2000-12-19 Sven Neumann <sven@gimp.org>
* app/gimpbrushpipe.c (gimp_brush_pipe_load): set the spacing and
other info from the first brush just as the old code used to do.
2000-12-18 Sven Neumann <sven@gimp.org>
Jens Lautenbacher <jtl@gimp.org>
* app/Makefile.am
* app/gimpbrushlistP.h
* app/gimpbrushpipeP.h
* app/gimpobjectP.h: removed these three files
* app/parasitelistP.h
* app/channels_dialog.c
* app/docindex.c
* app/gimpdrawable.c
* app/gimpdrawableP.h
* app/gimpimage.c
* app/gimpimageP.h
* app/gimplist.[ch]
* app/gimpobject.c
* app/gimpobject.h
* app/gimpsetP.h: changed according to header removal
* app/airbrush.c
* app/brush_select.[ch]
* app/brushes_cmds.c
* app/gimpbrush.[ch]
* app/gimpbrushgenerated.[ch]
* app/gimpbrushlist.[ch]
* app/gimpbrushpipe.[ch]
* app/gimpcontextpreview.c
* app/paint_core.c
* app/paintbrush.c
* app/pencil.c
* tools/pdbgen/pdb/brushes.pdb: Big Brushes Cleanup.
The GimpBrush* object hierarchy and the file formats were broken by
"design". This made it overly difficult to read and write pixmap
brushes and brush pipes, leading to the situation that The GIMP was
not able to read it's very own file formats. Since the GimpBrush
format did support arbitrary color depths, the introduction of a
file format for pixmap brushes was unnecessary.
The GimpBrushPixmap object is dead. GimpBrush has an additional
pixmap temp_buf and handles pixmap brushes transparently. The file
format of pixmap brushes is not any longer a grayscale brush plus
a pattern, but a simple brush with RGBA data. The old brushes can
still be loaded, but the .gpb format is deprecated.
GimpBrushPipe derives from GimpBrush. The fileformat is still a text
header, followed by a number of brushes, but those brushes are stored
in the new GimpBrush format (no pattern anymore). The pipe does not
care about the depth of the contained GimpBrushes, so we get
grayscale BrushPipes for free. Since the brush loader still loads the
old format, old .gih files can also still be loaded.
Since the brushes in the GimpBrushPipe do not any longer contain a
pointer to the pipe object, we do only temporarily switch brushes
in the paint_core routines. This is not very elegant, but the best
we can do without a major redesign.
* app/patterns.[ch]: changed the loader to work with a filedescriptor
instead of a filehandle to make it work with the new brush loading
code.
* plug-ins/common/.cvsignore
* plug-ins/common/Makefile.am
* plug-ins/common/plugin-defs.pl
* plug-ins/common/gih.c: new plug-in that saves GIH files in the
new format (loader will follow soon)
* plug-ins/common/gpb.c: removed since Pixmap Brushes are no longer
supported as a special file format.
* plug-ins/common/gbr.c: load and save brushes in the new brush format
which allows RGBA brushes too.
* plug-ins/common/pat.c: load and save grayscale patterns too
2000-12-16 Sven Neumann <sven@gimp.org>
* app/gimppreviewcache.h
* app/channel.c
* app/layer.c: put the preview cache priming code back into place,
but disabled it for small images to avoid unnecessary upscaling.
2000-12-15 Sven Neumann <sven@gimp.org>
* app/gimppreviewcache.[ch]: added new function gimp_preview_scale().
* app/channel.c
* app/channels_dialog.c
* app/gimpimage.c
* app/layer.c
* app/layers_dialog.c
* app/lc_dialog.c
* app/nav_window.c: unified preview drawing code. Previews are never
generated larger than the canvas size. Image and channel previews are
scaled up to the desired size, so does the navigation window. The
layer previews are not yet scaled up again since we can not use
gimp_preview_scale here. I have removed the preview_cache priming
code since we don't need it any longer and it caused bad results.
2000-12-15 Michael Natterer <mitch@gimp.org>
* configure.in
* help/*: Updated Gimp-Help. See help/ChangeLog for details.
* app/channels_dialog.c
* app/gdisplay.c
* app/gradient.c
* app/layers_dialog.c
* app/menus.c
* app/paths_dialog.c
* app/preferences_dialog.c
* app/resize.c: fixed menu paths.
* plug-ins/helpbrowser/helpbrowser.c: load a special
"undocumented_filter.html" page if a filter was not found.
Adapted to the new help structure.
2000-12-15 Michael Natterer <mitch@gimp.org>
* makedummyhelpfile.sh
* makeindex.sh: removed.
* images/*: added some images.
* help/*: Updated to the latest version of the gimp-help CVS module.
This Version instoduces a bunch of new toplevel HTML files for
general topics, an autogenerated index, TOC, glossary,
an "image" directory on the "C" level, peace, love, and stuff...
Thanks to all help writers: Prof, Bex, Piers, ...
2000-12-14 Sven Neumann <sven@gimp.org>
* app/layers_dialog.c
* app/lc_dialog.c
* app/nav_window.c: another try on bug #31098. This time limited the
preview scale ratio in the calling functions. Please test!
* app/gimpimage.c: small cleanup
2000-12-14 Sven Neumann <sven@gimp.org>
* app/gimpimage.c
* app/layer.c
* app/layers_dialog.c
* app/nav_window.c: cleaned up these files while trying to find a
fix for #31098. No real changes, the bug is still present :-(
Mon Dec 11 22:52:43 GMT 2000 Andy Thomas <alt@gimp.org>
* app/gimpimage.c
* app/layers_dialog.c
Backed out some changes that caused problems with the Nav preview
and small sized images.
2000-12-11 Sven Neumann <sven@gimp.org>
* app/channel.c (channel_invalidate_previews): check for NULL pointer
* app/fileops.c: if saving and loading images at the same time, the
PDB obviously manages to mangle return values. By checking the
validity of the returned image id we avoid to crash as described in
bug #26788.
* app/module_db.c: unload modules from the idle_loop as suggested
by Austin in #5745.
2000-12-11 Sven Neumann <sven@gimp.org>
* app/edit_selection.c
* app/gimpimage.c
* app/layer_select.c
* app/layers_dialog.c: when computing a preview, limit the scale ratio
to a maximum of 1.0. By doing so we avoid to scale drawables up if the
image (canvas) size becomes larger than the drawable. Fixes bug #31098.
* app/gimppreviewcache.[ch]: indented
2000-12-03 Sven Neumann <sven@gimp.org>
* app/gimpimage.c: do not set the layer mode of the bottom layer to
NORMAL_MODE if it is DISSOLVE_MODE since the latter does not work
on the projection with the lower layer but on the layers alpha
channel. Fixes bug #7829.
2000-12-03 Sven Neumann <sven@gimp.org>
* app/layer.c
* app/layers_dialog.c
* plug-ins/common/jpeg.c: reviewed code in an attempt to fix the
crash-on-jpeg-save. The only thing that I have changed (despite
lots of indentation) is that we now connect the preview rewdraw
using gtk_signal_connect_while_alive() which should keep gimp from
trying to create previews of nonexistant drawables.
Updated german translation.
2000-11-20 Sven Neumann <sven@gimp.org>
* app/convert.c (build_palette_button): changed the logic that sets
the default for "Custom Palette" so it goes through the whole list
instead of stopping at the "Web" palette. The Indexed-Conversion
dialog now also remembers palettes appearing after the "Web" palette.
2000-11-18 Michael Natterer <mitch@gimp.org>
* app/fileops.c
* libgimp/gimpquerybox.[ch]: Make sure that we don't try to destroy
query_boxes twice or try to disconnect not-any-more connected
handlers.
* app/color_notebook.c
* app/gimpcontext.[ch]
* app/gimphelp.[ch]
* app/lc_dialog.[ch]
* app/menus.h
* app/preferences_dialog.c
* app/tools.[ch]
* libgimp/gimpcolorbutton.[ch]
* libgimp/gimpdialog.[ch]
* libgimp/gimpexport.[ch]
* libgimp/gimpfileselection.[ch]
* libgimp/gimphelpui.[ch]
* libgimp/gimppatheditor.[ch]
* libgimp/gimppixmap.[ch]
* libgimp/gimpsizeentry.[ch]
* libgimp/gimpui.[ch]
* libgimp/gimpunitmenu.[ch]
* libgimp/gimpwidgets.[ch]: in a coding attack, changed help_data
and many other strings passed to UI functions to (const gchar *).
As a consequence, I had to fix lots of warnings ;)
* plug-ins/common/tga.c
* plug-ins/imagemap/imap_main.c: fixed warnings.
Code cleanup and indentation all over the place.
app/gimpbrushpipe.c
Double frees due to dangling pointers [gimpbrushpipe.c
CVS-1.27 Lines 396 ff]. Upon failure to load an n-th
pixmap, a GPattern holding the dangling pointers would
be released first, also releasing n-1 pixmap's name and
RGB mask. These would again be freed in pipe and GTK
deconstruction.
2000-10-28 Michael Natterer <mitch@gimp.org>
* app/gimpdrawable.[ch]
* libgimp/gimpsizeentry.c: did some code review: s/0/NULL/,
s/sprintf/g_snprintf/, spacing.
2000-10-26 Sven Neumann <sven@gimp.org>
* app/channel_ops.c (offset_ok_callback): avoid modulo operation on
negative values.
* app/channel_ops.c
* app/crop.c
* app/file_new_dialog.c
* app/layers_dialog.c
* app/preferences_dialog.c
* app/rotate_tool.c
* app/scale_tool.c: use RINT() when assigning the result of
gimp_size_entry_get_refval() to an integer.
2000-10-22 Sven Neumann <sven@gimp.org>
* app/edit_selection.[ch]: plugged memleak (similar to the one that
was present in gtkutil_compress_motion()) in the key snooper.
Round moves to nearest integer instead of truncating the value.
This seems to fix the reported redraw problems when moving
selections at low zoom levels.
Cleaned up the code a little and converted enum values to uppercase.
* app/bezier_select.c
* app/free_select.c
* app/fuzzy_select.c
* app/move.c
* app/rect_select.c
* app/text_tool.c: updated to use the new EditType enum values.
* app/gimprc.c: minor optimization in the GList handling.
* app/layer.[ch]: removed unused functions.
* app/menus.c: removed "Dump Items (Debug)" menu entry.
2000-10-01 Michael Natterer <mitch@gimp.org>
* app/gimpbrush.c
* app/gimpbrushpipe.c
* app/pattern_header.h
* app/patterns.c: (Hopefully) fixed#8150.
- If something goes wrong during loading, free the resp. object
immediately and close it's FD.
- When destroying, check each pointer before freeing it.
- Never use if (!gimp_brush_load_brush() || !pattern_load())
but check the return values of both calls separately.
(the original code tried to free an already destroyed temp_buf).
- stuff like below (s/sz_PatternHeader/sizeof(PatternHeader)/g etc.)
2000-09-29 Michael Natterer <mitch@gimp.org>
* app/brush_edit.[ch]
* app/brush_header.h
* app/gimpbrush.[ch]
* app/gimpbrushgenerated.[ch]
* app/gimpbrushlist.[ch]
* app/gimpbrushpipe.[ch]
* app/gimpcontext.h
* app/gimplist.[ch]
* app/gimpobject.[ch]
* app/gimpset.[ch]
* app/gimpsignal.[ch]
* app/parasitelist.h: Lotsa fixes in the brush referencing code
and cleanups in many GtkObjects:
- Reference brushes correctly (call gtk_object_sink() after adding
them to the brush list). Don't crash when renaming a brush
multiple times.
- Replaced all gtk_object_destroy() with gtk_object_unref|sink().
- Removed the "Fake Gimp object system" (gimp_object_destroy() et.al.)
all over the place (while it's a good idea to have a common Gimp
base object, it's totally useless to try to imitate Gtk's object
system).
- s/sz_BrushHeader/sizeof(BrushHeader)/g
- And of course various indentation and coding style paranoia changes
in all files I visited.
2000-08-23 Garry R. Osgood <gosgood@idt.net>
* app/image_map.c
A bit of refresh drawable code was misplaced
these many years ago, leading to the persistence
of color tool effects, as reported in Mihail
Milushev's nicely composed bug report. Closes#7626
which has the gory details.
2000-08-23 Sven Neumann <sven@gimp.org>
* app/fileops.c: make the file_save dialog sensitive again if the
user clicked "No" when asked if he wishes to overwrite a file.
* app/gimpimage.c: when saving an undoable parasite, just do it
without calling undo_push_cantundo(). Undoable parasites are used
by many save plug-ins to attach a comment, but this operation
shouldn't prevent the user from using the undo_stack afterwards.
The JPEG plug-in has an additional bug that I couldn't solve, so
undoing after a Save As JPEG is still impossible.
* app/undo_history.c: lame attempt to draw the clean symbol at the
correct undo step. Not perfect but much better as before...
2000-08-16 Garry R. Osgood <gosgood@idt.net>
* app/gdisplay.c
* app/image_map.c
Addresses Bug Report #14704, and an
undisclosed undo system bug. Freezes
undo stack before interactive tools
commence idle rendering; thaws stack
on cleanup and immediately sets menu
sensitivity.
Not an image lock, but Yet Another UI
Lock. Oh well; merrilly we patch along...
See http://bugs.gnome.org/db/14/14704.html
for gory details. Closes#14704.
Tue Jul 18 23:34:34 BST 2000 Austin Donnelly <austin@gimp.org>
* plugs-ins/common/gif.c: The gif save plugin normally asks the
user if it's ok to crop the image if there are layers larger
than the image size. This isn't a good idea if we're running
with run_mode == NON_INTERACTIVE. Fix similar to that
suggested by Bryan Livingston <bryan@cooltext.com> on
gimp-developer mailing list.
* app/layer.c: Fix for Bug#17347: Scale Layer leaves artifacts.
Basic problem is that scaling a layer when there's a selection
active (eg there's a floating selection) doesn't invalidate
the layer bounds. Same problem with resizing a layer,
although the bug report didn't mention that. Fixed by changing
layer_resize() and layer_scale_lowlevel() to call
layer_invalidate_boundary() once done changing the layer.
2000-06-14 Michael Natterer <mitch@gimp.org>
* Makefile.am
* app/cursorutil.[ch]
* app/tools.c
* cursors/*: added lots of new cursors and removed old ones.
* app/gdisplay.[ch]: enabled the cursor setting parameters in
gdisplay_install_tool_cursor().
* app/bezier_select.c
* app/blend.c
* app/bucket_fill.c
* app/by_color_select.c
* app/clone.c
* app/color_picker.c
* app/crop.c
* app/disp_callbacks.c
* app/edit_selection.c
* app/eraser.c
* app/flip_tool.c
* app/ink.c
* app/iscissors.c
* app/magnify.c
* app/measure.c
* app/move.c
* app/paint_core.c
* app/rect_select.c
* app/text_tool.c
* app/transform_core.c: use the new cursors. Only the transform
tools are still using old cursors.
* app/layers_dialog.c: a tooltip for "Keep Trans."
* app/user_install.c: set the ctree's selection mode to BROWSE.
2000-06-09 Michael Natterer <mitch@gimp.org>
Cursor patch II:
This is only the logic inside the cursor system and not yet used.
* app/cursorutil.[ch]: [gimp]_change_win_cursor() take lots of
parameters now and compose cursors from up to three cursor
bitmaps/masks.
* app/gdisplay.[ch]: As a test, create a hardcoded example cursor
if "Cursor Mode" is set to "Tool Icon with Crosshair" in prefs.
* app/curves.c
* app/dialog_handler.c
* app/scroll.c: changed the calls to the win_cursor function.
* app/tools.[ch]: added a cursor and a toggle cursor to the ToolInfo
structure of all tools.
* app/toolsF.h: new ToolType TOOL_TYPE_NONE.
* app/gimpdnd.c
* app/interface.c: check for silly filenames in the file dnd
callback. Closes#13733.
* Makefile.am
* cursors/bucket_fill_small.xbm
* cursors/bucket_fill_small_mask.xbm
* cursors/crop_small.xbm
* cursors/crop_small_mask.xbm
* cursors/crosshair_small.xbm
* cursors/crosshair_small_mask.xbm
* cursors/ellipse_select_small.xbm
* cursors/ellipse_select_small_mask.xbm
* cursors/eraser_small.xbm
* cursors/eraser_small_mask.xbm
* cursors/free_select_small.xbm
* cursors/free_select_small_mask.xbm
* cursors/fuzzy_select_small.xbm
* cursors/fuzzy_select_small_mask.xbm
* cursors/intersect.xbm
* cursors/intersect_mask.xbm
* cursors/minus.xbm
* cursors/minus_mask.xbm
* cursors/move.xbm
* cursors/move_mask.xbm
* cursors/paintbrush_small.xbm
* cursors/paintbrush_small_mask.xbm
* cursors/pencil_small.xbm
* cursors/pencil_small_mask.xbm
* cursors/plus.xbm
* cursors/plus_mask.xbm
* cursors/rect_select_small.xbm
* cursors/rect_select_small_mask.xbm
* cursors/resize_small.xbm
* cursors/resize_small_mask.xbm
* cursors/zoom.xbm
* cursors/zoom_mask.xbm
* cursors/zoom_small.xbm
* cursors/zoom_small_mask.xbm: new files extracted from Tigert's
gimp-tool-cursors.xcf created at GimpCon.
Tigert, I'll commit the xcf as soon as I've added empty layers
with the names of the cursors that are missing.
* cursors/mouse.xbm
* cursors/mouse_mask.xbm: made it 32x32 to allow for cursor
composition.
2000-06-08 Sven Neumann <sven@gimp.org>
* app/qmask.c: call undo_push_qmask() before removing the
channel since the callback will set the qmask_state
incorrectly otherwise. Fixes bug #13472.
* app/undo.[ch]: code cleanup
* plug-ins/common/gif.c: put save_comment into the values
that are stored between calls to the GIF plug-in, so it
remembers if the user wants to save the comment. Fixes
bug #12755.
* app/layers_dialog.c: Disable the "Add Layermask" menuentry
for indexed images.
* app/gimpimage.c: Leave gimp_image_add_layer_mask(...) after
spitting out messages when trying to add a layermask to an
indexed image.
-Yosh
2000-06-06 Michael Natterer <mitch@gimp.org>
* Makefile.am
* cursors/crosshair.xbm
* cursors/crosshair_mask.xbm: new files.
* app/cursorutil.[ch]
* app/gdisplay.[ch]
* app/gimprc.[ch]
* app/preferences_dialog.c: Started to add the new set of
consistent tool cursors. The crosshair one is the easy part
but I wanted to see a result :) The other two options in
the pref's cursor menu are still dummys.
2000-05-31 Michael Natterer <mitch@gimp.org>
* app/gimpunit.c
* libgimp/*: all libgimp headers are included via gimp.h or
gimpui.h, so include <gtk/gtk.h> there and in the *.c files.
Various cleanups.
2000-05-29 Sven Neumann <sven@gimp.org>
* app/Makefile.am
* app/paint_core.c
* app/paint_core_kernels.h: moved brush subsampling kernels
into its own header file and generated a slightly different
kernel (using the new kernelgen tool, see below). The new
kernel simulates a circular brush instead of a rectangular
one and gives slightly different (lighter) results.
* app/gimage_mask.c: when stroking a selection, offset the
points by 0.5 to align the brushes with the pixel grid. This
lets you create 1-pixel wide rectangles and ellipses.
* tools/Makefile.am
* tools/kernelgen.c: simple hack to generate subsampling
kernels.
* app/gimpimage.c
--Fixed Fujita Yuji's bug (reported 27May2000 to
gimp-developer) -larger than 64x64 brush and
selection mask crashes smudge/dodgeburn/convolve
This was a rowstrides bug in gimp_image_replace_image.
2000-05-21 Michael Natterer <mitch@gimp.org>
These changes enable help support for 3rd party plug-ins which
install their help files outside GIMP's main help dir.
Instead of calling gimp_help(), gimp_plugin_help_func() etc.,
all help callbacks now have to call gimp_standard_help_func()
which has different implementations in the app and in libgimp.
There is a new function gimp_plugin_help_register() which can
be called during plug-in query. plug_in.c keeps a list of
executable_name/help_path pairs. Plug-ins have to pass their
exec. name to gimp_help() which uses the list to find the plug-in's
help directory.
* app/gimphelp.[ch]: gimp_help() now takes a help_path parameter.
help_path == NULL means the standard help directory. Various
changes to pass the help_path to the help browser.
* app/gimprc.c: save the plug-in's help_path in the pluginrc file.
* app/menus.c: ugly hack to enable help_paths in the "F1" callback.
* app/plug_in.[ch]: many help_path related changes. Use g_basename()
instead of strrchr(str,G_DIR_SEPARATOR), cosmetic cleanups.
* app/internal_procs.c
* app/gimphelp_cmds.c
* tools/pdbgen/pdb/gimphelp.pdb: new procedure
gimp_plugin_help_register(). gimp_help() takes a second parameter
which is the executable name (not the help_path).
* app/color_notebook.c
* app/commands.c
* app/lc_dialog.c
* app/preferences_dialog.c
* app/tools.c: call gimp_standard_help_func() instead of gimp_help().
* libgimp/gimp.c: new function gimp_get_progname() which returns
the full path of the plug-in's executable.
* libgimp/gimp.h: export the new function,
removed gimp_plugin_help_func(), gimp_help() takes the executable
name as second parameter.
* libgimp/gimpcompat.h: added gimp_plugin_help_func().
* libgimp/gimphelp.c: a wrapper for gimp_plugin_help_register(),
changed the calls to gimp_help.
* libgimp/gimphelpui.[ch]: call gimp_standard_help_func() instead
of gimp_help().
* plug-ins/helpbrowser/helpbrowser.c: now called with an additional
help_path parameter. Various changes to enable
help_path != gimp_standard_help_path.
Unrelated stuff:
* app/batch.h: added missing GPL header.
* app/gimpunit.c: had a LGPL header, merged some fprintf's into
one call.
* app/procedural_db.[ch]: cosmetic: g* types, s/g_malloc/g_new/,
prototypes, indentation.
* app/resize.c: use less packing widgets. didn't find the "offset"
redraw bug :(
2000-05-12 Sven Neumann <sven@gimp.org>
* gimpdrawable.c: enabled the (commented out) signal
"invalidate_preview".
* app/layers_dialog.c: connect to the "invalidate_preview"
signal to catch changes that need to be shown in the layer
previews. Synthetize an expose event when a layer changes.
Expose events are optimzed away by GTK+ if the widget is not
visible. Therefore, previews not visible in the layers_dialog
are not redrawn when they invalidate. Later the preview gets
validated by the image_preview in lc_dialog but is never
propagated to the layer_pixmap. We work around this by using an
additional flag "layer_pixmap_valid" so that the pixmap gets
updated once the preview scrolls into sight.
Fixes bugs #10549, #10300 and #8787.
* app/channel.[ch]
* app/layer.[ch]: code review and indentation
--Sven
Sven Neumann