Wed Jan 27 22:59:07 1999 Owen Taylor <otaylor@redhat.com>
* gtk/Makefile.am (static_SOURCES): Move gtkintl.h
from $source_headers to $static_sources since
it should not be installed.
Wed Jan 27 22:51:23 1999 Owen Taylor <otaylor@redhat.com>
* gdk/gdkevents.c (expose_predicate): Allow exposure
compression across GravityNotify events to improve
the appearance of a GTK+ window unshading.
Wed Jan 27 21:56:07 1999 Owen Taylor <otaylor@redhat.com>
[ patch from Takashi Matsuda <matsu@arch.comp.kyutech.ac.jp> ]
* acconfig.h configure.in gdk/gdkim.c: Add configure
flag for disabling XRegisterIMInstantiateCallback(), default
it to off for Solaris. Always try to open the
input method immediately, and only if that fails,
register than instantiate callback.
* gdk/gdkentry.c: Some tweaks to make sure that we
only get and set the IC attributes when appropriate.
Wed Jan 27 20:08:54 1999 Owen Taylor <otaylor@redhat.com>
* gtk/gtkfilesel.c (open_user_dir): Patch from
Sebastian Wilhelmi <wilhelmi@ira.uka.de> to
use g_get_home_dir() instead of calling getpwuid()
ourselves.
Wed Jan 27 18:57:57 1999 Owen Taylor <otaylor@redhat.com>
* gdk/gdk.h gdk/gdkdnd.c: Change signatures for
gdk_drag_begin() and gdk_drag_motion() so that the set
of possible actions is passed on each motion, not just at
the beginning of the drag. We do this so that
we can restrict the set of possible drag events
when the user presses a modifier key during a
drag.
* gdk/gdkdnd.c: Send a motif Operation-changed
message when the set of possible actions change,
as well as when the suggested action change.
* gdk/gdkdnd.c: Change the XdndActionList whenever
the set of actions change.
* gdk/gdkdnd.c: Add a filter to catch changes to
the XdndActionList property for the source widget.
* gtk/gtkdnd.c: Change the set of possible actions
when the user presses Control, Shift, or Control-Shift,
to only include the corresponding action.
* gtk/gtkdnd.c (gtk_drag_button_release_cb):
Disconnect button signals before we possibly
free info structure.
Wed Jan 27 18:40:50 1999 Owen Taylor <otaylor@redhat.com>
* gtk/gtkmain.c (gtk_main_do_event): Hack scrollwheel
support into the eventloop, by treating button 4-5
presses on scrolled windows and scrollbars specially.
Put it here to keep the ugliness small and confined.
(We need to intercept the normal propagation in any case
to have this work reliably right now, since some
subwidgets will trap the button presses)
* gtk/gtkviewport.c (gtk_viewport_realize): Select for
BUTTON_PRESS to support the above hack.
Tue Jan 26 14:02:50 1999 Owen Taylor <otaylor@redhat.com>
* gdk/gdkwindow.c (gdk_window_show/hide): Remove TimJ's
checks for already being mapped or hidden. See
the ICCCM's section on state changes for why this
is a bad idea.
Wed Jan 27 17:05:40 1999 Owen Taylor <otaylor@redhat.com>
* Separate out GLIB_LIBS and GLIB_DEPLIBS. This
may fix the --with-glib option. GLIB_LIBS are
the libs to link example programs against, while
GLIB_DEPLIBS are dependency libraries.
Wed Jan 20 11:19:00 1999 Owen Taylor <otaylor@redhat.com>
* gtk/gtklabel.c: Use floor() instead of truncating
to integer values so we get translation invariance.
* gtk/gtklayout.c (gtk_layout_size_allocate): Set upper
and lower values for adjustments in size_allocate().
* gdk/gdkwindow.c gdk/gdk.h gdk/gdkprivate.h: New
function gdk_window_set_static_gravities() to set
up a window for guffaw scrolling.
* gdk/gdkwindow.c (gdk_window_internal_destroy): Set flags
indicating destroyed state before cleanup.
* gtk/gtkprivate.h gtk/gtkwidget.c: Add a new
private flag IS_OFFSCREEN. If set, this indicates
to GTK+ that the widget is not to be considered
viewable regardless of its map state. Queued draws
on offscreen widgets are suppressed.
Added new function static gtk_widget_is_offscreen() to
check this flag on a widget and its ancestors.
* gtk/gtklayout.[ch]: Major revisions.
- Use gdk_window_set_static_gravities to set static gravity
on all child windows, and thus avoid having to create a window
for NO_WINDOW children.
- Adjust allocations of children as we scroll them
so queued draws work correctly.
- Don't allocate our children directly in a put()
or move(); just queue a resize() like every other
widget.
* gtk/testgtk.c: Make the arrows on the scrollbars
work, create a larger and more demanding test.
Mon Jan 25 20:05:22 1999 Owen Taylor <otaylor@redhat.com>
* gtk/gtkwidget.c (gtk_widget_set_{visual,colormap}):
New functions to set the colormap or visual
of an existing widget. These functions should not
be called on a widget that has previosly been
realized.
Mon Jan 25 20:05:22 1999 Owen Taylor <otaylor@redhat.com>
* gtk/gtkwidget.c (gtk_widget_set_{visual,colormap}):
New functions to set the colormap or visual
of an existing widget. These functions should not
be called on a widget that has previosly been
realized.
* gtk/gtkcolorsel.c (gtk_color_selection_dialog_init):
Set the visual and colormap on the toplevel dialog,
then push them for the child widgets. As opposed
to push/pop in _new(), this way, things will work
properly with gtk_widget_new().
Mon Jan 25 19:41:56 1999 Owen Taylor <otaylor@redhat.com>
* gtk/gtknotebook.c (gtk_notebook_page_allocate): Queue
a redraw on the tab area when a tab changes size.
(Includes improvements from Lars Hamann)
Mon Jan 25 19:00:13 1999 Owen Taylor <otaylor@redhat.com>
* gtk/gtkcolorsel.c (gtk_color_selection_value_events):
Back out the last solution to updating problem. And
try a different one that is less dependent on
the state of the widget. Just redraw the value_marker
when it gets mapped. Somebody needs to analyze
this widget more carefully and figure out
Mon Jan 25 17:48:03 1999 Owen Taylor <otaylor@redhat.com>
* gtk/gtkcolorsel.c: Removed some GTK_WIDGET_DRAWABLE()
tests - we need to update the value bar's contents
even when the widget is unmapped.
Mon Jan 25 15:47:46 1999 Owen Taylor <otaylor@redhat.com>
* acinclude.m4 (ac_result): Don't define ENABLE_NLS
unless we have a working NLS. (This should keep
libintl.h from being included when it doesn't
exist).
Mon Jan 25 11:05:10 1999 George Lebl <jirka@5z.com>
* gtk/gtkbutton.c: set RECEIVES_DEFAULT on init, and a
button press on the widget doesn't grab default anymore
* gtk/gtkcheckbutton.c,gtk/gtkoptionmenu.c,gtk/gtkradiobutton.c:
unset RECEIVES_DEFAULT on init
* gtk/gtkwidget.[ch]: add a RECEIVES_DEFAULT flag to widget
flags
* gtk/gtkwindow.c,gtk/gtkplug.c: on pressing <enter>, activate
the focused widget if it is a RECEIVES_DEFAULT widget,
no matter what the current default_widget is
* gtk/gtkwindow.c: on focus change, migrate the HAS_DEFAULT
flag to CAN_DEFAULT widgets if they RECEIVE_DEFAULT as well
or migrate it back to the default_widget if the focus widget
doesn't "RECEIVE_DEFAULT"
Sun Jan 24 12:17:39 1999 Owen Taylor <otaylor@redhat.com>
* gdk/gdkcolor.c (gdk_colormap_real_destroy): Fix
reversed assertion.
(Pointed out by Austin Donnelly <austin@greenend.org.uk>)
* gtk/gtkruler.c (gtk_ruler_expose): Remove unecessary
and harmful call to gtk_paint_box(). We are going
to paint onto the backing store later anyways.
Sat Jan 23 11:15:39 1999 Owen Taylor <otaylor@gtk.org>
* gdk/gdkdnd.c (gdk_window_register_dnd): X expects
longs, not uint32 for XChangeProperty, when
format == 32.
* gdk/gdkdnd.c (motif_read_initiator_info): Byte-swap
the MotifInitiatorInfo structure if necessary.
* gdk/gdkdnd.c: When extracting and constructing
Motif DND messages, don't use data.s[], data.l[]
since on 64bit architectures, these won't necessarily
be in the same place with data.b[].
* gtk/testdnd.c (target_drag_drop): Return TRUE for
drag_drop within the drop zone.
* gtk/gtkdnd.c (gtk_drag_dest_handle_event): Only send
a XmTRANSFER_FAILURE if no drop site is found, not
on every drop!
Fri Jan 22 19:22:39 1999 Owen Taylor <otaylor@redhat.com>
* gtk/gtkeditable.c: Add flags field to
GtkTargetEntry table.
* gdk/gdkwindow.c (gdk_window_is_viewable): Treat foreign
windows as always viewable, since we can't track their
map-state reliably without speed problems. (Fixes
problem where updating didn't occur properly within
GtkPlug)
Sat Jan 23 13:41:00 1999 Jeff Garzik <jgarzik@pobox.com>
* gtk/gtkctree.c
(tree_delete): Don't search selection list twice.
(real_select_row): Remove extra ';', add default for clarity.
Sat Jan 23 01:44:43 1999 Lars Hamann <lars@gtk.org>
* gtk/gtkspinbutton.c (gtk_spin_button_get_value_as_int)
(gtk_spin_button_get_value_as_float): removed extra
gtk_spin_button_update calls again.
* gtk/gtkclist.c (gtk_clist_unmap): unblock dnd signal handler if
neccessary
* gtk/gtknotebook.c (gtk_notebook_button_press): grab_focus before
switch_page
(gtk_notebook_set_shape): don't use cur_page if it's NULL.
Fri Jan 22 07:33:44 1999 Tim Janik <timj@gtk.org>
* gtk/gtkvscale.c (gtk_vscale_pos_trough): take value_spacing into
account for GTK_POS_LEFT.
* gtk/gtkradiobutton.c (gtk_radio_button_set_group): use
gtk_toggle_button_set_active () to set the new state.
(gtk_radio_button_new): only _set_group if group != NULL.
(gtk_radio_button_clicked): reference the widget.
(gtk_radio_button_init): make radio buttons active upon
startup.
Thu Jan 21 17:13:12 1999 Jeff Garzik <jgarzik@pobox.com>
* gtk/gtkselection.c:
(gtk_selection_property_notify): Check args for NULL.
(gtk_selection_default_handler):
Replace manual list count with call to g_list_length.
Remove signedness of temp var 'count' to agree with usage.
Slight spacing adjustment for readability.
* gtk/gtkbox.c, gtk/gtklayout.c, gtk/gtkselection.c:
Init a few ptrs to NULL, to quiet warnings.
Thu Jan 21 16:03:02 1999 Owen Taylor <otaylor@redhat.com>
* gtk/gtktreeitem.c: Add a paint routine, fix up bugs
in drawing where if the expose area was contained
completely in the right side of the tree the background
wasn't redraw, etc.
Thu Jan 21 16:03:02 1999 Owen Taylor <otaylor@redhat.com>
* gtk/gtkbutton.c (gtk_button_expose): Code cleanup.
* gtk/gtktree.c: Add a paint routine, fix up bugs
in drawing where if the expose area was contained
completely in the right side of the tree the background
wasn't redraw, etc.
* gtk/gtkwidget.c (gtk_widget_clip_rect): Fix a bug
in clipping draw regions while propagating them
upwards.
Thu Jan 21 12:20:51 1999 Owen Taylor <otaylor@redhat.com>
* gtk/gtkcolorsel.c (gtk_color_selection_dialog_new):
Pushing and popping the visual/cmap in the init()
function does no good, because the window itself needs
to be create with that visual/cmap.
So do it in the new() function instead.
This will require people doing gtk_widget_new() to do
the push/pop themselves, but I don't see any way around
it other than digging in and tweaking the window
widget's colormap and visual ourselves, which
is highly non-kosher, but would be safe...
Thu Jan 21 16:04:35 1999 Timur Bakeyev <mc@bat.ru>
* gtk/gtkrc.c (gtk_rc_init): It was reported by Owen Taylor
<otaylor@redhat.com>, that LC_ALL produce unparsable garbage
on some glibc'ed systems and HP/UX. So, falling to LC_CTYPE.
Thu Jan 21 02:11:44 1999 Owen Taylor <otaylor@redhat.com>
* Set and read the XdndActionList property defined
in the Xdnd spec. We do this always, not just for
XdndActionAsk like the spec says, but if we don't
find it, we fall back to actions = suggested_action,
which is compatible with the language of the spec.
* gdk/gdkdnd.c (gdk_drag_context_unref):
Move not-null assertion before pointer de-ref.
* gdk/gdkevents.c (gdk_event_queue_find_first):
Fix infinite loop bug. Someone should look at this further to
make sure the events_pending tests are correct.
Thu Jan 21 00:44:47 1999 Owen Taylor <otaylor@redhat.com>
* gtk/gtkdnd.c (gtk_drag_dest_drop): Default handling
for drops should be keyed on GTK_DRAG_DEFAULT_DROP
not GTK_DRAG_DEFAULT_MOTION.
* gtk/gtkclist.c (real_remove_row):
Don't search entire row list twice for element being removed.
* gtk/gtkmenushell.c (gtk_menu_shell_insert):
Replaced hand-coded implementation of g_list_insert with
call to g_list_insert.
* gtk/gtktooltips.c
(gtk_tooltips_destroy_data, gtk_tooltips_layout_text):
tooltipsdata->row test not-null test now properly encompasses
both g_list_* calls.
Thu Jan 21 02:47:36 1999 Lars Hamann <lars@gtk.org>
* gtk/gtknotebook.c (gtk_notebook_focus): few cleanups.
(gtk_notebook_key_press): call select page function for
GDK_DOWN key in case of tab_pos == TOP. Changed other
positions accordingly.
Wed Jan 20 19:57:14 1999 Owen Taylor <otaylor@redhat.com>
* gtk/gtkdnd.c (gtk_drag_begin): Start off with the
cursor for "no drop" since if we'll get a status
if we are over a drag window, but none if we aren't
over one.
.
* gtk/gtkcombo.c (gtk_combo_entry_key_press):
Free 'nprefix', closing a memory leak.
Spaced out the code a bit to make it more readable.
* gtk/gtkhbox.[ch] (gtk_hbox_new),
gtk/gtkvbox.[ch] (gtk_vbox_new):
Change 'homogenous' arg from gint to gboolean.
* gtk/gtktogglebutton.[ch]:
New function gtk_toggle_button_get_active().
1999-01-20 Federico Mena Quintero <federico@nuclecu.unam.mx>
* gtk/gtkdnd.c (gtk_drag_get_event_action): Allow button 2 to act
as GDK_ACTION_ASK. Also, allow MOD1 to specify GDK_ACTION_ASK
only if that action is enabled in the actions argument.
* configure.in (gtklocaledir): New variable. Define to
$prefix/$DATADIRNAME/locale. This will point to the directory
where po/Makefile actually installs the message translation
tables.
* gtk/Makefile.am (INCLUDES): Define GTK_LOCALEDIR to
$(gtklocaledir) rather than $prefix/share/locale.
Under some systems (notably, BSDI), there is no LC_MESSAGES category in
locale.h. In this case it is good guess to relay on the value of LC_ALL
(in fact, LC_CTYPE will be more precise).
Wed Jan 20 13:06:51 1999 Owen Taylor <otaylor@redhat.com>
* gtk/gtktext.[ch] (clear_[focus_]area): Instead of drawing
a background pixmap in with gdk_draw_pixmap(), set
it as a tile pixmap for a GC. This should speed
up things considerably and also will hide a bug
where text->first_onscreen_ver_pixel is being
corrupted, causing the bg pixmap drawing code to generate
graphics exposes.
Tue Jan 19 22:15:10 1999 Lars Hamann <lars@gtk.org>
* gtk/gtklist.c
(gtk_list_button_press): only call grab_add if pointer_grab succeeds.
(gtk_list_unmap): remove pointer/widget grabs if needed.
(gtk_list_signal_focus_lost): removed.
(gtk_list_focus) (gtk_list_set_focus_child): set last_focus_child.
(gtk_list_remove_items_internal) (gtk_list_clear_items): unset
undo_focus_child if necessary. In case of SELECTION_BROWSE/EXTENDED
select a new item if selection is empty.
