Tim Janik [Thu, 25 Feb 1999 14:08:50 +0000 (14:08 +0000)]
still request the extra width or height if the corresponding scrollbar is
Thu Feb 25 13:38:38 1999 Tim Janik <timj@gtk.org>
* gtk/gtkscrolledwindow.c (gtk_scrolled_window_size_request): still
request the extra width or height if the corresponding scrollbar is
already visible (the change from 09:50 needs to apply to policy ==
GTK_POLICY_AUTOMATIC only).
Tim Janik [Thu, 25 Feb 1999 08:58:23 +0000 (08:58 +0000)]
don't add extra width or height if the requested size is mandatory (i.e.
Thu Feb 25 09:50:38 1999 Tim Janik <timj@gtk.org>
* gtk/gtkscrolledwindow.c (gtk_scrolled_window_size_request): don't add
extra width or height if the requested size is mandatory (i.e. set
through child->usize).
Owen Taylor [Thu, 25 Feb 1999 04:06:15 +0000 (04:06 +0000)]
Make prototypes for drag_data_get and drag_data_received handlers match
Wed Feb 24 23:12:09 1999 Owen Taylor <otaylor@redhat.com>
* gtk/gtkwidget.h gtk/gtkclist.c: Make prototypes
for drag_data_get and drag_data_received handlers
match the types used for gtk_singal_new; time needs
to be a uint, not a uint32.
Tim Janik [Wed, 24 Feb 1999 23:34:14 +0000 (23:34 +0000)]
added function to check if a specific signal handler connection exists
Thu Feb 24 23:58:44 1999 Tim Janik <timj@gtk.org>
* gtk/gtksignal.c (gtk_signal_handler_pending_by_id): added function
to check if a specific signal handler connection exists (similar to
gtk_signal_handler_pending_by_func). and yes, this is a neccessary
addition, since without this function, signal handlers can not be
(un)blocked or disconnected in a generic fashion without producing
g_warnings().
made gtk_signal_get_handlers() an inline function since it's used in
a lot of places, especially within signal emission.
Tim Janik [Wed, 24 Feb 1999 18:47:49 +0000 (18:47 +0000)]
if the widget is not editable, don't silently eat up GDK_Return, GDK_Home
Wed Feb 24 19:35:15 1999 Tim Janik <timj@gtk.org>
* gtk/gtktext.c (gtk_text_key_press): if the widget is not editable,
don't silently eat up GDK_Return, GDK_Home and GDK_End, return FALSE
instead so the user can still hook up key bindings.
Tim Janik [Wed, 24 Feb 1999 16:29:47 +0000 (16:29 +0000)]
warn about &widget->requisition, even for toplevels.
Wed Feb 24 11:19:12 1999 Tim Janik <timj@gtk.org>
* gtk/gtkwidget.c (gtk_widget_size_request): warn about
&widget->requisition, even for toplevels.
* gtk/gtkwindow.c (gtk_window_move_resize): request size into an
automatic variable and check if our size has changed.
when we decide to reallocate everything, we need to resize our
own GdkWindow as well.
* gtk/gtkmain.c:
* gtk/gtkobject.c: hack time, make --gtk-debug=objects work even
after gtk_type_init(), this is to "fix" gnome's arg parsing oddities.
* gtk/gtktypeutils.h: don't segfault in GTK_CHECK_CLASS_TYPE and
GTK_CHECK_TYPE upon NULL pointers.
* gtk/gtklayout.c: removed bogus prototypes of gtk_layout_gravity_works,
gtk_layout_set_static_gravity, gtk_layout_add_child_cb and
gtk_layout_remove_child_cb to shut up compiler.
* gtk/Makefile.am: added an empty rule
gtkrc:
:
so make doesn't try to "update" gtkrc from gtkrc.c.
Tim Janik [Tue, 23 Feb 1999 19:42:53 +0000 (19:42 +0000)]
fix implicit branch creation, we need the *real* parent path for this and
Tue Feb 23 17:52:37 1999 Tim Janik <timj@gtk.org>
* gtk/gtkitemfactory.c (gtk_item_factory_create_item): fix implicit
branch creation, we need the *real* parent path for this and
gtk_item_factory_parse_path() only returns the parent path with "_?"
stripped off.
Owen Taylor [Tue, 23 Feb 1999 00:43:26 +0000 (00:43 +0000)]
When sending events to windows with a proxy, set the window field in the
Mon Feb 22 19:06:30 1999 Owen Taylor <otaylor@redhat.com>
* gdk/gdkdnd.c (xdnd_send_*): When sending events to
windows with a proxy, set the window field in the
sent event to the original window, not to the proxy.
(This gives consistency with Motif, and also provides
information to the destination that it wouldn't have
otherwise).
* gtk/gtkdnd.c: Fixed some refcount leaks for windows.
* gdk/gdkdnd.c (xdnd_send_xevent): When sending
Xdnd event to the root window, set the event mask
to ButtonPressMask. Hopefully, this should give us
compatibility with kfm, if it uses the Qt style
of root window drops.
Owen Taylor [Sun, 21 Feb 1999 22:12:00 +0000 (22:12 +0000)]
Always set the max width or height to at least 1, since 0 triggers bugs in
Sun Feb 21 17:14:44 1999 Owen Taylor <otaylor@redhat.com>
* gdk/gdkwindow.c (gdk_window_set_geometry_hints): Always
set the max width or height to at least 1, since 0
triggers bugs in some window managers. (The window
"pops out" of the frame in fvwm)
Owen Taylor [Sun, 21 Feb 1999 19:25:53 +0000 (19:25 +0000)]
removed debugging statements about embedded XID's. They could have been
Sun Feb 21 14:21:46 1999 Owen Taylor <otaylor@redhat.com>
* gtk/gtkwindow.c: removed debugging statements
about embedded XID's. They could have been
moved to the new PLUGSOCKET debug key, but they weren't
really useful anyways.
* gdk/gdk.h: Added gdk_error_trap_push/pop to provide
a clean interface to the gdk_error_code functionality
of trapping X errors.
* gtk/gtkplug.c gtk/gtksocket.c: Add error
traps around accesses to foreign windows.
Tim Janik [Sat, 20 Feb 1999 20:11:23 +0000 (20:11 +0000)]
hm, set_modal doesn't really work with random gtk_grab_add/gtk_grab_remove
Sat Feb 20 20:59:25 1999 Tim Janik <timj@gtk.org>
* gtk/gtkwindow.c: hm, set_modal doesn't really work with random
gtk_grab_add/gtk_grab_remove calls. either the window user uses
the set_modal interface or he places grab_add/grab_remove on his own.
(gtk_window_set_modal): enforce modality state either way.
(gtk_window_hide): revert previous change.
Tim Janik [Sat, 20 Feb 1999 19:09:21 +0000 (19:09 +0000)]
increment years and months by two not three on double clicks, patch
Sat Feb 20 20:04:48 1999 Tim Janik <timj@gtk.org>
* gtk/gtkcalendar.c (gtk_calendar_button_press): increment years and
months by two not three on double clicks, patch provided by Nick Lamb
<njl98r@ecs.soton.ac.uk>.
Tim Janik [Sat, 20 Feb 1999 19:00:53 +0000 (19:00 +0000)]
added ::modal argument. (gtk_window_hide): reset window->modal from the
Sat Feb 20 19:44:36 1999 Tim Janik <timj@gtk.org>
* gtk/gtkwindow.c: added ::modal argument.
(gtk_window_hide): reset window->modal from the current grab state
(this is only relevant when grab_add/grab_remove is called on the
window by bypassing gtk_window_set_modal).
(gtk_window_remove_embedded_xid):
(gtk_window_add_embedded_xid): don't use g_print() but g_message()
within #ifdef G_ENABLE_DEBUG. when the hack got these functions
added (ChangeLog doesn't know them)?
Owen Taylor [Thu, 18 Feb 1999 22:36:40 +0000 (22:36 +0000)]
When queueing clears relative to widget->window, actually store
Thu Feb 18 17:26:09 1999 Owen Taylor <otaylor@redhat.com>
* gtk/gtkwidget.c (gtk_widget_queue_clear_area):
When queueing clears relative to widget->window,
actually store widget->window so we know that
the coordinates are window-relative.
Owen Taylor [Thu, 18 Feb 1999 16:43:42 +0000 (16:43 +0000)]
Fix glib_config_args gtk_config_args confusion. (pointed out by Roland
Thu Feb 18 11:46:57 1999 Owen Taylor <otaylor@redhat.com>
* gtk.m4: Fix glib_config_args gtk_config_args
confusion. (pointed out by
Roland Dreier <droland@mail.math.okstate.edu>)
* gtk/gtksocket.c gtk/gtkmain.c gtk/gtkdebug.h
(gtk_socket_filter_func): Move the plug-socket
debugging code to a new "plugsocket" debug keys
from the previous hardcoded DPRINTF.
Tim Janik [Wed, 17 Feb 1999 23:31:34 +0000 (23:31 +0000)]
only translate widget relative coordinates to window relative if the
Thu Feb 18 00:27:46 1999 Tim Janik <timj@gtk.org>
* gtk/gtkwidget.c (gtk_widget_queue_clear_area): only translate widget
relative coordinates to window relative if the widget is not a toplevel
(i.e. it has ->parent != NULL), otherwise the area size gets screwed up
for toplevels.
Owen Taylor [Tue, 16 Feb 1999 04:52:21 +0000 (04:52 +0000)]
Use a ScrolledWindow test how Text widgets work within ScrolledWindow.
Mon Feb 15 23:55:27 1999 Owen Taylor <otaylor@redhat.com>
* gtk/testgtk.c (create_text): Use a ScrolledWindow
test how Text widgets work within ScrolledWindow.
(POLICY_AUTOMATIC drew very badly, so it uses
POLICY_ALWAYS for the vscrollbar)
* gtk/gtktext.c (gtk_text_unrealize): Free the cache
in the unrealize handler. Some code in gtktext.c
assumes that if text->line_start_cache is non-NULL,
the widget is realized. We'll recompute it again
anyways if we later re-realize.
* gtk/gtkhandlebox.c: Grab on widget->window,
not on handlebox->bin_window. This means we don't
have to worry about ungrabbing while reparenting.
Don't make a GTK+ grab - that wasn't necessary.
* gtk/gtkhandlebox.c: Only compute the root -> desk
offset once, when starting the drag, instead
of continuously. This should solve the *Bad Window*
errors that were occurring because of race conditions
with the WM.
* gtk/gtkhandlebox.c: Completely reworked the
the logic of reattachment. We introduce the
concept of the "snap edge", which is an edge
that needs to be aligned for reattachmed.
We set a reasonable default from
handlebox->handle_position, but an application
might want to change this in some circumstances.
If the focus widget is the same, but does not currently
have the focus, call gtk_window_set_focus(). This is needed
so that click focusing with GtkPlug works correctly.
Tim Janik [Wed, 10 Feb 1999 16:37:09 +0000 (16:37 +0000)]
applied patch from Paolo Molaro to fix GTK_TYPE_POINTER args of
Wed Feb 10 15:49:16 1999 Tim Janik <timj@gtk.org>
* gtk/gtkwindow.c (gtk_window_class_init):
* gtk/gtkctree.c (gtk_ctree_class_init): applied patch from Paolo
Molaro to fix GTK_TYPE_POINTER args of ::tree_collapse and
::set_focus signals.
* docs/gtk_tut.sgml: s/menufactory/itemfactory, minor fixups
to the item factory example.
Tim Janik [Wed, 10 Feb 1999 08:06:30 +0000 (08:06 +0000)]
asure that ->value_buf always points to an allocated area.
Wed Feb 10 08:35:36 1999 Tim Janik <timj@gtk.org>
* gtk/gtkcolorsel.c (gtk_color_selection_draw_value_bar): asure that
->value_buf always points to an allocated area.
(gtk_color_selection_drag_end): new function hooked up to drag_end to
destroy the drag_window.
* gtk/gtkdnd.c: some coding style fixups.
Wed Feb 10 06:47:17 1999 Tim Janik <timj@gtk.org>
* gdk/gdkwindow.c: various indentation fixes.
* gdk/gdkevents.c: coding style fixups.
NULL initialize event_func, event_notify and event_chunk.
(expose_predicate): made this function static.
(gdk_event_handler_set): perform destroy notification even if no
event_func is set.
Owen Taylor [Wed, 10 Feb 1999 02:35:09 +0000 (02:35 +0000)]
Fixed some bugs with set_default_size.
Sun Feb 7 19:49:21 1999 Owen Taylor <otaylor@redhat.com>
* gtk/gtkwindow.c (gtk_window_move_resize): Fixed some
bugs with set_default_size.
Sat Feb 6 13:23:51 1999 Owen Taylor <otaylor@redhat.com>
* docs/Changes-1.2.txt: Added information about
the change to gtk_widget_size_request().
* gtk/gtkentry.c: Call gtk_widget_get_child_requisition
explicitely since we differentiate between the usize
set by the user and what we got. (Ugh)
* gtk/gtkwidget.[ch] (gtk_widget_get_child_requisition):
New function to return the effective size of a widget
as it looks to its parent.
* gtk/gtkwidget.c (gtk_widget_size_request): Leave
widget->requisition set to exactly what the widget
asked for, and then make a copy of that into
the requisition argument. Allow a NULL requisition
argument, and, if G_ENABLE_DEBUG, warn if
requisition == &widget->requisition.
Lars Hamann [Mon, 8 Feb 1999 21:03:54 +0000 (21:03 +0000)]
changed to 64 (gtk_clist_draw_focus) (hadjustment_value_changed)
Mon Feb 8 21:00:20 1999 Lars Hamann <lars@gtk.org>
* gtk/gtkclist.c (CLIST_OPTIMUM_SIZE): changed to 64
(gtk_clist_draw_focus) (hadjustment_value_changed) (draw_row):
few fixes to respect GTK_CAN_FOCUS flag.
Owen Taylor [Fri, 5 Feb 1999 18:40:14 +0000 (18:40 +0000)]
Adjust saved positions of children when resizing a guffaw_gravity = TRUE
Fri Feb 5 13:23:50 1999 Owen Taylor <otaylor@redhat.com>
* gdk/gdkwindow.c (gdk_window_move_resize): Adjust
saved positions of children when resizing a
guffaw_gravity = TRUE window.
* gtk/gtkdnd.c (gtk_drag_highlight): Rework the
highlighting so that it is done in callbacks for
draw and expose. This should solve problems with
multiple highlights for the same GdkWindow.
* gtk/gtkdnd.c (gtk_drag_highlight): Fix bug where
the highlight was being drawn with an incorrect
width/height for !NO_WINDOW widgets.
Owen Taylor [Fri, 5 Feb 1999 15:48:25 +0000 (15:48 +0000)]
Don't assume any more that the gtkwidget.c code will queue a redraw on us
1999-02-05 Owen Taylor <otaylor@gtk.org>
* gtk/gtk[hv]paned.c (gtk_[hv]paned_size_allocate):
Don't assume any more that the gtkwidget.c code
will queue a redraw on us during a size_allocate -
queue clear/draws as appropriate for our groove.
Owen Taylor [Tue, 2 Feb 1999 18:40:47 +0000 (18:40 +0000)]
Make sure we set the icon window position before it is shown to prevent
Tue Feb 2 12:39:42 1999 Owen Taylor <otaylor@redhat.com>
* gtk/gtkdnd.c: Make sure we set the icon window position
before it is shown to prevent annoying flicking to 0,0.
Make it consistent that info->cur_x/y do not include
info->hot_x/y displacement.
Owen Taylor [Mon, 1 Feb 1999 21:05:32 +0000 (21:05 +0000)]
Cast uints to ints in geometry calculations. This should prevent FP
Mon Feb 1 14:54:28 1999 Owen Taylor <otaylor@redhat.com>
* gtk/gtklabel.c (gtk_label_expose): Cast uints
to ints in geometry calculations. This should
prevent FP exceptions on FreeBSD when quantities
become negative. [ Actually, this is probably
a workaround with a gcc bug with handling of
bitfields, according to Ben Jackson <ben@ben.com> ]