Thu Apr 2 04:06:50 EST 1998 Matthew Wilson <msw@gimp.org>
* app/disp_callbacks.c: Always ungrab the pointer on button release.
This is a quick and dirty hack to try to nuke the pointer grab
bug.
Thu Apr 2 01:49:34 EST 1998 Matthew Wilson <msw@gimp.org>
* app/transform_core.c: added a change to make a note of the
current drawable in the button_release. This prevents the tool
from being restarted when the transform_core makes a floating
selection.
* app/tools.c: Initialize the current drawable to NULL if
there is gdisp == NULL.
--Matt
* app/brightness_contrast.c
* app/by_color_select.c
* app/color_balance.c
* app/color_picker.c
* app/curves.c
* app/histogram_tool.c
* app/hue_saturation.c
* app/levels.c
* app/posterize.c
* app/threshold.c
* app/transform_core.c: modified to call the cancel callback
instead of the ok callback when freed. modified to save the
last used drawable so that we might be able to check later
and restart the tool if need be.
* app/disp_callbacks.c
* app/gimage.c: modified tool restart/destruction code
* app/tools.c
* app/tools.h: added tools_initialize, made changes to
be able save the last used drawable in the tool.
--Matt
I will hopefully finish my changes to the dialog tonight and will upload a
new version of my patch. Could you (the core developers) please make up your
mind whether to include it in 1.0 or not...
The Blend-Tool dialog now changes the repeat_mode_menu to insensitive when
the Gradient-Type is one of the Shapeburst-types.
--Sven
Wed Apr 01 04:07:29 EST 1998 Matthew Wilson <msw@gimp.org>
* app/disp_callbacks.c: Only destroy the current tool if you change
gdisplays and the tool caches image data.
--Matt "flood the commits-list" Wilson
Wed Apr 01 03:40:15 EST 1998 Matthew Wilson <msw@gimp.org>
* app/disp_callbacks.c: If the current gdisp->ID does not match
the ID of the gdisp_ptr in the active_tool, restart it.
--Matt "I need sleep" Wilson
Tue Mar 31 03:10:12 EST 1998 Matthew Wilson <msw@gimp.org>
* app/iscissors.c: Set tool->preserve to TRUE in tools_new_iscissors.
--Matt "forgets files when committing" Wilson
Tue Mar 31 02:21:15 EST 1998 Matthew Wilson <msw@gimp.org>
* app/gimage.c
* app/tools.c
* app/tools.h: Added a field in the Tools struct, preserve.
During gimage_dirty, if this flag is not set the tool will be
reset. This is good for tools that keep a copy of the image
in cache for local manipulation like transform_core.
* app/bezier_select.c
* app/blend.c
* app/brightness_contrast.c
* app/bucket_fill.c
* app/color_balance.c
* app/color_picker.c
* app/crop.c
* app/curves.c
* app/ellipse_select.c
* app/free_select.c
* app/histogram_tool.c
* app/hue_saturation.c
* app/levels.c
* app/move.c
* app/paint_core.c
* app/posterize.c
* app/rect_select.c
* app/text_tool.c
* app/transform_core.c: Set the preserve flag to the correct
values in the new functions and wrapped calls to functions that
dirty the gimage to prevent tool restarts.
* app/disp_callbacks.c
* app/menus.c: Removed Nether's tool patch.
--phew. Matt
* app/brush_select.c
* app/iscissors.c: get rid of unused variable warnings
* app/fileops.c: refresh the filesel better
* app/plug_in.c: make sure everything gets initialized to something in the
plug-in struct
-Yosh
This fixes the crashes when closing an image with the by-color-select-dialog
associated to it still open and pressing "Reset" or clicking into the preview.
The checks produce some warnings like
** WARNING **: invalid class type (unknown)' in cast to impDrawable'
when failing, but I consider this less evil than a crash.
There is still a problem: Sometimes the by-color-select-dialog is closed
when the image is closed and can't be opened anymore. Anyone?
--Sven
Fri Mar 27 13:41:55 EST 1998 Matthew Wilson <msw@gimp.org>
* app/floating_sel.c: Always update the drawable when converting
a floating selection to a layer
--Matt
Fri Mar 26 03:13:15 EST 1998 Matthew Wilson <msw@gimp.org>
* app/paint_core.c: Offset brush buffer when painting on edges
of the canvas.
* app/text_tool.c: Set window to auto_shrink = FALSE
-Matt
Fri Mar 26 01:30:45 EST 1998 Matthew Wilson <msw@gimp.org>
* app/iscissors.c
* app/bezier_select.c: Change to select the bezier tool properly
when converting from iscissors to bezier
-Matt
* app/iscissors.c: This version works much better.
outstanding bugs <VERY IMPORTANT>:
1. the edgemap does not seem to be constructed quite right
(construct_edge_map) -- uncomment #define
ISCISSORS_STILL_DOES_NOT_WORK and take a look at the edgemap -- it
will be saved to the file "dump"
2. sigsegvs if the layer offset isn't 0,0. dunno why.
3. convert to bezier still does not work! Any tools experts wanna
help me with it? (it converts to bezier fine, it just doesn't set
the active tool to bezier currectly.)
* app/iscissors.c: fixes the display artifact where the preview
outline of the selected area would not appear or would appear in
the wrong place or have the wrong shape
* iscissors.c: This should hopefully fix most of the UMRs
(uninitialized memory reads) found with Purify. No noticable
change in iscissors's behavior, unfortunately.
* app/app_procs.c app/brush_select.c app/brushes.c
app/brushes.h app/convert.c app/gradient.c app/gradient.h
app/palette.c app/palette.h app/pattern_select.c app/patterns.c
app/patterns.h: Fixed up --no-data so that the various *_init()
routines always get called, and so that when in interactive mode
the data can be loaded by the user even when started with --no-data.
Fixes the blend bug too...
yeah, sven scooped me... but i think this is still a needed patch...
-adrian
started with the --no-data option. This is fixed now. grad_get_color_at()
now returns a totally transparent black when no gradient is present.
Additionally the gradient editor now behaves better when the window is
resized. The resize code is still an ugly hack though.
--Sven
1998-03-25 Federico Mena Quintero <federico@nuclecu.unam.mx>
* app/asupsample.c (adaptive_supersample_area): Removed Sopwith's
"speed optimizations", as they introduced bugs (exposed by using a
high threshold value for supersampling). They weren't much of a
speed improvement, anyway (memcpy on very small regions, etc.).
Also, assuming that all-bytes-zero is a floating-point 0 value is
not portable.
Wed Mar 25 16:22:00 EST 1998 Matthew Wilson <msw@gimp.org>
* app/app_procs.c: now prints messages when starting up --no-splash
* app/gimprc.c: now prints messages when starting up --no-splash
* app/plug-in.c: does not print messages when starting up with
splash screen
* app/fileops.c: refresh the open/save as dialogs upon open
* applied gimp-quinet-980323-0, fixing more M_PI madness
* applied gimp-quinet-980323-1 which fixes redundant inline checks
* plug-ins/gfig/gfig.c: changed stipple to type unsigned char to quell warnings
-Yosh
* configure.in: corrected test for libXmu for some systems; added test for
difftime
* app/main.c: use glib ATEXIT macro
* app/text_tool.c: applied gimp-stric-980321-0 (text preview refresh)
* plug-ins/script-fu/script-fu-console.c: don't need to init gtkpreview stuff,
since we don't use them
* plug-ins/script-fu/*: many portability fixes
* plug-ins/pnm/pnm.c: sprintf portability patch
* plug-ins now #define RAND_MAX if needed
* plug-ins/sparkle/sparkle.c: applied gimp-joke-980322-1
-Yosh
* app/gimage_mask.c
* app/paint_core.c: applied patch from Ben Jackson to fix fractional pixel
errors and reverted the old fix.
* app/paint_funcs.h: changed the #defines for ERASE_MODE and REPLACE_MODE to
correctly match layer_modes[]
-Yosh
Mon Mar 16 21:52:46 CST 1998 Larry Ewing <lewing@gimp.org>
* app/convert.c (indexed_ok_callback): added gimage_ID to the
dialog struct to check if the gimage was still around.
* app/commands.c: fixed border/grow/shrink to check for valid
gimage. (Once again object if you don't like this method).
Mon Mar 16 17:05:54 CST 1998 Larry Ewing <lewing@gimp.org>
* app/commands.c: changed select_feather_cmd_callback to pass the
gimage->ID rather than the gimage to prevent a fatal error. If
you have objections to this speak now.
