George Lebl [Sun, 3 Sep 2000 05:49:34 +0000 (05:49 +0000)]
Add some documenting comments about GTK_CAN_DEFAULT, GTK_HAS_DEFAULT and
Sun Sep 03 00:47:08 2000 George Lebl <jirka@5z.com>
* gtk/gtkwidget.h: Add some documenting comments about
GTK_CAN_DEFAULT, GTK_HAS_DEFAULT and GTK_RECEIVES_DEFAULT
just as owen said on irc.
* gtk/gtkwindow.c (gtk_window_set_default): Don't blindly
set HAS_DEFAULT. Only set it if there isn't another window
holding it (a focus window which RECEIVES_DEFAULT) and only unset
it if the widget is not the focused window with RECEIVES_DEFAULT.
This fixes weird double default bugs in dialogs which has things
call gtk_window_set_default at times.
Owen Taylor [Sat, 2 Sep 2000 02:43:50 +0000 (02:43 +0000)]
Apply patch from Nils Barth and David Santiago to improve submenu
Fri Sep 1 22:39:07 2000 Owen Taylor <otaylor@redhat.com>
* gtk/gtkmenu.[ch] TODO.xml: Apply patch from
Nils Barth and David Santiago to improve submenu
navigation. The patch does this by creating a triangular
region from the point where the pointer leaves the
menu to the submenu. While the pointer is in
that region and a timeout has not expired, events
that would cause the active submenu to change are
ignored.
Owen Taylor [Fri, 1 Sep 2000 20:10:58 +0000 (20:10 +0000)]
Fix bug where moving a toplevel window immediately after sizing it would
Fri Sep 1 15:34:46 2000 Owen Taylor <otaylor@redhat.com>
* gdk/x11/gdkwindow-x11.c (gdk_window_move): Fix bug where
moving a toplevel window immediately after sizing it would
result in it being sized back to the old size since
impl->width, impl->height aren't updated until we hear
back from the server. There is still a (longstanding) bug
here that gdk_window_get_size() will give the wrong size
until we do hear back.
Havoc Pennington [Thu, 31 Aug 2000 04:20:59 +0000 (04:20 +0000)]
cache end line, for rapid testing whether an iterator is the end iterator
2000-08-31 Havoc Pennington <hp@pobox.com>
* gtk/gtktextbtree.c (GtkTextBTree): cache end line,
for rapid testing whether an iterator is the end iterator
(gtk_text_btree_new): init end line cache
(gtk_text_line_is_last): use fast cached line to
see if we're the last line.
(get_last_line): use cached line here too, should
speed up some random unrelated code.
Havoc Pennington [Mon, 28 Aug 2000 03:44:06 +0000 (03:44 +0000)]
wrote lots of API docs, minor indentation fixes, inserted some FIXME about
2000-08-27 Havoc Pennington <hp@pobox.com>
* gtk/gtktextiter.c, gtk/gtktextiter.h: wrote lots of API docs,
minor indentation fixes, inserted some FIXME about stuff I
realized was broken while documenting it.
(gtk_text_iter_get_pixmap): allow out params to be NULL, following
GTK convention
Elliot Lee [Wed, 23 Aug 2000 00:50:32 +0000 (00:50 +0000)]
Try to fix a bug where a window move, with a non-rectangular region
Try to fix a bug where a window move, with a non-rectangular region unobstructed, would cause
pieces of the window to be moved twice. Effectively, sort of implement a
gdk_draw_drawable_region(drawable, gc, drawable, region, dx, dy);
Am going blind trying to find the off-by-one thing, especially since it doesn't happen except for
the preview tests. Still not fixed, have at it.
Tor Lillqvist [Sat, 19 Aug 2000 21:46:05 +0000 (21:46 +0000)]
gdk/gdkimage.h No need any longer on Win32 for the shared memory
2000-08-20 Tor Lillqvist <tml@iki.fi>
* gdk/gdkimage.h
* gdk/win32/gdkimage-win32.c: No need any longer on Win32 for the
shared memory image+pixmap GdkImage type, or
gdk_image_bitmap_new(). They were used in the gdk_imlib port, but
I am dropping that.
* gtk/gtkmain.h: On Win32, use a #define to map gtk_init() to
actually call gtk_init_abi_check(), passing also
sizeof(GtkWindow). Ditto for gtk_init_check().
* gtk/gtk.def
* gtk/gtkmain.c: (gtk_init_abi_check, gtk_init_check_abi_check):
New functions, used to check that the GTK+-using code has been
compiled using the correct compiler and switches. In particular,
with gcc one has to use the -fnative-struct switch as GTK+ is
compiled with that.
Tor Lillqvist [Sat, 12 Aug 2000 07:25:55 +0000 (07:25 +0000)]
Add more calls to g_filename_{to,from}_utf8() to convert back and forth
2000-08-11 Tor Lillqvist <tml@iki.fi>
* gtk/gtkfilesel.c: Add more calls to g_filename_{to,from}_utf8()
to convert back and forth from UTF-8 and the "system" (or "current
locale") charset when necessary. Indentation/spacing cleanup.
Owen Taylor [Wed, 2 Aug 2000 19:07:00 +0000 (19:07 +0000)]
Move is_viewable() test further inside the function, since removing the
Wed Aug 2 14:57:38 2000 Owen Taylor <otaylor@redhat.com>
* gdk/gdkwindow.c (gdk_window_process_updates_internal): Move
is_viewable() test further inside the function, since removing the
window from the redraw list without clearing ->update_area
was also very broken.
Owen Taylor [Mon, 31 Jul 2000 22:59:29 +0000 (22:59 +0000)]
Fix problem where computation was depending on widget->allocation instead
Mon Jul 31 18:37:36 2000 Owen Taylor <otaylor@redhat.com>
* gtk/gtkentry.c (entry_adjust_scroll): Fix problem where computation
was depending on widget->allocation instead of the width of text_area,
which was incorrect for subclasses like GtkSpinButton.
* gdk/gdkpango.c (gdk_draw_layout): Fix a couple of typos
with width == 1 tests that should be width == -1.
Owen Taylor [Mon, 31 Jul 2000 20:48:38 +0000 (20:48 +0000)]
Replace broken, (broken, Elliot, every single window blank on X) custom
Mon Jul 31 16:45:48 2000 Owen Taylor <otaylor@redhat.com>
* gdk/gdkwindow.c (gdk_window_process_updates_internal): Replace broken,
(broken, Elliot, every single window blank on X) custom code with call
to gdk_window_is_viewable() which a) prevents code duplicate and
b) works.
Elliot Lee [Mon, 31 Jul 2000 17:18:36 +0000 (17:18 +0000)]
We really do need to include gdkfb.h in order to get the GDK_PARENT_ROOT()
* gtk/gtkcolorsel.c: We really do need to include gdkfb.h in order to get the GDK_PARENT_ROOT()
macro to work.
* gtk/gtkmarshal.list: Add NONE:STRING,POINTER
* gdk/gdkwindow.c: Don't send expose events, or invalidate pieces of, windows that aren't shown.
* gdk/linux-fb: Redraw-when-menu-hidden solved, plus a bunch of pointer grab stuff.
Tor Lillqvist [Sun, 30 Jul 2000 08:11:28 +0000 (08:11 +0000)]
Add makefile.mingw.in to EXTRA_DIST.
2000-07-30 Tor Lillqvist <tml@iki.fi>
* gdk/Makefile.am: Add makefile.mingw.in to EXTRA_DIST.
* gdk/win32/gdkevents-win32.c (gdk_events_init, gdk_events_queue):
Remove the g_pipe_readable_msg stuff. Not needed with new GLib
GIOChannel implementation.
* gdk/win32/gdkinput.c
* gdk/win32/gdkpango-win32.c: Update license name here, too.
* gdk/win32/gdkwin32id.c (gdk_win32_handle_table_lookup): Fix
definition to match prototype.
* gtk/makefile.mingw.in: Add libiconv CFLAGS and LIBS.
Elliot Lee [Fri, 28 Jul 2000 00:09:36 +0000 (00:09 +0000)]
Add gdk_pixbuf_loader_new_with_type function, to allow handling image
2000-07-27 Elliot Lee <sopwith@redhat.com>
* gtk/gdk-pixbuf-loader.[ch]: Add gdk_pixbuf_loader_new_with_type
function, to allow handling image formats that cannot be detected
by looking at content alone.
gdk-pixbuf/
2000-07-27 Elliot Lee <sopwith@redhat.com>
* gdk-pixbuf-io.h, gdk-pixbuf-io.c: Add gdk_pixbuf_get_named_module() function to facilitate above change
* io-bmp.c: Don't malloc a temporary buffer - use stack instead.
* io-wbmp.c, pixbufloader_wbmp.defs, gdk-pixbuf-io.c, Makefile: Implement loader for WBMP format.
**NOTE: I haven't tested this yet, I need to sync it to another machine, but it if it is broken
I'm already working on fixing it
Tim Janik [Wed, 26 Jul 2000 11:33:08 +0000 (11:33 +0000)]
applied patch from Andreas Persenius <ndap@swipnet.se> that updates the
Wed Jul 26 12:59:31 2000 Tim Janik <timj@gtk.org>
* *.[hc]: applied patch from Andreas Persenius <ndap@swipnet.se> that
updates the license headers to the GNU Lesser General Public License,
as well as updating the copyright year to 2000.
Robert Brady [Wed, 26 Jul 2000 00:55:42 +0000 (00:55 +0000)]
Cause GETTEXT_PACKAGE to be consistent in config.h and po/Makefile, was
2000-07-26 Robert Brady <rwb197@zepler.org>
* configure.in: Cause GETTEXT_PACKAGE to be consistent in config.h
and po/Makefile, was installing and looking for them under gtk20
and gtk2 respectively.
* gtk/testtext.c, gtk/testtextbuffer.c: update to reflect text
widget changes.
* gtk/gtktextview.h: To be consistent with usage of "line"
throughout the API to mean "newline-terminated thingy", change
MOVEMENT_LINE to be MOVEMENT_WRAPPED_LINE, and MOVEMENT_PARAGRAPH
to MOVEMENT_LINE.
(GtkTextView): Add flags for default editability, and whether to
show the cursor. Add functions to get/set that. Add
(gtk_text_view_get_iter_location): new function
* gtk/gtktexttypes.h: Move GtkTextLine typedef from here to
gtktextlayout.h
(g_convert): Add g_convert temporarily, will go in glib in a bit
* gtk/gtktexttagtable.h: include gtktexttag.h, and define
GtkTextTagTableForeach instead of brokenly using GHFunc.
Change gtk_text_tag_table_foreach() so it doesn't use GHFunc.
* gtk/gtktexttagprivate.h: Remove GtkTextStyleValues from here,
moved to public header.
* gtk/gtktexttag.h: Rename the "elide" attribute of tags to
"invisible", since "elide" was a bad name.
(gtk_text_tag_get_priority): Added
(GtkTextStyleValues): put this in public header, along with
functions to use it.
* gtk/gtktextmarkprivate.h: Include more headers, since we no
longer include gtktextbtree.h.
* gtk/gtktextlayout.h: Don't include the "really private" headers,
only buffer/iter. Forward declare GtkTextLIne and GtkTextLineData
to make this possible. Now we only need to install
gtktextlayout.h, not gtktextbtree.h and gtktext*private.h.
(However the Makefile.am isn't changed yet because of the
logistics of merging gtk-hp-patches piecemeal)
* gtk/gtktextiterprivate.h: include btree header, so it compiles;
rename gtk_text_iter_get_line to gtk_text_iter_get_text_line since
gtk_text_iter_get_line is now used in the public API for a
different purpose.
* gtk/gtktextiter.h: Clean up function names to be more
consistent. Always call char offset "offset" and byte index
"index". A "line" is always a line number.
(gtk_text_iter_is_last): new function, more efficient than
the existing way to check
(gtk_text_iter_is_first): new function, also more efficient
(gtk_text_iter_up_lines, gtk_text_iter_down_lines): Remove these
(gtk_text_iter_next_char, gtk_text_iter_prev_char): Renamed from
gtk_text_iter_forward_char, etc.
(gtk_text_iter_forward_to_tag_toggle): Renamed from
forward_find_tag_toggle, since this isn't a linear search
(GtkTextCharPredicate): rename from GtkTextViewCharPredicate
(gtk_text_iter_forward_search, gtk_text_iter_backward_search):
New functions, search for a buffer substring.
* gtk/gtktextbuffer.h: Add fields to store whether a paste is
interactive and default editable (since we need to store that info
until we receive the selection data).
Remove all the _at_char and at_line etc. versions of functions;
only have iterator versions.
Add _interactive() versions of functions, that consider the
editability of text. (FIXME add interactive flag to the
insert/delete signals per Darin's suggestion)
(gtk_text_buffer_get_tag_table): new function, demand-creates the
tag table if necessary
Remove declaration of gtk_text_buffer_get_iter_from_string
Implement new sane, 5-function API for using GtkProgressBar. See
2000-07-25 Havoc Pennington <hp@redhat.com>
* gtk/gtkprogressbar.h: Implement new sane, 5-function API for
using GtkProgressBar. See Changes-2.0.txt for details.
* gtk/gtkprogressbar.c: Add object arguments "fraction" and
"pulse_step" which are the equivalent of
gtk_progress_bar_set_pulse_step and gtk_progress_bar_set_fraction.
Implement new API.
* gtk/gtkprogress.h (struct _GtkProgress): Add a field
(use_text_format) to mark whether text set on the progress bar is
a format string. Deprecate entire GtkProgress interface.
* gtk/gtkprogress.c (gtk_progress_init): init use_text_format to TRUE
(gtk_progress_build_string): make this a no-op if use_text_format
is FALSE
* docs/Changes-2.0.txt: Describe progress bar changes.
Elliot Lee [Mon, 24 Jul 2000 16:19:00 +0000 (16:19 +0000)]
(Part 2) Remove gdk_*_lookup() defines, since they are defined by the
(Part 2)
* gdk/gdkprivate.h: Remove gdk_*_lookup() defines, since they are defined by the
individual backends already.
* gdk/gdkregion-generic.h, gdk/gdktypes.h: Put gdkregionbox & gdksegment back together
again. Yes, there really is a good reason for this, if you are using the gdkregion
internals, and if you're not, why do you care?
* gdk/gdkwindow.c: Fix inverted condition
* gdk/linux-fb: Compiles (for me - it will not work elsewhere most likely).
* gtk/gtkcolorsel.c, gtk/gtkwindow.c: Add include for linux-fb
* gtk/gtkrange.c: Redraw trough when moving.
* gtk/gtktypeutils.c: Fix warning by adding const cast.
* modules/linux-fb/basic.c: Fix unknown glyph retrieval.
* gdk/makefile.mingw.in
* gdk/win32/makefile.mingw.in: New files.
* gdk/Makefile.am
* gdk/win32/Makefile.am: Generate and distribute them.
* gdk/win32/gdkwindow-win32.h
* gdk/win32/gdkdrawable-win32.h
* gdk/win32/gdkpixmap-win32.h: New files.
* gdk/win32/gdkcolor-win32.c
* gdk/win32/gdkcursor-win32.c
* gdk/win32/gdkgc-win32.c
* gdk/win32/gdkimage-win32.c
* gdk/win32/gdkprivate-win32.h
* gdk/win32/gdkwin32.h: Update data structures. Cosmetics:
don't use fields and variables called xcursor, xid, etc (leftovers
from the X11 backend).
* gdk-pixbuf-io.c (gtk_win32_get_installation_directory):
New function, to get the GTK+ installation directory from
the Windows Registry, where the installation program
for whatever software includes GTK+ has stored it.
Used to avoid having hardcoding the directory where to look for
loaders. This function is needed by gtk, too, so it should
really be just in one place. Maybe a small static library
one level up from here?
* gdk-pixbuf-animation.c (gdk_pixbuf_animation_new_from_file)
* gdk-pixbuf-io.c (gdk_pixbuf_new_from_file):
Open file in binary mode. This *is* standard C. (No-op on Unix,
of course.)
* io-jpeg.c: If we don't HAVE_SIGSETJMP, use normal setjmp().
* io-tiff.c: Use g_get_tmp_dir(). If we don't HAVE_MKSTEMP, use
mktemp() and open().
(gdk_pixbuf__tiff_image_stop_load): g_free() also the tempname.
* pixops/*.S: The Gas from mingw32 doesn't like the .type
pseudo-op. Conditionalise on __MINGW32__, but probably
should conditionalise on Gas version instead? Or could we
do without .type on all systems?
* pixops/timescale.c: Use g_get_current_time()
instead of gettimeofday().
Owen Taylor [Fri, 21 Jul 2000 19:30:22 +0000 (19:30 +0000)]
Use new PangoAttrShape to reserve space for pixmaps, add GSList *pixmaps
Fri Jul 21 15:28:13 2000 Owen Taylor <otaylor@redhat.com>
* gtk/gtktextlayout.[ch], gtk/gtktextdisplay.c: Use new
PangoAttrShape to reserve space for pixmaps, add
GSList *pixmaps to GtkTextLineDisplay, draw the pixmaps
in the display functions.
* gtk/testgtk.c (create_layout): Set some more exotic
scrolled window options
* gtk/gtkscrolledwindow.[ch] (gtk_scrolled_window_set_shadow_type):
Add function to set the shadow type of a window, to allow
putting a shadow around widgets such as GtkLayout or GnomeCanvas
which don't draw their own frame.
Include files outside of the extern "C" block. Makes some C++ compiler
2000-07-19 Sebastian Wilhelmi <wilhelmi@ira.uka.de>
* gdk-pixbuf/gdk-pixbuf.h, gtk/gtkhsv.h, gtk/gtktextbtree.h,
gtk/gtktextbuffer.h, gtk/gtktextchild.h, gtk/gtktextdisplay.h,
gtk/gtktextiter.h, gtk/gtktextiterprivate.h, gtk/gtktextlayout.h,
gtk/gtktextmarkprivate.h, gtk/gtktextsegment.h, gtk/gtktexttag.h,
gtk/gtktexttagtable.h, gtk/gtktexttypes.h, gtk/gtktextview.h,
gtk/gtktypeutils.h: Include files outside of the extern "C"
block. Makes some C++ compiler happy. Reported by Denis Vakatov
<vakatov@peony.nlm.nih.gov>.
I really hope not to have caused a merging nightmare for all
the branches out there. Otherwise I'll be available for crucifixion.
- Move most of the functionality from the theme vtable
into GtkRcStyleClass and GtkStyleClass. The moved
vtable functions were changed a bit in the move to
work better in their new home.
- Get rid of the engine and engine_data fields from
GtkRcStyle and GtkStyle; instead the theme
engine derives theme-specific subclasses of GtkRcStyle
and GtkStyle
- Add extra dlsym() found entry point to themes,
theme_create_rc_style().
* gtk/gtkstyle.c: Copy xthickness, ythickness fields
in gtk_style_real_copy.
* gtk/themes.[ch]: add a function gtk_theme_engine_register_type()
to register a type associated with an engine. (The engine
won't be unloaded as there is an instance of the type.)