2005-07-18 Matthias Clasen <mclasen@redhat.com>
* gtk/gtkmain.c (gtk_parse_args, gtk_init_with_args): Call
gettext_initialization before checking gtk_initialized, otherwise
there are scenarios where it is not called at all. For an
example, see tests/testfilechooserbutton.c. (#310323,
Arkady L. Shane)
2005-07-18 Tor Lillqvist <tml@novell.com>
* configure.in (GDK_EXTRA_LIBS): [Win32] Bypass libtool, use
-Wl,-luuid. This avoids a libtool warning as libuuid is static.
* gdk/win32/gdkgc-win32.c (gdk_win32_hdc_get): Be sure to remove
any possible leftover clipping region in the DC if we don't want
any clipping. Thanks to Robert Ögren. (#309823)
2005-07-18 Matthias Clasen <mclasen@redhat.com>
* gtk/gtkcombobox.c (gtk_combo_box_set_focus_on_click): Propagate
the focus-on-click value to the button. (#310543, Christian Persch)
2005-07-18 Matthias Clasen <mclasen@redhat.com>
* gtk/gtkwidget.c (gtk_widget_set_double_buffered): Add a note
about the need to handle expose events. We really need a chapter
on the GTK+ drawing model that we can refer to here... (#310765,
Steve Chaplin)
2005-07-16 Tor Lillqvist <tml@novell.com>
* configure.in (GDK_EXTRA_LIBS): [Win32] Put -luuid back, it is
needed in case one builds without the ie55 stuff. (#310582)
2005-07-15 Matthias Clasen <mclasen@redhat.com>
* gtk/gtkentry.c (gtk_entry_move_forward_word):
(gtk_entry_move_backward_word): Don't treat whitespace as
words for keynav, pointed out by Owen Taylor.
2005-07-15 Federico Mena Quintero <federico@ximian.com>
* gtk/gtkfilechooserdefault.c
(should_respond_after_confirm_overwrite): Take the parent folder
path instead of the display name, to avoid doing duplicated work
in gtk_file_chooser_default_should_respond().
2005-07-15 Federico Mena Quintero <federico@ximian.com>
* gtk/gtkfilechooserdefault.c
(should_respond_after_confirm_overwrite): Take the parent folder
path instead of the display name, to avoid doing duplicated work
in gtk_file_chooser_default_should_respond().
2005-07-15 Federico Mena Quintero <federico@ximian.com>
* gtk/gtkfilechooserdefault.c
(confirm_dialog_should_accept_filename): Take the display name of
the folder as well so that we can present a better message. Use a
stock cancel button, "_Replace", and GTK_STOCK_SAVE_AS.
(gtk_file_chooser_default_should_respond): Pass the file part and
folder display name directly to the function above. Oops, only do
the confirmation dialog if the file exists.
2005-07-15 Federico Mena Quintero <federico@ximian.com>
* gtk/gtkfilechooserdefault.c
(confirm_dialog_should_accept_filename): Use header capitalization
for the buttons, per the HIG.
2005-07-15 Federico Mena Quintero <federico@ximian.com>
* gtk/tmpl/gtkfilechooserdialog.sgml: Add a "save" example to the
typical usage.
* gtk/gtk-sections.txt: Add GtkFileChooserConfirmation,
gtk_file_chooser_set_do_overwrite_confirmation,
gtk_file_chooser_get_do_overwrite_confirmation.
* gtk/tmpl/gtkfilechooser.sgml: Document the confirmation signal
and enumeration; provide an example.
2005-07-14 Federico Mena Quintero <federico@ximian.com>
Add overwrite confirmation for SAVE mode. Fixes bug #152850:
* gtk/gtkfilechooser.h: Add prototypes for
gtk_file_chooser_set/get_do_overwrite_confirmation().
(GtkFileChooserConfirmation): New enum for the result of the
"confirm-overwrite" signal.
* gtk/gtkmarshalers.list: Add ENUM:VOID.
* gtk/gtkfilechooser.c (gtk_file_chooser_class_init): Add a
"do-overwrite-confirmation" boolean property.
(gtk_file_chooser_set_do_overwrite_confirmation): Implement.
(gtk_file_chooser_get_do_overwrite_confirmation): Implement.
(confirm_overwrite_accumulator): New accumulator for the signal.
* gtk/gtkfilechooserutils.h (GtkFileChooserProp): Add
GTK_FILE_CHOOSER_PROP_DO_OVERWRITE_CONFIRMATION.
* gtk/gtkfilechooserutils.c
(_gtk_file_chooser_install_properties): Override the
do-overwrite-confirmation property.
* gtk/gtkfilechooserprivate.h (struct _GtkFileChooserIface): Add a
confirm_overwrite signal to the vtable.
* gtk/gtkfilechooserdefault.c
(gtk_file_chooser_default_set_property): Handle the new property.
(gtk_file_chooser_default_get_property): Likewise.
(get_selected_file_info_from_file_list): New helper function; code
taken from list_selection_changed().
(list_selection_changed): Use get_selected_file_info_from_file_list().
(should_respond_after_confirm_overwrite): New function.
(gtk_file_chooser_default_should_respond): Confirm when necessary.
2005-07-14 Tor Lillqvist <tml@novell.com>
* gtk/gtkfilesystemwin32.c (filename_get_info): Implement getting
the MIME type of a file. Look it up in the Registry. Bug reported
by Hans Oesterholt.
2005-07-14 Tor Lillqvist <tml@novell.com>
* configure.in (GDK_EXTRA_LIBS): Don't need -luuid on Win32. It
was some leftover, and in fact occasionally caused issues with
libtool as it made its way into the libgdk-win32-2.0.la file and
that way into the .la files of all libraries that depend on
GDK. (This happened only when building from CVS and not using a
full "make install" which would run the sanitize-la.sh script.)
2005-07-14 Matthias Clasen <mclasen@redhat.com>
* gdk/x11/gdkevents-x11.c (set_screen_from_root): Don't assert if
the xrootwin is invalid. (#309997, Barbie LeVille)
(gdk_event_translate): Ignore events with an invalid root field.
2005-07-14 Matthias Clasen <mclasen@redhat.com>
* configure.in: Set interface age to 0.
* gtk/gtkimmodule.h: Move G_BEGIN_DECLS to the right spot, noticed
by Akira Tagoh.
2005-07-14 Kristian Rietveld <kris@gtk.org>
* gtk/gtktreeviewcolumn.[ch] (gtk_tree_view_column_queue_resize): new
function (fixes#169961, reported by Morten Welinder).
2005-07-14 Kristian Rietveld <kris@gtk.org>
* gtk/gtktreeview.[ch] (gtk_tree_view_get_visible_range): new
function. (fixes#105252, suggestion from Mikael Hallendal).
2005-07-13 Kristian Rietveld <kris@gtk.org>
* gtk/gtktreeview.c (gtk_tree_view_bin_expose): paint a flat box
on the area the treeview got allocated extra vertically (fixes
#142063, reported by Brian Bober).
* gtk/gtktreeviewcolumn.c (gtk_tree_view_column_cell_process_action):
make this function calculate real_cell_area and real_background_area
correctly. (fixes#309249, reported by Bernd Demian).
2005-07-13 Matthias Clasen <mclasen@redhat.com>
* gtk/gtkcalendar.c (calendar_realize_arrows): Fix an
unintentional case of width-for-height... (#310133, Luis Villa)
2005-07-13 Kristian Rietveld <kris@gtk.org>
* gtk/gtktreeview.c (gtk_tree_view_new_column_width): let's take
the MIN of width and max_width here, not width and max_width != -1.
(fixes#144480, reported by Mikael Magnusson).
2005-07-13 Matthias Clasen <mclasen@redhat.com>
* gtk/gtkicontheme.c (load_themes): Don't keep the caches
for unthemed directories in a separate list, instead append
them at the end of the dir_mtimes list, so that we can check
them for staleness like the other caches. (#310221, Mark
McLoughlin)
2005-07-13 Robert gren <gtk@roboros.com>
Fix handling of Aiptek and Aiptek-like graphical tablets such as Trust
on Windows. (#167004, thanks to "pnohant" for testing)
Code cleanup: Remove the unused and unmaintained code for using a
non-system tablet context (USE_SYSCONTEXT ifdef:s), suggested
by Tor Lillqvist.
* gdk/win32/gdkinput-win32.c (_gdk_input_wintab_init_check): Handle
tablet drivers that don't support WTI_DDCTXS/WTI_DSCTXS gracefully.
Sligthly reorganize and simplify the code to support this.
(_gdk_input_wintab_init_check, _gdk_input_other_event): Remove
USE_SYSCONTEXT ifdef:s and the code for the case it was not defined.
(gdk_input_window_find_within): Removed.
2005-07-13 Kristian Rietveld <kris@gtk.org>
* gtk/gtktreeview.c (validate_row): when !separator, don't add
the focus_line_width to width/height (happens in
_column_cell_get_size), but do add the vertical-separator to height,
we weren't taking this into account before.
* gtk/gtkentrycompletion.c (_gtk_entry_completion_resize_popup):
take vertical-separator into account. (#309137, Ross Burton).
2005-07-13 Kristian Rietveld <kris@gtk.org>
* gtk/gtktreeselection.c (gtk_tree_selection_selected_foreach): quit
maintaining the iter on every iteration, only get the iter when
we are about to call the foreach_func. Gives us a 10x speedup,
since maintaining iters is a lot more expensive than maintaining
paths. We lose a bit of sanity checking though. Thanks go to
Billy Biggs for pointing this out.
