column_update_button): draw
the indicator if the column has a sort column id (and the model is
sortable) or if the user explicitly requested the indicator
to be shown. (Fixes #352738, Chris Vine).
* gtk/gtkfilechooserdefault.c (shortcuts_reload_icons):
(shortcuts_insert_path):
* gtk/gtkfilechooserbutton.c (change_icon_theme):
(model_add_bookmarks):
(model_update_current_folder):
(update_label_and_image):
If the bookmark points to a remote file, don't call get_info(),
since that may a) take a long time and b) pop up an auth dialog.
Instead, just use a folder icon and create a display name
from the uri.
* gtk/gtkfilechooserdefault.c (_gtk_file_chooser_label_for_uri):
New function to create a suitable display name for a remote
uri. This should really be done in GtkFileSystem.
Michael Natterer [Thu, 21 Sep 2006 17:05:33 +0000 (17:05 +0000)]
Implement lots of value setters for GdkGC, based on a heavily modified
2006-09-21 Michael Natterer <mitch@imendio.com>
Implement lots of value setters for GdkGC, based on a heavily
modified patch from Thomas Broyer (bug #328853):
* gdk/quartz/gdkcolor-quartz.c: removed functions which set colors
on the CGContext. Instead, added gdk_quartz_get_rgba_from_pixel()
which simply returns RGBA values from a GdkColor's pixel value.
See gdk_quartz_update_context_from_gc() below.
* gdk/quartz/gdkprivate-quartz.h (struct GdkGCQuartz): added lots
of members for the newly suppored GC values. Added enum
GdkQuartzContextValuesMask which is used for setting up the
CGContext for filling and/or stroking.
* gdk/quartz/gdkgc-quartz.c (gdk_quartz_gc_get_values)
(gdk_quartz_gc_set_values)
(_gdk_windowing_gc_copy): support a lot more GC values.
(gdk_quartz_update_context_from_gc): added
GdkQuartzContextValuesMask parameter and set filling/stroking
parameters accordingly. This function also gained full control
over the FG and BG colors (they can't be set separately any more).
The stipple mask part of the patch doesn't work but seems to take
the right approach and doesn't make things worse, so I applied it.
Did *not* apply the clipping part of the patch since I don't
understand it (I don't understand the version in CVS either, but
it at least works :-)
* gdk/quartz/gdkdrawable-quartz.c: pass the right masks to
gdk_quartz_update_context_from_gc() and removed separate color
setting calls. Some minor fixes.
* gdk/quartz/gdkwindow-quartz.c
(gdk_window_impl_quartz_begin_paint_region): set the CGContext's
fill color manually. We don't have/need a GC here.
Michael Natterer [Thu, 21 Sep 2006 11:19:25 +0000 (11:19 +0000)]
fix cast: it's drawable_impl->wrapper that is the GdkWindow, not the
2006-09-21 Michael Natterer <mitch@imendio.com>
* gdk/quartz/gdkwindow-quartz.c
(gdk_window_impl_quartz_begin_paint_region): fix cast: it's
drawable_impl->wrapper that is the GdkWindow, not the
drawable_impl itself.
just check the first shift-level of keyboard layout for RTL and LTR
2006-09-20 Behnam Esfahbod <behnam@zwnj.org>
* gdk/x11/gdkkeys-x11.c:
(get_direction): just check the first shift-level of keyboard
layout for RTL and LTR keysyms() (compliment to #116626)
Tim Janik [Wed, 20 Sep 2006 15:02:01 +0000 (15:02 +0000)]
always update the bin_window position and size in size_allocate, even if
Wed Sep 20 16:59:38 2006 Tim Janik <timj@imendio.com>
* gtk/gtkviewport.c: always update the bin_window position and size
in size_allocate, even if we don't have a visible child. this fixes
expose artefacts as described in #313508.
* gtk/gtkscrolledwindow.c: make hiding of automatic scrollbars when we
have no visible child explicit.
projects / ~andy / gtk / log