Sat Sep 26 04:46:22 1998 Tim Janik <timj@gtk.org>
* configure.in: set binary age and interface age to 0.
* gtk/gtkspinbutton.h:
* gtk/gtkspinbutton.c: changed gtk_spin_button_construct to
gtk_spin_button_configure. gtk_spin_button_construct is still defined
as an alias for gtk_spin_button_configure for source compatibility,
but actuall use is deprecated.
typeness corrections, i.e. gint->gboolean and gint->guint.
added object arguments: GtkSpinButton::value, GtkSpinButton::wrap,
GtkSpinButton::shadow_type, GtkSpinButton::update_policy,
GtkSpinButton::numeric, GtkSpinButton::snap_to_ticks,
GtkSpinButton::adjustment, GtkSpinButton::digits and
GtkSpinButton::climb_rate.
GtkType and macro fixups.
Sat Sep 25 23:33:55 1998 Tim Janik <timj@gtk.org>
* gdk/gdkwindow.c (gdk_window_get_root_origin): new function to get
the *real* geometry position of a window, taken possible window
manager offsets into account.
this has been succesfully tested with fvwm, fvwm-2, bowman, olwm,
olvwm, twm, ctwm, mlvwm, windowmaker and enlightenment.
it does fail though for amiwm which adds windows to a pseudo root
window, and for icewm by a small offset because it defines the
geometry position whithin its border.
* gtk/testgtk.c: added "saved position" test to figure how
gdk_window_get_root_origin() interacts with window managers (repopup
this window to figure ;).
Wed Sep 23 06:09:19 1998 Tim Janik <timj@gtk.org>
* configure.in: update to Gtk+ version 1.1.3, binary age 1,
interface age 1.
* gtk/gtkobject.c
(gtk_object_ref):
(gtk_object_unref): check for ->ref_count to be greater than 0 (the
existing code was buggy in letting ->ref_count==0 pass through and
then failing the living_objs_ht assertion).
* gtk/gtkclist.c (draw_row): fix chopping bug for right justified
pixtext fields (Roman Pozlevich <roma@botik.ru>).
Mon Sep 21 07:44:30 1998 Tim Janik <timj@gtk.org>
* gtk/gtkobject.c (gtk_object_class_add_signals): reallocate
the required memory block, rather than allocating it newly from
scratch and doing a full-blown block copy on it.
Mon Sep 21 02:30:06 1998 Tim Janik <timj@gtk.org>
* NEWS file update for upcoming release of Gtk+ version 1.1.2,
binary age 0, interface age 0.
* gtk/gtkmain.c (gtk_init): enforce gtk-modules to always be resident.
Sun Sep 20 09:52:39 1998 Owen Taylor <otaylor@gtk.org>
* HACKING: indent other than GNU indent should be OK
now.
* gtk/genmarshal.pl gtk/Makefile.am: Call indent on file
instead of piping through indent. If indent wasn't
found in ./configure don't bother indenting at
all.
* gtk/Makefile.am (source_headers): Move `gtkfeature.h' ...
(gtkinclude_HEADERS): ... here.
(gtk.defs): Pass `g' flag to sed, to replace all occurances.
Tue Sep 15 14:57:30 1998 Owen Taylor <otaylor@redhat.com>
* Makefile.am gtk-config.in l*: Update to libtool-1.2b,
change library versioning scheme to drop LT_RELEASE
from the -l line, while keeping it in the soname.
Tue Sep 15 14:10:33 1998 Owen Taylor <otaylor@redhat.com>
* configure.in Makefile.in: define a MAINT-like variable REBUILD
which is set if Perl and awk are found, and autogenerated sources
can be rebuilt. Remove the auto-generated sources from
CVS.
Add dependencies so that the autogenerated sources are
rebuilt if the files they depend on change.
(This unfortunately currently results in the entire
gtk/ directory being rebuilt when any header changes.
Moving all enums to gtkenums.h would fix this.)
Thu Sep 17 05:12:48 1998 Tim Janik <timj@gtk.org>
* gtk/gtkobject.h: added gtk_object_set_data_destroy and
gtk_object_set_data_destroy_by_id functions to set the destroy
function pointer for an existing data portion.
* gtk/gtkobject.c: removed the GtkObjectData implementation and
made all gtk_object_*data* functions use the GLib equivalents.
1998-09-16 Federico Mena Quintero <federico@nuclecu.unam.mx>
* gtk/gtktoolbar.c (gtk_toolbar_insert_element): Test whether type
is not equal to GTK_TOOLBAR_CHILD_RADIOBUTTON before checking that
(widget != NULL), otherwise radio button insertion will not work
-- the widget argument is NULL for the first radio button in the
group.
Tue Sep 15 13:09:24 1998 Owen Taylor <otaylor@redhat.com>
* gdk/gdkimage.c (gdk_image_new): Reverted global changes
of g_new to malloc(), added back in the one place where
it actually mattered. #include <stdlib.h>.
* gdk/gdkrgb.c: g_malloc() to malloc() as above.
Tue Sep 15 11:06:13 1998 Owen Taylor <otaylor@redhat.com>
* gtk/gtkwindow.c (gtk_window_move_resize): Don't request
a new size from the WM if we asked for the same size
before and our request was rejected.
* gtk/gtkwindow.c (gtk_window_configure_event): If
in response to a request, we get a configure event
with unchanged dimensions, send a synthetic expose
since ForgetGravity won't expose our window.
Mon Sep 14 14:28:24 1998 Owen Taylor <otaylor@redhat.com>
* gtk/gtktext.c (adjust_adj): Change the way we
set the adjustments on resize so that it tries to
keep the beginning of the text in approximately
the same place.
Removed assertion that the above change made invalid.
Mon Sep 14 21:47:35 CDT 1998 Shawn T. Amundson <amundson@gtk.org>
* gdk/gdkimage.c: changed g_new()s to malloc()s and
g_free()s to free()s to avoid mixing g_new()s with
free()s. X itself uses free() on some of the
malloced memory.
* gtkfilesel.c: Maintain a list of directories like
/afs we know contain only directories, and avoid
stat'ing files in those directories. (Because
stat'ing all files in /afs is extremely expensive)
To support automounters, try to open directories,
even if we couldn't find them when reading their
parent directory.
* gtk/gtkclist.c (gtk_clist_set_selectable): new function
to set/unset the selectable flag of a single row.
(gtk_clist_get_selectable): new function
to get the state of the selectable flag
* gtk/gtkctree.c (gtk_ctree_node_set_selectable)
(gtk_ctree_node_get_selectable): clist analogons
* configure.in: added -posix check for NeXTStep
* gtk/testthreads.c: applied part of gtk-joke-980907-0 to use
a unique buffer for each thread.
* docs/gtk-config.1: clarified copyright
-Yosh
Tue Sep 8 05:19:51 1998 Tim Janik <timj@gtk.org>
* gdk/gdk.c (gdk_init): use g_atexit() instead of ATEXIT.
* gtk/gtkmain.c (gtk_init): use g_atexit() instead of ATEXIT.
* gtk/gtkobject.c (gtk_object_init_type): use g_atexit() instead of
ATEXIT.
* ltconfig: use GLib's ltconfig which honours lcc.
Sat Sep 5 16:01:19 1998 Owen Taylor <otaylor@gtk.org>
* gdk/gdkinputxfree.h gdk/gdk.c gdk/gdkinputgxi.h:
When the pointer is grabbed on an input window,
either explicitely, or through press-grab, and
then grabbed on a non-input window, ungrab the
devices.
Sat Sep 5 07:13:19 1998 Tim Janik <timj@gtk.org>
* gtk/gtktypeutils.h: don't include any gdk headers.
added structure definitions for GtkTypeObject and GtkTypeClass,
which cover the very basic fields of GtkObject and GtkObjectClass,
which are actually used by the type system.
* gtk/gtktypeutils.c: don't include gtkobject.h or gtkcontainer.h,
we only refer to our own structures. improved class inheritance
sanity checking a bit.
* gtk/gtkobject.h: placed appropriate comments at the beginnings
of the GtkObject and GtkObjectClass structures, which explain
their relationship to GtkTypeObject and GtkTypeClass.
* gtk/gtktypeutils.h:
* gtk/gtkobject.h:
moved GTK_STRUCT_OFFSET(), GTK_CHECK_CAST(), GTK_CHECK_CLASS_CAST(),
GTK_CHECK_TYPE() and GTK_CHECK_CLASS_TYPE() macro definitions from
gtkobject.h to gtktypeutils.h.
* gtk/gtkobject.h:
* gtk/gtkobject.c:
removed gtk_object_check_cast() and gtk_object_check_class_cast()
functions. to keep source compatibility, we can provide macro aliases
for the corresponding gtk_type_* functions. but then again, people
shouldn't have been using these functions anyways since they were part
of Gtk's internal API.
* gtk/gtktypeutils.h:
* gtk/gtktypeutils.c:
implemented gtk_type_check_object_cast() and gtk_type_check_class_cast()
functions to check GtkTypeObject and GtkTypeClass types.
Thu Sep 3 04:22:20 1998 Tim Janik <timj@gtk.org>
* gtk/gtkmenushell.c (gtk_real_menu_shell_move_current):
GTK_MENU_DIR_CHILD: check for the existance of
menu_shell->active_menu_item before accessing its child.
GTK_MENU_DIR_PREV:
GTK_MENU_DIR_NEXT: if we haven't had an active item and still
don't, make a default selection.
Wed Sep 2 00:28:58 1998 Tim Janik <timj@gtk.org>
* gtk/gtkwidget.c (gtk_widget_propagate_state): iterate
the children with _forall for sensitivity changes and with
_foreach on pure state changes. this fixes a lot of the
old inclusions of internal widgets into _foreach calls.
* gtk/gtktree.c: removed gtk_tree_foreach, let gtk_tree_forall
do the work. don't walk the subtrees of first level children.
* gtk/gtktreeitem.c: provide a _forall implementation,
which walks the subtrees as well for include_internals.
* gtk/gtkmenuitem.c: provide a _forall implementation, which walks
the submenus as well for include_internals.
* gtk/gtkscrolledwindow.c: removed gtk_scrolled_window_foreach and
implemented gtk_scrolled_window_forall, which will iterate over
the viewport and the scrollbars for gtk_container_forall or
iterate over the viewports children for gtk_container_foreach.
* gtk/gtktoolbar.c:
* gtk/gtktable.c:
* gtk/gtkpaned.c:
* gtk/gtkpacker.c:
* gtk/gtkmenushell.c:
* gtk/gtklist.c:
* gtk/gtkfixed.c:
* gtk/gtkclist.c:
* gtk/gtkbox.c:
* gtk/gtkbin.c:
* gtk/gtknotebook.c:
removed the old gtk_*_foreach functions and provided gtk_*_forall.
* gtk/gtknotebook.c:
(gtk_notebook_real_switch_page): expose tabs.
(gtk_notebook_page_num): new function to return the page number
of a distinct child.
(gtk_notebook_focus): minor fixups. foxus handling is still screwed
under some circumstances.
* gtk/gtktreeitem.c:
(gtk_real_tree_item_select):
(gtk_real_tree_item_deselect): major fixes.
some general fixups wrt queue_redraw, and tree items not being
NO_WINDOW widgets.
* gtk/gtklistitem.c:
(gtk_real_list_item_select):
(gtk_real_list_item_deselect):
(gtk_real_list_item_toggle):
removed unneccessary queue_redraw calls.
Wed Aug 30 09:42:07 1998 Tim Janik <timj@gtk.org>
* gtk/gtkoptionmenu.c: allow optionmenus to have the focus and
automatically popup the menu on space bar.
Wed Aug 26 06:40:34 1998 Tim Janik <timj@gtk.org>
* gtk/gtkcontainer.h:
* gtk/gtkcontainer.c: implemented gtk_container_forall() (as a class
method), which acts similar to gtk_container_foreach(), but iterates
over internal children. the GtkContainer::foreach signal vanished in
favour of a new class method ->forall() that optionally includes
internal widgets.
* gtk/gtkclist.c (gtk_clist_init): provide no _foreach implementation
but a _forall implementation, since all child widgets we have are
internal ones.
(column_button_create): set the parent window prior
to gtk_widget_set_parent().
* gtk/gtkwidget.c:
exchanged all calls to gtk_container_foreach() with
gtk_container_forall().
* gtk/gtkwidget.h:
* gtk/gtkwidget.c: added the GTK_COMPOSITE_CHILD, exported through
the GtkWidget::composite_child argument. to have a widget created
with the flag initially, two new functions got added to wrap a widgets
creation:
gtk_widget_push_composite_flag() and gtk_widget_pop_composite_flag().
Wed Aug 25 23:37:39 1998 Tim Janik <timj@gtk.org>
* gtk/gtktooltips.h:
* gtk/gtktooltips.c: exported gtk_tooltips_create_window() as
gtk_tooltips_force_window(), so tooltips->tip_window can be accessed
prior to the first tip being set.
don't put an extra reference on the window, since it is a toplevel,
it wont get destroyed from anywhere else.
* overall macro and GtkType fixups.
1998-09-02 Federico Mena Quintero <federico@nuclecu.unam.mx>
* gtk/gtkcolorsel.c: Replaced "foo == TRUE" and "foo == FALSE"
constructs by more forgiving ones. (Does anyone else feel like
the color selector need rewriting?).
Tue Sep 1 10:53:52 1998 Owen Taylor <otaylor@redhat.com>
* gtk/gtkprogress.c (gtk_progress_finalize): Unref
the adjustment.
* gtk/gtkmenu.c: Modified reference counting so
that the reference count from the toplevel that
the menu creates to the menu is invisible to the
application.
Tue Aug 25 16:52:47 1998 Owen Taylor <otaylor@redhat.com>
* gtk/gtktext.c (recompute_geometry): Make sure the
mark we pass to line start is the beginning of a
real line, not a wrapped one.
Mon Aug 24 18:37:15 1998 Owen Taylor <otaylor@redhat.com>
* gtk/gtktext.c: Reference count the fonts used in
the text widget.
* gdk/gdk.h gdk/gdkcolor.c: Rename the color allocation
freeing functions to be more consistent, and more
convenient; leave the old names in for backwards compatibility.
* gdk/gdkcolor.c gdk/gdkprivate.h: Reference count the
allocations in pseudo-color colormaps to greatly reduce
calls to XAllocColor. Keep a per-colormap hashtable to
speed up finding if there is an already-allocated matching
color.
* gdk/gdkcolor.c: Don't just match read the system colormap
when the colormap is created, but synchronize our copy
with the system colormap periodically.
* gdk/gdk.c gdk/gdktypes.h gtk/gtkentry.c gtk/gtktext.c:
Change XIM constants names to match GDK conventions
* gtk/testinput.c: Allow the drawing area to get the focus.
* gtk/testgtk.c: Change around the Text test to demonstrates
multiple fonts, use more colors.
* gtk/gtkwidget.c: Improve gtk_widget_get_colormap()/visual()
so they work after a widget is unrealized.
* gtk/gtktext.[ch]: Remove the requirement that the text
widget be realized before adding text (!) Allocate colors
ourself, instead of requiring the caller allocate them.
Allow changing styles to work properly by keeping track
of the values for a certain property are default or
set explicitely.
* gtk/gtkmenu.h: Added some comments.
* gtk/gtkentry.c: Changes to match XIM constants.
* gtk/gdk.h gdk/gdkwindow.c: Add gdk_drawable_set_data(),
for adding keyed data to drawables. (Uses g_dataset
internally)
* gdk/gdkpixmap.c: Keep track of the colors we allocate,
when creating an XPM - store them as user data for the GdkPixmap,
so we don't leak colors when we create pixmaps from XPM's.
Allocate memory for color information in large blocks instead of
as many little pieces.
Mon Aug 24 02:36:53 1998 Tim Janik <timj@gtk.org>
* gtk/gtkarg.h:
* gtk/gtkarg.c:
* gtk/gtkwidget.h:
* gtk/gtkwidget.c:
* gtk/gtkobject.h:
* gtk/gtkobject.c:
* gtk/gtkcontainer.h:
* gtk/gtkcontainer.c:
(gtk_object_new):
(gtk_object_set):
(gtk_widget_new):
(gtk_widget_set):
(gtk_container_add_with_args):
(gtk_container_child_set):
begin the parameter elipsis (...) after the first argument name. this
change is source compatible, since it was always required, even as NULL.
(gtk_args_collect):
(gtk_object_args_collect):
(gtk_container_child_args_collect):
changed prototypes to pass first_arg_name, also, pass va_list variable
by value (portability concerns). callers changed.
* gtk/gtkargcollector.c: implemented gtk_arg_collect_value() as
a huge macro GTK_ARG_COLLECT_VALUE() <shrug>. this is needed because we
can't pass va_list variables by reference for portability reasons.
Fri Aug 21 02:31:26 1998 Tim Janik <timj@gtk.org>
* gtk/gtkmain.h:
* gtk/gtkmain.c (gtk_check_version): added a function to perform a
binary version check, returning a static string for error descriptions,
or NULL on success.
Thu Aug 20 20:08:15 1998 Owen Taylor <otaylor@redhat.com>
* gtk/gtktext.c ([un]draw_cursor): Check to see if we've
computed our geometry yet, and if not, ignore the request.
gtk/gtktext.c (correct_cache_insert): Fiddled around]
some more. All the bug-test codes seem to work at once
now. (fingers crossed...)
