Mon Jul 2 16:53:25 2001 Owen Taylor <otaylor@redhat.com>
* gtk/xembed.h gtk/gtkplug.c gtk/gtksocket.c: For
XEMBED embedding add a _XEMBED_INFO property to the
client with version number and a "mapped" flags.
Use the mapped flag instead of the racy MapRequestEvent
* gtk/gtksocket.c: Clean up the gtk_socket_steal()
code to reliably set things (when the child is a passive
embedder participating in the XEMBED protocol) intead
of just being a hack for embedding non-participating
programs. Fix various bugs and race conditions.
* gtk/gtksocket.[ch] gtk/gtkplug.[ch]: Make local embedding
work by simply making the GtkSocket the gtk parent
of the GtkPlug. Set a flag in this case and make
the GtkPlug work like a normal container by overriding
methods such as check_resize and "chaining past" GtkWindow
to GtkBin.
* gtk/gtkentry.c (gtk_entry_real_activate)
gtk/gtkmain.c (gtk_propagate_event):
Eliminate use of gtk_widget_get_ancestor (widget, GTK_TYPE_WINDOW).
* gtk/gtkwidget.c (gtk_widget_get_toplevel,
gtk_widget_get_ancestor):
Explain why gtk_widget_get_ancestor (widget, GTK_TYPE_WINDOW)
might not give the expected result and recommend
an alternative.
* tests/testsocket.c tests/testsocket_child.c
tests/testsocket_common.c tests/Makefile.am: Extended
to test different type of adding plugs to sockets
(local,active,passive), and to test mapping/unmapping
the plug.
* gdk/gdkwindow.c (_gdk_window_destroy_hierarchy): Don't
mark the window as destroyed until after we
called _gdk_windowing_window_destroy().
(_gdk_windowing_window_destroy() may use GDK functions
on the window.)
* gdk/x11/gdkinput.c: Remove the check for finalization -
devices can be finalized under some circumnstances.
* gdk/x11/gdkinput-x11.c (gdk_input_device_new): Fix
small problem with GDK_TYPE_DEVICE.
Thu Jun 7 18:25:42 2001 Jonathan Blandford <jrb@redhat.com>
* gtk/gtktreeselection.c: Now it's a GObject instead of a
GtkObject. The GtkTreeSelection::selection_changed signal is now
the GtkTreeSelection::changed signal.
* gtk/gtktreeview.c: Modified to deal with new GtkTreeSelection
object.
* tests/gtktree*.c: Modified to deal with new GtkTreeSelection
object.
Wed Apr 18 14:23:14 2001 Owen Taylor <otaylor@redhat.com>
* gdk/x11/gdkimage-x11.c (gdk_image_new): Try changing
mode on shm segments to 0600. We'll see who complains.
* gdk/gdkwindow.c (_gdk_window_destroy_hierarchy): Call
_gdk_windowing_window_destroy() AFTER recursing through
children.
* tests/Makefile.am (noinst_PROGRAMS): Build testsocket,
testsocket_child on X.
* tests/testsocket[_child].c: Fix uses of
gtk_window_get_default_accel_group().
[ Merge patch from Ramiro Estrugo <ramiro@eazel.com> from gtk-1-2 ]
* gdk/gdkimage.c: (gdk_image_get):
Deal with the possibility that XGetImage() might return NULL.
Allocate the GdkImagePrivate structure only after XGetImage()
succeeds in order not to dereference a NULL ximage pointer. This
prevents a core dump when XGetImage() fails - which is unlikely,
but can happen due to race conditions accessing the geometries of
drawables. An x error will still be triggered, but the gdk image
wrapper at least wont seg fault.
Sat Mar 3 13:21:27 2001 Owen Taylor <otaylor@redhat.com>
* tests/Makefile.am tests/testsocket.c tests/testsocket_child.c:
Add plug-socket tests. (Won't work worth a darn until I commit
the rest of the plug-socket changes, but no reason these
can't go into CVS now.)
