2001-02-15 Havoc Pennington <hp@redhat.com>
* gdk/gdkpango.c (gdk_pango_layout_line_get_clip_region): convert
the X coordinates so they're with respect to the line, rather than
with respect to the layout.
* gtk/gtkalignment.c: Convert to new property API, patch from Lee
Mallabone
* gtk/testgtk.c (create_range_controls): add vscale tests,
and inverted test
* gtk/gtkrange.c (gtk_range_set_inverted): new function to fix
#50806
* gtk/gtkentry.c (gtk_entry_get_text): add G_CONST_RETURN
* gtk/gtktextiter.h (gtk_text_iter_is_last): rename
gtk_text_iter_is_end
* gtk/gtktextbuffer.h (gtk_text_buffer_get_last_iter):
rename gtk_text_buffer_get_end_iter
* gtk/testgtk.c (create_labels): Add test for selectable
* gtk/gtkentry.c (gtk_entry_draw_text): Use new GDK API to draw
the selection stuff. This code is kind of broken since it doesn't
use the theme engine.
* gdk/gdkpango.c (gdk_pango_layout_line_get_clip_region):
fix infinite loop and y offset problem
(gdk_draw_layout_line_with_colors): fix foreground color handling
* gtk/gtklabel.h, gtk/gtklabel.c: Implement a "selectable" flag
that makes the label selectable.
* gtk/gtklabel.c (gtk_label_style_set): recreate the label's
layout when the style is set, since fonts etc. could have changed.
2001-02-13 Alexander Larsson <alla@lysator.liu.se>
* gdk/linux-fb/gdkdrawable-fb2.c (gdk_fb_clip_region):
Add a shortcut if the window and the current clip_region doesn't
overlap. This is needed when there are a lot of windows in a
window. The layout test in testgtk is a good test.
* gdk/linux-fb/gdkwindow-fb.c:
Whitespace fix.
2001-02-13 Alexander Larsson <alexl@redhat.com>
* gdk/linux-fb/gdkcursor-fb.c (gdk_fb_cursor_unhide):
Don't reference last_cursor if it is null. Fixes crash when rotating
the screen before having moved the mouse.
* gdk/linux-fb/gdkcc-fb.c: Zapp!
* gdk/linux-fb/Makefile.am: Don't compile gdkcc-fb.c
* gdk/linux-fb/gdkwindow-fb.c: Fix typo. width->height. Fixes redraw when
using 90 or 270 degrees rotation.
2001-02-12 Havoc Pennington <hp@pobox.com>
* gdk/gdkpango.c (gdk_pango_get_gc): fix bug where
it always set the foreground, even if we were only using
a stipple.
(gdk_draw_layout_line_with_colors): new function, allow
override colors
(gdk_draw_layout_with_colors): new function, allow override
colors
(gdk_pango_layout_line_get_clip_region): function to get the clip
region for a logical text range
(gdk_pango_layout_get_clip_region): get the clip region for a
logical text range
* gdk/x11/gdkcolor-x11.c: forward declare gdk_colormap_sync(),
(gdk_colormap_new): fix call to gdk_colormap_sync() so it has
the right number of arguments.
* gtk/gtktextbtree.c (gtk_text_btree_node_check_consistency):
enhance the function to check that node data corresponds to a
view still belonging to the tree.
* gtk/gtktreeview.c (gtk_tree_view_changed): we were leaking the
GtkTreePath
(gtk_tree_view_inserted): ditto
(gtk_tree_view_child_toggled): ditto
* gtk/gtktreemodel.c (gtk_tree_path_append_index): use realloc to
simplify this code.
* gtk/gtkcellrenderertext.c (get_layout): fix leak of
a PangoAttrList
* demos/gtk-demo/main.c (load_file): Fix leak of a GString
* gtk/gtkentry.c (gtk_entry_realize): Fix leak of a GdkCursor
* gtk/gtkmenubar.c (gtk_menu_bar_size_request): consider toggle
size in the size request
(gtk_menu_bar_size_allocate): consider toggle size here
* gtk/gtkimagemenuitem.h, gtkimagemenuitem.c: Menu item
that displays a widget in the toggle slot
* gtk/testgtk.c: test GtkImageMenuItem
* gtk/Makefile.am, gtk/gtk.h: Add GtkImageMenuItem
* gtk/gtkmenuitem.h: Use "gint" not "guint16" for toggle size
request and allocation
* gtk/gtkmenu.c (gtk_menu_size_request): use gint not guint16
* gtk/gtkcheckmenuitem.c
(gtk_check_menu_item_toggle_size_request): ditto
2001-02-11 Tor Lillqvist <tml@iki.fi>
* gdk/win32/gdkcolor-win32.c (gdk_colormap_query_color): New
function, copied from X11 version.
* gdk/win32/gdkwindow-win32.c (gdk_window_set_cursor): Eliminate
DestroyCursor failures. The reason for the problem (despite the
inconsistent messages from GetLastError()) was that
DestroyCursor() cannot be called while a cursor is set in a
window. Set the new cursor first and then destroy the old one.
Thanks to jpe@archaeopteryx.com.
* gdk/win32/gdkcc-win32.c: Removed.
* gdk/gdk.def
* gdk/makefile.mingw.in: Update. Godbye GdkColorContext.
* gtk/gtk.def
* gtk/makefile.mingw.in: Updates.
2001-02-08 Havoc Pennington <hp@redhat.com>
* gtk/gtkbin.c (gtk_bin_add): better error message if you try to
add a widget to a bin that already has a child.
* gdk/gdkcc.h, gdk/gdkcc-x11.c: Remove GdkColorContext, #50512
* docs/Changes-2.0.txt: note that GdkColorContext is gone.
* gdk/x11/Makefile.am (libgdk_x11_la_SOURCES): remove gdkcc-x11.c
* gdk/gdk.h: remove gdkcc.h
* gdk/Makefile.am (gdk_public_h_sources): remove gdkcc.h
* gtk/gtkpixmap.c (build_insensitive_pixmap): rewrite using
GdkPixbuf, since the previous implementation was GdkColorContext
dependent.
* gtk/testgtk.c (create_pixmap): add test for insensitive GtkPixmap
Mon Feb 5 22:05:57 2001 Owen Taylor <otaylor@redhat.com>
* gdk/linux-fb/gdkimage-fb.c (_gdk_fb_get_image): Fix syntax
error introduced with earlier commit) (pointed out by
Ollie Lho)
Sun Feb 4 07:55:17 2001 Tim Janik <timj@gtk.org>
* gtk/gtksignal.h: removed gtk_signal_init() compat define.
* gtk/gtktypeutils.[hc]: gtk_tpye_init() take an additional argument
GTypeDebugFlags debug_flags and pass it on to g_type_init ();
deprecated this function.
* gdk/gdk.c (gdk_init_check): call g_type_init () with 0 as
argument.
Sun Feb 4 07:45:45 2001 Tim Janik <timj@gtk.org>
* gtk/gtkdebug.h: nuke GTK_DEBUG_OBJECTS and GTK_DEBUG_SIGNALS.
Sun Feb 4 07:59:57 2001 Tim Janik <timj@gtk.org>
* gdk-pixbuf.h: DOH! don't include glib/gobject.h but glib-object.h,
we finally need the gobject->gruntime rename.
* *.c: scratched calls to g_type_init(), there's simply no point in
doing that, use gtk_init().
Sat Feb 3 11:28:03 2001 Owen Taylor <otaylor@redhat.com>
* configure.in: Remove confusing comment about X11.
Fri Feb 2 19:21:23 2001 Owen Taylor <otaylor@redhat.com>
* gdk/gdkwindow.c (gdk_propagate_shapes): Surround additional
regions with HAVE_SHAPE_EXT. (#50511, Eric Ding)
Fri Feb 2 18:31:54 2001 Owen Taylor <otaylor@redhat.com>
* gdk/x11/gdkimage-x11.c: Remove definition of _XOPEN_SOURCE, since
it breaks things on BSD. If you want to compile GTK+ with
-ansi -pedantic, you should supply CPP_FLAGS=_XOPEN_SOURCE
when configuring, since GTK+ simply can't be "pure ANSI"
in the header files it uses. (#8170, Greg Hudson)
Fri Feb 2 17:16:09 2001 Owen Taylor <otaylor@redhat.com>
* gdk/x11/gdkcolor-x11.c (gdk_colormap_sync): Make
gdk_colormap_sync private since it was never exported
in a header file.
* gdk/x11/gdkcolor-x11.c (gdk_colormap_new): Fill in
colormap->colors[] for StaticGray, StaticColor colormaps.
* gdk/gdkpixbuf-drawable.c: Fix problems where image->bpp
was being used as if it was image->bits_per_pixel.
* gdk/gdkimage.h, gdk/x11/gdkimage-x11.c: Save the bits-per-pixel
for the image in the GdkImage structure since it isn't
reconstructable, and we need it to handle packed types
* gdk/win32/gdkimage-win32.c: Set image->bits_per_pixel. (I'm making
the assumption that on Win32 image->bits_per_pixel == image->depth,
always.
* gdk/linux-fb/gdkimage-fb.c: Set image->bits_per_pixel.
* gdk/gdkrgb.c (gdk_rgb_select_conv): Exit with an informative
warning message if no converter can be found.
Fri Feb 2 12:26:50 2001 Owen Taylor <otaylor@redhat.com>
* gtk/gtkrc.c (gtk_rc_add_initial_default_files): Get rid of
a bunch of g_strdup_printf("%s%s") in favor of g_strconcat().
* gtk/gtkrc.c Makefile.am: Use $(libdir), not $(exe_prefix),
since some people set $(libdir) separately. (#1290, David Kaelbling)
Thu Feb 1 18:25:46 2001 Owen Taylor <otaylor@redhat.com>
* gtk/gtkfilesel.c: If PATH_MAX and MAXPATHLEN are not
defined, define MAXPATHLEN to 2048. (The Hurd doesn't have
MAXPATHLEN, but the code here depends on a fixed value.)
(#4524)
Wed Jan 31 22:01:04 2001 Owen Taylor <otaylor@redhat.com>
* gtk/gtkhandlebox.c (gtk_handle_box_button_changed): Handle the case
where child == NULL and handle_position == RIGHT or BOTTOM. (#8041g)
Wed Jan 31 21:20:39 2001 Owen Taylor <otaylor@redhat.com>
* gtk/gtkctree.c (real_tree_move): If the node being moved isn't
viewable there is no way that moving the node will cause the
focus row to become not viewable, so omit check on the visibility
of new_sibling, which is irrelevant. (Fixes#8002, David Helder)
Wed Jan 31 20:38:17 2001 Owen Taylor <otaylor@redhat.com>
* gtk/gtkentry.c (gtk_entry_commit_cb): Delete the current
selection before inserting new text.
Wed Jan 31 18:49:33 2001 Owen Taylor <otaylor@redhat.com>
* gtk/gtkoptionmenu.c (gtk_option_menu_item_state_changed_cb):
Make the sensitivity of the reparented child track that of
the original parent menu item. (#34218, David Hodson)
* gtk/gtkoptionmenu.c (gtk_option_menu_item_destroy_cb): Handle
the case where the current item is destroyed properly.
* gtk/gtkoptionmenu.c: Some additional code cleanups and fix
some edge cases with child-less menuitems.
Wed Jan 31 17:16:13 2001 Owen Taylor <otaylor@redhat.com>
* gtk/gtkcombo.c (gtk_combo_window_key_press): Make Return
key pop down window. (#12074, Jon K Hellan)
Wed Jan 31 16:21:42 2001 Owen Taylor <otaylor@redhat.com>
* gtk/gtklist.c (gtk_list_signal_item_toggle): Don't allow
toggling of rows off in BROWSE or EXTENDED mode. (#12072, Jon K Hellan)
The solution here isn't perfect - you get an extraneous
emission of "toggle", which could conceivably confuse an app,
but better than the current situation. LXR search seems to
indicate that no apps in GNOME CVS connect to "toggle".
Wed Jan 31 15:46:13 2001 Owen Taylor <otaylor@redhat.com>
* gtk/Makefile.am (libgtkinclude_HEADERS): Move gtkcompat.h from
gtk_public_h_sources to directly here to avoid warning when
building srcdir != builddir. (#9656)
Tue Jan 30 19:49:02 2001 Owen Taylor <otaylor@redhat.com>
* gtk/gtkrange.c: Patch from Kipp Hickman to make the event
handlers in gtkrange.c return the proper values (TRUE == handled)
(#10316).
This is just the tip of the iceberg, but gtkrange.c is the
most common place where the propagation is problematical,
and also a place where it is almost certainly safe to change
this in the stable branch.
(You don't want right click popups on a range control or anything...)
Tue Jan 30 18:57:59 2001 Owen Taylor <otaylor@redhat.com>
* gtk/gtktext.c (clear_focus_area): We need to clear the focus
area on focus out, even if a background pixmap isn't set.
(#13941)
Tue Jan 30 18:24:10 2001 Owen Taylor <otaylor@redhat.com>
* gtk/gtknotebook.c (gtk_notebook_set_shape): Fix from Sean Cunningham
to deal with setting the shape properly when scrolling arrows are
turned on, but not visible because there is sufficient space.
(#13432)
Tue Jan 30 16:39:25 2001 Owen Taylor <otaylor@redhat.com>
* gtk/gtkitemfactory.c (gtk_item_factory_delete_item): For menu
items with submenus, destroy the item along with the submenu.
(#7841, Brian Masney(?)) Also, handle paths of the form '<foo>/abcd...'
properly.
* gtk/testgtk.c (menu_items): Add a dummy branch that we delete
later.
Tue Jan 30 15:51:25 2001 Owen Taylor <otaylor@redhat.com>
* gtk/gtkwindow.c (gtk_window_real_set_focus): Fix a problem where
the focus widget sometimes wasn't drawn with the default if there
was no default widget.
* gtk/gtkstyle.c (gtk_style_real_unrealize): free colors,
unreference pixmaps.
* gtk/gtkstyle.c (gtk_style_realize): Reference colormap
for some extra safety.
Mon Jan 29 19:00:01 2001 Owen Taylor <otaylor@redhat.com>
* gtk/gtk{ctree.c,clist.c} (set_cell_contents): Handle setting
the text of a cell to the old pointer value better, by
copying the new text before freeing the old text. Some code
cleanup. (#8079, Karl Nelson)
Mon Jan 29 16:50:19 2001 Owen Taylor <otaylor@redhat.com>
* gtk/gtklabel.[ch] gtk/gtkframe.[ch]: Make gtk_label_get_text()
gtk_frame_get_label() non strdup'ing, and G_CONST_RETURN.
Mon Jan 29 15:22:51 2001 Owen Taylor <otaylor@redhat.com>
* gtk/gtkmenu.c (gtk_menu_remove): When removing an
item from a menu, check to see if it matches
menu->old_active_menu_item, and if so, unref and clear
old_active_menu_item (Patch from Pavel Cisler)
* gtk/gtkmenushell.c (gtk_menu_shell_remove): Unset
menu_shell->active_menu_item, if it is the child being
removed. (Patch based on that of Gene Ragan, #50337)
2001-01-26 Havoc Pennington <hp@redhat.com>
* gtk/gtktextlayout.c (convert_color): adapt to handle PangoColor
* gtk/gtktreeview.c (gtk_tree_view_widget_to_tree_coords): fix to
not offset by TREE_VIEW_HEADER_HEIGHT
(gtk_tree_view_tree_to_widget_coords): fix to not offset by
TREE_VIEW_HEADER_HEIGHT
* configure.in (included_loaders): for me, --with-included-loaders
generates the error "the specified loader yes does not exist",
i.e. the arg defaults to "yes", so change test for value ""
to test for value "yes", and include all loaders in that case.
* gtk/gtkrbtree.c (_gtk_rbtree_get_depth): new function
* gtk/gtktreeview.c (gtk_tree_view_get_cell_rect): fix to properly
handle TREE_VIEW_VERTICAL_SEPARATOR
(gtk_tree_view_bin_expose): fix to consider the row offset as
pointing halfway into vertical separator.
(gtk_tree_view_draw_node_focus_rect): ditto
* gtk/gtkdebug.h, gtk/gtkmain.c (gtk_init_check): Add
--gtk-debug=updates, which causes gdk_window_set_debug_updates
(TRUE) to be called.
* gdk/gdkwindow.c (gdk_window_set_debug_updates): Allow enabling a
debug mode where the invalid region is colored in on invalidate,
so you can see the flicker and know whether your redraw code is
doing a good job.
* gtk/gtktreeview.c (gtk_tree_view_queue_draw_node): Work in
tree window coordinates (clip rect is in tree window coords)
* gtk/Makefile.am: add gtktreednd.[hc]
* gtk/gtkliststore.c: implement gtktreednd interfaces.
* gtk/gtktreednd.c, gtk/gtktreednd.h: New interface to support
drag-and-drop data operations on a model (so we can set up tree
drag-and-drop automatically)
* gtk/testgtk.c: Add a window to change sensitivity in the
GtkLabel test; add a way to change the entry frame in GtkEntry
test
* gtk/gtkentry.c (gtk_entry_set_has_frame):
(gtk_entry_get_has_frame): new functions to remove the frame
around an entry
(gtk_entry_size_request): shrink requisition if no frame
(gtk_entry_draw_focus): don't draw frame if no frame
* gtk/gtkstyle.c (gtk_default_draw_check): draw custom look for
checks inside a cell renderer
(gtk_default_draw_option): ditto for options
* gtk/gtktreeviewcolumn.c (update_button_contents): add/remove
children from the alignment, not the button
(gtk_tree_view_column_init): ref/sink the column, to emulate
GObject refcounting.
* gtk/gtkcellrenderer.c (gtk_cell_renderer_init): ref/sink
* gtk/gtkcellrenderertoggle.c (gtk_cell_renderer_toggle_render):
Use theme functions to draw the toggles
* gdk/gdkpango.c (gdk_pango_get_gc): use GdkRGB to alloc colors
* gdk/gdkpango.h, gdk/gdkpango.c: Add GdkPangoAttrStipple and
GdkPangoAttrEmbossed to use in rendering insensitive text
* gdk/gdkpango.c (gdk_draw_layout_line): render new properties
* gtk/gtkstyle.c (gtk_default_draw_layout): handle sensitivity
using new GDK features
2001-01-22 <alexl@redhat.com>
* gdk/gdkwindow.c (gdk_window_invalidate_region):
Don't leak visible_region. Use visible_region instead of region.
* gdk/linux-fb/gdkwindow-fb.c (gdk_window_new):
Don't do extra gdk_drawable_ref(). This is not X.
(gdk_fb_window_find_focus, gdk_fb_change_focus): make sure
gdk_fb_focused_window is ref:ed.
(gdk_fb_window_send_crossing_events): Watch out for destroyed
windows.
2001-01-18 Tor Lillqvist <tml@iki.fi>
* gtk/gtkmain.c
* gtk/gtkrc.c: (Win32) Use the new
g_win32_get_package_installation_(sub)directory() functions.
* config.h.win32.in: New file.
* config.h.win32: Removed.
* configure.in: Output config.h.win32.
* Makefile.am: Add it to EXTRA_DIST.
* gtk/gtk.def: Update.
2001-01-16 Tor Lillqvist <tml@iki.fi>
* gdk/win32/gdkwindow-win32.c (gdk_window_show, gdk_window_hide):
Eliminate redraws when input-only windows are shown or
hidden. Thanks to jpe@archaeopteryx.com.
2001-01-18 Tor Lillqvist <tml@iki.fi>
* gdk-pixbuf-io.c (get_libdir): (Win32) Use the new
g_win32_get_package_installation_subdirectory() function.
(gtk_win32_get_installation_directory): Remove this then.
Tue Jan 9 11:20:48 2001 Owen Taylor <otaylor@redhat.com>
* gdk/x11/gdkdnd-x11.c: Remove unused variable.
* gtk/gtkdnd.c: Encapsulate storing and retrieving info
structures in functions. Fixes bug where gtk_dataset_*
was accidentally still being used in one place, causing
every dest side event to be treated independently.
* gtk/gtkdnd.c: Remove last vestages of handling
::draw (fixes warning)
* gtk/gtkentry.[ch]: Add drag and drop support.
* gtk/gtkdnd.[ch] gtk/gtktextview.c gtk/gtkentry.c: Add
new function gtk_check_drag_threshhold() for checking
to check if (dx,dy) has exceeded the threshhold for starting
a drag and use it everywhere.
2001-01-17 <alexl@redhat.com>
* gdk/linux-fb/gdkdrawable-fb2.c (gdk_shadow_fb_draw_arc):
Doh! Mixed up x and y.
* gdk/linux-fb/gdkglobals-fb.c:
Removed gdk_fb_pointer_grab_window_events,
added _gdk_fb_pointer_grab_owner_events and
_gdk_fb_keyboard_grab_owner_events.
* gdk/linux-fb/gdkmain-fb.c (gdk_fb_pointer_grab):
Set all button mask if GDK_BUTTON_MOTION_MASK set. Send
crossing events before grabbing. Set _gdk_fb_pointer_grab_owner_events.
(gdk_fb_pointer_ungrab): Unset _gdk_fb_pointer_grab_owner_events
Send crossing events after ungrab finished.
(gdk_keyboard_grab): Set ..._owner_events
(type_masks): Move out of function.
(gdk_fb_other_event_window, gdk_fb_pointer_event_window,
gdk_fb_keyboard_event_window): New functions, return the
window an event should be targeted at. Handles grabs and
event propagation. Can return NULL.
(gdk_event_make): Remove event_mask checking. Now always
returns an event.
* gdk/linux-fb/gdkkeyboard-fb.c:
* gdk/linux-fb/gdkproperty-fb.c:
* gdk/linux-fb/gdkselection-fb.c:
Use new event_window/gdk_event_make() behaviour.
* gdk/linux-fb/gdkmouse-fb.c:
Use new event_window/gdk_event_make() behaviour.
Only send motion events if in the same window.
If grabbed, use cursor from window if sibling of grabbed
window, and cursor from grabbed window otherwise.
* gdk/linux-fb/gdkprivate-fb.h:
Update gdk_fb_window_send_crossing_events, gdk_event_make
and grab varaible declarations. Add gdk_fb_*_event_window().
* gdk/linux-fb/gdkwindow-fb.c:
Use new event_window/gdk_event_make() behaviour.
Keep correct track of focus window.
(gdk_window_set_events): Set all specific button motion masks
if GDK_BUTTON_MOTION_MASK set.
* gtk/gtkdnd.c (gtk_drag_update):
Free info->last_event if gdk_drag_motion returned FALSE.
2001-01-16 Tor Lillqvist <tml@iki.fi>
* gdk/win32/gdkvisual-win32.c: Include <stdlib.h> for getenv().
* gdk/win32/gdkevents-win32.c (gdk_event_translate): Don't use
GDK_COLORMAP_PRIVATE_DATA on a potentially NULL GdkColormap.
* gdk/win32/gdkfont-win32.c (gdk_font_load_internal): Adapt calls
to g_{locale,filename}_{to,from}_utf8 to match current API.
* gtk/makefile.msc.in (gtk_OBJECTS)
* gtk/makefile.mingw.in (gtk_OBJECTS): Add gtkwindow-decorate.
* gtk/gtk.def: Update.
2001-01-15 Tor Lillqvist <tml@iki.fi>
* gdk/win32/gdkevents-win32.c (synthesize_crossing_events): New
implementaion that more closely mimics the events that an X server
would generate. Especially, properly generate GDK_NOTIFY_VIRTUAL and
GDK_NOTIFY_NONLINEAR_VIRTUAL crossing events. This makes buttons
with nested preview widgets, for instance, more responsive when
you move the mouse quickly into them.
(find_window_for_pointer_event): Make it actually compile.
(gdk_event_translate): Set send_event to the value of
InSendMessage (). This doesn't really mean the same as X11's
send_event flag, but is close enough, says jpe@archaeopteryx.com
2001-01-15 Alexander Larsson <alexl@redhat.com>
* gdk/linux-fb/gdkdnd-fb.c:
Implemented drag and drop.
Seems to be mostly working. I sometimes get an assert from gtkdnd.c:2664
when it tries to free and then copy the same event (info->last_event).
I am not sure about why this happens, but will continue looking.
2001-01-15 Alexander Larsson <alexl@redhat.com>
* gdk/linux-fb/gdkwindow-fb.c:
Use gdk_window_invalidate_rect instead of
gdk_window_invalidate_rect_clear to minimise flashing.
When hiding windows you need to clear the root window though.
2001-01-15 Alexander Larsson <alexl@redhat.com>
* gdk/linux-fb/gdkdrawable-fb2.c:
For optization, add an full_shapes argument to gdk_fb_clip_region.
If this is false, only the cliprects of the shapes are used.
* gdk/linux-fb/gdkprivate-fb.h:
Export gdk_fb_window_peek_shape.
Add full_shapes to gdk_fb_clip_region.
* gdk/linux-fb/gdkwindow-fb.c (gdk_fb_window_move_resize):
When moving a window, don't clear everything under the it, just the
root window. Makes stuff faster and flash less.
(gdk_window_get_pointer): Correctly handle offseted shapes, due to
USE_CHILD_SHAPE. (gdk_fb_window_peek_shape): Return the correct
offset of the shape.
2001-01-12 Havoc Pennington <hp@redhat.com>
* gdk/x11/gdkgeometry-x11.c (gdk_window_scroll): offset the
current invalid region, fixes redraw bug while scrolling the
text widget
* gtk/gtktextview.c, gtk/gtktextview.h:
Rearrange the scrolling/validation/etc. code in a major way,
so it seems to make sense to me. Probably isn't genuinely that
much better, but...
* gtk/gtktexttag.c (set_fg_color): fix name of property used for
notifies
* gtk/gtktextmark.c (gtk_text_mark_get_left_gravity): new function
2001-01-12 Alexander Larsson <alla@lysator.liu.se>
* gdk/linux-fb/gdkmain-fb.c (gdk_fb_set_rotation):
Hide mouse pointer while rotation. Show it afterwards.
2001-01-12 Alexander Larsson <alexl@redhat.com>
* gdk/linux-fb/gdkdrawable-fb2.c (gdk_fb_clip_region): Take shaped
windows in consideration when generating clip region.
* gdk/linux-fb/gdkfb.h:
Huge HACK. Added GDK_FB_USE_CHILD_SHAPE.
* gdk/linux-fb/gdkprivate-fb.h:
Add shape to GdkWindowFBDatat.
exported gdk_fb_window_get_abs_shape().
removed gdk_fb_draw_lines declaration.
* gdk/linux-fb/gdkwindow-fb.c (gdk_window_impl_fb_init):
Init shape to NULL. (gdk_fb_window_send_crossing_events): Don't
send focus changes when there is a grabbed window.
(gdk_window_get_pointer): Handle shaped windows.
(gdk_fb_region_create_from_bitmap): Convert bitmap to region.
(gdk_fb_window_peek_shape): Returns the shape for a window,
handles GDK_FB_USE_CHILD_SHAPE.
(gdk_fb_window_get_abs_shape): Returns the shape of the window,
offseted to absolute positions. Handles GDK_FB_USE_CHILD_SHAPE.
(gdk_window_shape_combine_mask): Implement.
* gtk/gtkwindow-decorate.c:
HACK! Set GDK_FB_USE_CHILD_SHAPE on window->frame to make sure
setting the shape of a window makes the window transparent.
2001-01-11 Alexander Larsson <alexl@redhat.com>
* acconfig.h:
Added ENABLE_SHADOW_FB
* configure.in:
Added --disable-shadowfb
* gdk/linux-fb/gdkcursor-fb.c:
Update shadowfb when updating cursor
* gdk/linux-fb/gdkdrawable-fb2.c:
Added wrappers for shadowfb that calls the normal drawable
methods, but calls gdk_shadow_fb_update(bounding box) when
GdkWindows are drawed to.
Moved gdk_draw_glyphs implementation to _gdk_draw_glyphs
which also returns the bounding box.
* gdk/linux-fb/gdkfb.h:
Added GdkFBAngle type and gdk_fb_set_rotation declaration.
* gdk/linux-fb/gdkgeometry-fb.c:
Update shadowfb when scrolling window.
* gdk/linux-fb/gdkglobals-fb.c:
Add _gdk_fb_screen_angle.
* gdk/linux-fb/gdkkeyboard-fb.c:
Test code for screen rotation. Shift-F2 in the xlate driver
rotates the screen.
* gdk/linux-fb/gdkmain-fb.c:
Handle shadowfb. Add gdk_fb_set_rotation(). Remove CM and RP.
* gdk/linux-fb/gdkmouse-fb.c:
Use fb_width/height instead of modeinfo.xres/yres.
* gdk/linux-fb/gdkprivate-fb.h:
Added fb_men, fb_width, fb_height & fb_stride. When using
shadow fb these can differ from the framebuffer stuff.
Declarations for gdk_shadow_fb_update, gdk_shadow_fb_init,
gdk_shadow_fb_stop_updates, gdk_fb_recompute_all,
_gdk_fb_screen_angle. Removed CM, RP.
* gdk/linux-fb/gdkrender-fb.c:
Added code for shadowfb handling and screen rotation using
shadowfb.
* gdk/linux-fb/gdkwindow-fb.c:
Use fb_mem, fb_stride, fb_width, fb_height.
Added recompute_rowstride to reset the rowstride of all windows.
Added gdk_fb_recompute_all() which recomputes rootwindow size,
window abs positions and window rowstrides. Usefull when the
rotation has changed.
2001-01-09 Tor Lillqvist <tml@iki.fi>
* gdk/win32/gdkevents-win32.c: Implement better mouse
capturing. Do use SetCapture(). Thanks to jpe@archaeopteryx.com.
(find_window_for_pointer_event): New function.
(gdk_event_translate): Use it when handling mouse events.
2001-01-08 Alexander Larsson <alla@lysator.liu.se>
* gdk/linux-fb/gdkwindow-fb.c (_gdk_fb_window_find_focus,
gdk_fb_window_find_focus, gdk_fb_send_focus_change,
gdk_fb_window_send_crossing_events):
Possibly send focus events when the mouse moves from one
window to another.
2001-01-07 Tor Lillqvist <tml@iki.fi>
* gdk/win32/gdkwindow-win32.c (gdk_window_set_title,
gdk_window_new): Adapt calls to g_{locale,filename}_{to,from}_utf8
to match current API.
* gdk/win32/gdkvisual-win32.c (gdk_visual_init): If a secret
environment variable is set, don't call g_error when on a
palettized (PseudoColor) display. Some people want to use GTK+
in 256-colour mode even though works only partially.
* gdk/gdk.def
* gtk/gtk.def
* gtk/makefile.mingw.in: Update.
2001-01-05 Alexander Larsson <alexl@redhat.com>
* gdk/linux-fb/gdkdrawable-fb2.c (gdk_fb_drawable_finalize): free
colormap in here. Sync from X port.
(gdk_fb_set_colormap): allow the colormap to be NULL if you like.
Sync from X port.
(gdk_fb_clip_region): Comment out g_warning about empty clip region.
This happens now and then, and i don't think it is a bug.
(gdk_fb_draw_polygon): Correctly add extra point if startpoint
!= endpoint.
(gdk_fb_draw_lines): Don't draw dashed lines if line_style is
GDK_LINE_SOLID.
* gdk/linux-fb/gdkfont-fb.c: Return bogus GdkFont that has
some of the fields filled in at least.
* gdk/linux-fb/miwideline.c: The widths argument to miLineArcI, and
miLineArcD is no more, remove it from the function definition.
2001-01-03 Havoc Pennington <hp@redhat.com>
* gtk/gtktextbtree.c: Adapt to uscore-ification of gtktextiterprivate
* gtk/gtktextdisplay.c (gtk_text_layout_draw): remove use
of private functions; remove inclusion of private headers.
* gtk/gtktextlayout.c (gtk_text_layout_get_iter_at_line): Add this
function, so we don't need private functions in gtktextdisplay.c
* gtk/gtktextiterprivate.h: underscore-ification
* gtk/gtkwidget.c: Clean up a bunch of docs that said "INTERNAL"
to instead say "only useful to implement widgets"
* gtk/gtkenums.h (GtkMovementStep): Rename GTK_MOVEMENT_CHARS,
GTK_MOVEMENT_POSITIONS to GTK_MOVEMENT_LOGICAL_POSITIONS,
GTK_MOVEMENT_VISUAL_POSITIONS. Resolves bug 40249.
* gdk/x11/gdkwindow-x11.c (gdk_window_impl_x11_set_colormap):
This function was completely broken
* gtk/testtext.c (line_numbers_expose): use gtk_paint_layout
* gtk/gtkvscale.c (gtk_vscale_draw_value): use gtk_paint_layout
* gtk/gtkvruler.c (gtk_vruler_draw_ticks): use gtk_paint_layout
* gtk/gtklabel.c (gtk_label_expose): use gtk_paint_layout
* gtk/gtkhscale.c (gtk_hscale_draw_value): use gtk_paint_layout
* gtk/gtkhruler.c (gtk_hruler_draw_ticks): use gtk_paint_layout
* gtk/gtkcellrenderertext.c (gtk_cell_renderer_text_render): use
gtk_paint_layout
* gtk/gtkaccellabel.c (gtk_accel_label_expose_event): use
gtk_paint_layout
* gtk/gtkstyle.h: Add draw/paint expander to draw tree expanders.
Progress on bug 40103. Add draw_layout to draw a PangoLayout.
(struct _GtkStyleClass): Remove draw_cross, draw_oval, draw_ramp,
which were not implemented.
* gtk/gtktextbuffer.h (struct _GtkTextBufferClass): Add
insert_pixbuf signal. Rename delete_text to delete_range since it
also deletes pixbufs and child anchors. This almost closes bug
40245 (still need to deal with child anchors)
* gtk/gtktextbuffer.c (gtk_text_buffer_class_init): Add
insert_pixbuf, change signal names as appropriate, change types of
signals taking marks/tags to have the specific type, not just
G_TYPE_OBJECT
* gtk/gtkmain.c (gtk_get_current_event_state): Add this function;
closes bug 40051
* gtk/gtkoptionmenu.c (gtk_option_menu_set_history): clean up
unnecessary remove_contents() call
(gtk_option_menu_class_init): add a "changed" signal, closes
bug 40039
(gtk_option_menu_update_contents): emit "changed" if the
active menu item changes
* gtk/gdk-pixbuf-loader.c (gdk_pixbuf_loader_frame_done): fix bad
cast to GtkObject, reported by Jonas Borgstrom
(gdk_pixbuf_loader_finalize): don't close the loader on finalize;
we can't do stuff with side effects in finalize. Instead, spew a
warning if the loader isn't closed.
* gdk/x11/gdkdrawable-x11.c (gdk_drawable_impl_x11_finalize): free
colormap in here, non-X ports probably need to sync to this change
* gdk/x11/gdkdrawable-x11.c (gdk_x11_set_colormap): remove
assertion that colormap != NULL, you can set the colormap to NULL
if you like.
* Makefile.am: remove gtk-config-2.0
* configure.in: Use pkg-config to locate GLib. Remove
separated-out GMODULE_LIBS, GMODULE_CFLAGS; these were not used,
and the gmodule libs/cflags were in GLIB_LIBS GLIB_CFLAGS anyhow.
Use pkg-config to locate Pango. Output correct Pango libs to
gdk-2.0.pc.in. Fix test for FREETYPE_CONFIG (typo fix).
* Makefile.am (pkgconfig_DATA): install only target-specific pc
files
(install-data-local): symlink gtk+-2.0.pc and gdk-2.0.pc to the
X11 pc files
* gtk+-2.0.pc.in (Requires): require the GDK for the current target
unref from here
2000-12-30 Tor Lillqvist <tml@iki.fi>
* gdk/win32/gdkwindow-win32.c
(gdk_window_impl_win32_get_visible_region): New function, as in
the X11 backend.
(gdk_window_new, gdk_window_set_title): Use g_locale_from_utf8()
instead of duplicating the same code here.
* gdk/win32/gdkwin32.h (GDK_COLORMAP_PRIVATE_DATA): Use
GDK_COLORMAP macro.
* gdk/win32/gdkevents-win32.c (gdk_event_funcs): Use NULL as
destroy function.
(gdk_event_translate): Use GDK_COLORMAP_PRIVATE_DATA to correctly
access that.
* gtk/gtk.def: Update.
* gtk/makefile.mingw.in: Use glib-genmarshal from $(GLIB)/gobject.
2000-12-26 Tor Lillqvist <tml@iki.fi>
* gdk/win32/gdkvisual-win32.c (gdk_visual_init): Call g_error if
we are on a palettized display ("PseudoColor" in X11), as the code
for that doesn't work anyway, and never has.
2000-12-30 Tor Lillqvist <tml@iki.fi>
* gdk_pixbuf.def: Update.
2000-12-30 Havoc Pennington <hp@pobox.com>
* gdk/x11/gdkwindow-x11.c (gdk_window_new): Set the system
colormap on input only windows, to avoid special cases all over
the code for these windows.
Tue Dec 19 23:47:27 2000 Owen Taylor <otaylor@redhat.com>
* configure.in gdk/x11/Makefile.am: Check for FreeType
using freetype-config, since we can't rely on Xft
to pull in the libraries for us, or rely on it being
in the default include path first.
* gtk/gtkframe.c: Fix some problems where the gap wasn't
being drawn in the right place, especially for right-aligned
frame labels.
Thu Dec 14 21:02:59 2000 Owen Taylor <otaylor@redhat.com>
* gtk/gtkfontsel.c (gtk_font_selection_select_font): Update
the font_name entry as needed.
2000-12-18 Havoc Pennington <hp@redhat.com>
* gdk/gdkevents.c (gdk_event_get_state): wow, that implementation
was pretty non-working.
* gtk/testtext.c (fill_file_buffer): make this a bit more robust
by passing the length to g_utf8_validate().
2000-12-16 Havoc Pennington <hp@pobox.com>
* gtk/gdk-pixbuf-loader.h, gtk/gdk-pixbuf-loader.c:
Port to GObject, can go back in gdk-pixbuf after setting up
a gdk-pixbuf-marshal.h header over there.
* gtk/gtktreeview.c: s/SEPERATOR/SEPARATOR/g;
(gtk_tree_view_class_init): specify GTK_TYPE_ADJUSTMENT for signal
args
(gtk_tree_view_init): don't unset GTK_NO_WINDOW, it shouldn't be
set
(gtk_tree_view_realize_buttons): don't gtk_widget_show() buttons
here, do it when we create the buttons later
(gtk_tree_view_realize_buttons): add some g_return_if_fail
(gtk_tree_view_map): paranoia checks that column->button is shown
and unmapped
(gtk_tree_view_size_request): only request visible children.
Move header size calculation in here, for cleanliness, and
to maintain invariants for child widgets if we eventually
let users set different children inside the buttons
(gtk_tree_view_map_buttons): factor out code to map buttons,
since it was being called several times
(gtk_tree_view_size_allocate_buttons): move_resize the drag
windows instead of just moving them; their height may change
if we allow random widgets in there, or the theme changes.
(gtk_tree_view_size_allocate): move button size allocation
above emitting the scroll signals, to ensure a sane state when we
hit user code
(gtk_tree_view_button_release): remove queue_resize after
tree_view_set_size(), set_size() will handle any resize queuing
that's needed
(gtk_tree_view_focus_in): just queue a draw, don't fool with
draw_focus goo
(gtk_tree_view_focus): use gtk_get_current_event() and
gdk_event_get_state()
(gtk_tree_view_deleted): don't queue_resize() after calling set_size()
(gtk_tree_view_build_tree): fix a "if (foo); {}" bug - i.e. remove
semicolon
(gtk_tree_view_create_button): show the button here
(gtk_tree_view_button_clicked): actually emit the clicked signal
on the column
(_gtk_tree_view_set_size): return right away if the size is
unchanged, as a cheesy optimization
(gtk_tree_view_setup_model): rename set_model_realized to
setup_model to match the flag that indicates whether we've
called it
(gtk_tree_view_get_hadjustment): create adjustment if it doesn't
exist, because set_scroll_adjustment does that and it shouldn't
matter what order you call these in
(gtk_tree_view_get_vadjustment): ditto
(gtk_tree_view_set_headers_visible): canonicalize the bool,
for paranoia
(gtk_tree_view_set_headers_visible): call
gtk_tree_view_map_buttons() instead of using cut-and-paste code
(gtk_tree_view_append_column): clarify whether the return value
is the count of columns before or after, and do the increment
separately from the return statement so you can tell from the code.
(gtk_tree_view_remove_column): ditto
(gtk_tree_view_insert_column): ditto
(gtk_tree_view_get_column): remove g_return_if_fail for columns
outside the existing range, the docs say that outside-range
columns are allowed, so we handle them as documented. (Presumably
this allows a nice loop with column != NULL as test.)
(gtk_tree_view_move_to): document what 0.0, 0.5, 1.0 alignments
mean (left/right/center etc.).
(gtk_tree_view_collapse_all): only queue a draw if we're mapped
(gtk_tree_view_expand_row): add docs
(gtk_tree_view_collapse_row): add docs
* gtk/gtktreeviewcolumn.c (gtk_tree_view_column_clicked): new
function to emit the clicked signal on a column
* gdk/gdkevents.c (gdk_event_get_state): new function, to get the
state of an event
(gdk_event_get_time): don't treat GDK_SCROLL
as a button event, remove default case from switch so gcc
will whine if we don't explicitly handle all event types
* gtk/gtktreeselection.h: added some FIXME
* gtk/gtktreeprivate.h (struct _GtkTreeViewPrivate): rename
"columns" to "n_columns" and "column" to "columns" for clarity
Thu Dec 14 20:22:31 2000 Owen Taylor <otaylor@redhat.com>
* gdk/{gdkdrawable.[ch],gdkpixmap.c,gdkwindow.c,x11/gdkwindow.c}:
Add two virtualized functions gdk_drawable_get_clip_region - to
get the clip region when drawing.
* gdk/gdkwindow.c (gdk_window_invalidate_rect): Rewrite to simple
use invalidate_region.
* gdk/gdkwindow.c (gdk_window_invalidate_region): Clip to visible
region.
* acconfig.h configure.in: Check for Xft. For now, assume
that if Xft is found, Pango was compiled with Xft support
as well.
* gdk/gdkcolor.h gdk/x11/gdkcolor-x11.c: Add
gdk_colormap_query_color().
* gdk/x11/gdkdrawable-x11.c (gdk_x11_draw_glyphs): Draw
with Xft if appropriate.
* gdk/x11/gdkpango-x11.c (gdk_pango_context_get): Create
a pangoxft context if we have XFT and the environment
variable GD_USE_XFT is set.
* gdk/x11/gdkx.h (struct _GdkGCX11): Cache the fg_pixel
and also possibly an XftDraw structure.
* gtk/gtkfontsel.c: Handle the case where the font from the
style doesn't match any of the fonts a bit better.
* gtk/testgtk.c: Add tabs between directional segments for
hebrew/arabic test. (Not really necessary, just a little
prettier.)
2000-12-15 Tor Lillqvist <tml@iki.fi>
* gdk/makefile.{mingw.in,msc} (gdk_OBJECTS): Add gdkkeys.
* gtk/gtk.def
* gdk/gdk.def: Update.
* gdk/win32/gdkmain-win32.c
* gdk/win32/gdkkeys-win32.c: New file. Move some functions from
gdkmain-win32.c here.
* gdk/win32/makefile.mingw.in (all): No need to make
gdk-win32res.o here, the makefile one step up will call us to make
it.
* gdk/win32/makefile.{mingw.in,msc} (gdk_win32_OBJECTS): Add
gdkkeys-win32.
* gdk/win32/gdkcolor-win32.c: Silence gcc -Wall.
* gdk/win32/gdkevents-win32.c (gdk_events_init): Remove leftover
call to g_source_add().
* gdk/win32/gdkgc-win32.c (BitmapToRegion): Plug memory
leak. Thanks to Andreas Kemnade.
