Mon Sep 20 12:51:30 EDT 1999 Austin Donnelly <austin@gimp.org>
Honest, guv, it's not a feature - it's a tightly integrated
package of undo system cleanups and fixes.
NEW FILES:
* app/undo_history.c: window showing recent undo (and redo) steps
available.
* app/undo_types.h: broken out of undo.h to fix circular includes.
MODIFIED FILES:
* app/Makefile.am: compile undo_history.c
* app/channel.h: use enum for channel undo type, not just magic
numbers.
* app/layer.h: same for layer undos.
* app/commands.c: edit_show_undo_history_cmd_callback() function to
pull up undo history window.
* app/commands.h: prototype for above.
* app/gdisplay.c: make undo / redo menu items sensitive according
to whether they would do anything. Would be easy to change
the text to say what would be undone/redone, but I don't know
the GTK.
* app/gimpimage.c: new signal emitted by gimage:
UNDO_EVENT. gimp_image_undo_event() function to emit it.
* app/gimpimage.h: prototype for above.
* app/gimpimageP.h: pushing_undo_group member is now an undo_type,
not an int. Keep undo history widget here too (if created).
* app/menus.c: add "Edit/Undo history..." to image menu.
* app/undo.c: new types: enums undo_type and undo_state rather than
ints and magic numbers. All undo_pop_* and undo_free_*
functions made static. New static function
undo_type_to_name(). Issue undo event signals on various
important events (eg undo pushed, undo popped etc).
undo_push() now takes a "dirties_image" arg to say whether
image should be dirtied. Layer moves now dirty the image. A
couple of g_return_if_fails () on undo_pop and undo_redo to
assert we're not in the middle of an undo group.
undo_get_{undo,redo}_name() to peek at names of top items on
undo and redo stacks resp. undo_map_over_{undo,redo}_stack()
to run a function for each item or group on stack. Layer and
channel undos use symbolic names rather than 0 or 1. Array
mapping undo types to names.
* app/undo.h: split out undo types to undo_types.h. Prototypes
for functions described above. undo_event_t enum.
undo_history_new() prototype lives here too.
Random other fixes:
* app/gimpdrawable.c
* app/image_render.c: default labels in switches to keep egcs happy.
* app/nav_window.c: some fixes to (sort of) cope with image res !=
screen res. Still needs work to handle non-square pixels
properly.
* app/paths_dialog.c: bad idea to call gimp_image_dirty()
directly. Even though it's currently commented out.
Sun Feb 7 15:04:23 GMT 1999 Adam D. Moss <adam@gimp.org>
* app/fileops.c: More robust and kickin' thumbnail support.
* app/edit_selection.c app/layer.c app/layer.h: Working
on lazy opaque layer moves. Disabled for now.
* app/gdisplay.c
* app/gimage.h
* app/gimpimage.c
* app/gimpimage.h
* app/image_render.c
* app/tile_manager.c: Errr, I don't remember. No, seriously.
Nothing of consequence.
Sat Dec 5 21:31:57 GMT 1998 Austin Donnelly <austin@greenend.org.uk>
* app/commands.[ch]
* app/edit_selection.c
* app/gdisplay.[ch]
* app/gdisplay_ops.[ch]
* app/image_render.c
* app/info_window.c
* app/magnify.c
* app/menus.c
* app/scale.c: image rendering now happens with float scale
factors, independent in X and Y axes. Turning on dot-for-dot in view
menu does what gimp always used to do (ie one image pixel becomes
one monitor pixel). Dot-for-dot is on by default so people
shouldn't notice any difference unless they load an image that's
not at 72 dpi and also turn off dot-for-dot.
* app/app_procs.c
* app/gimprc.[ch]
* app/preferences_dialog.c: new gimprc options
(monitor-xresolution <float>) and corresponding
(monitor-yresolution <float>). Uglyness in preferences dialog to
add a "Monitor" page to the notebook, allowing user to set their
monitor's resolution or take it from the X server instead. This
badly needs cleaned up :(
* plug-ins/newsprint/newsprint.c: oops - this hasn't been working
for grayscale images since my last checkin. Now fixed.
ok, i modified *way* more files than CVS is telling me....
- add canvas_init machinery (tile_manager_validate equivalent)
- fix bugs in compositing code
- layers/channels dialog, previews
- all color handling should be fairly precision independent now
- replaced gimage_type and gimage_base_type with Tags
- remove concept of flat vs. layered gimages
ray lehtiniemi <rayl@netrover.com>
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
EDT 1999 Austin Donnelly