Sat Oct 3 21:01:34 BST 1998 Adam D. Moss <adam@gimp.org>
* app/channel.c app/channel_ops.c app/disp_callbacks.c
app/floating_sel.c app/gdisplay.c app/gdisplay.h
app/gdisplay_ops.c app/gimpimage.c app/image_map.c
app/interface.c app/layers_dialog.c app/plug_in.c app/undo.c
app/xcf.c:
Resizing image canvases should be less horrible to look at.
I've removed the implicit clear of the whole area when a
window is resized by the user, and clear the exposed area around
the image manually when appropriate.
Dialogs which want synchronous updates for previews and
such use displays_update_now().
Removed some old debugging nonsense which I don't want any more.
Sat Sep 26 20:46:18 BST 1998 Adam D. Moss <adam@gimp.org>
* app/channel.c app/channel_ops.c app/drawable.c
app/floating_sel.c app/gdisplay.c app/gdisplay.h
app/gimpimage.c app/layers_dialog.c app/undo.c:
Moved the idlerender stuff into gdisplay.c. Implemented
idlerender when doing floating_sel->layer, and undoing/redoing
layer deletion.
idlerender would be useful in many other places for improving
interactivity, if it weren't for the following problems:
* By definition, idlerender doesn't wait for a
gdisplays_update() call before starting work - it just
runs in idle time, which due to CPU contention with
plugins may not be genuinely available idle time when
things are 'noninteractive'.
* Most GIMP functions don't know whether they're
being run interactively or not. idlerender only
makes sense for interactive work. This is why
it is currently only applied to those functions which
would normally only be activated manually.
* Mixing idlerender and drawable_update() /
gdisplays_update_area() calls can lead to a region
being rerendered twice.
Hence, some slogwork is needed before idlerender can be
applied in the more general case.
bookkeeping without being used). Made copy_region more intelligent on
when to use tile sharing; some changes made to pixel_regions to
facilitate this. Fixed a refcount problem with xcf load and probably
a few other bugs that I've forgotten about. Added a sanity check in
set_undo_tiles to help with a problem larry is reporting with airbrush
and xinput. --sg
* app/tile_swap.c: made the g_message errors a little smarter instead
of popping up a ton of dialogs
* app/layer.c
* app/channel.c
* app/gdisplay.c
* app/gimprc.c: some code cleanups
-Yosh
Sun Jun 28 08:32:20 1998 Tim Janik <timj@gtk.org>
* app/layer.c (gimp_layer_mask_get_type):
(gimp_layer_get_type):
* app/drawable.c (gimp_drawable_get_type):
* app/channel.c (gimp_channel_get_type):
return a GtkType value and initialize the GtkTypeInfo structure
correctly.
* app/layer.h:
* app/drawable.h:
* app/channel.h:
provide GIMP_TYPE_* and GIMP_IS_*_CLASS() macros.
Applied a patch from Jay Cox. Reimplements selection_grow, selection_shrink
and selection_border using new algorithms => faster and nicer results.
--Sven
* app/indexed_palette.c: fix for wrong color selected in indexed
palette dialog
* app/xcf.c: don't crash on bad input (0 byte files)
* app/plug_in.h
* app/plug_in.c: fixes Gimp's most obscure bug. Failed plugin
queries are handle correctly now
* app/commands.c: added marching ants speed to preferences
* plug-ins/tiff/tiff.c: correction for inversion for MINISWHITE
images without alpha
* plug-ins/pcx/pcx.c: updated to new version
* app/paint_funcs.h: changed OPAQUE and TRANSPARENT to
OPAQUE_OPACITY and TRANSPARENT_OPACITY to avoid possible
conflicts. All affects .c files changed.
-Yosh
BST 1998 Adam D. Moss