2004-03-20 Sven Neumann <sven@gimp.org>
* app/widgets/gimppropwidgets.c (gimp_prop_size_entry_callback):
avoid to set the unit property with every size change; only set it
if it actually changed.
* app/core/gimpimage-undo-push.c (gimp_image_undo_push_text_layer):
allow to pass a GParamSpec that identifies a single text property
to be changed. In this case, don't store a GimpText object on the
undo stack but only the changed value.
* app/tools/gimptexttool.c: use the new undo feature to reduce the
memory footprint of text undo for the common case.
* app/text/gimptextlayer.c: changed accordingly.
2004-03-20 Simon Budig <simon@gimp.org>
* app/core/gimpimage-qmask.c: Applied slightly modified patch
from Sven. When the quickmask has a floating selection, anchor
it before loading the selection and deleting the qmask channel.
Fixes bug #137170.
2004-03-20 Simon Budig <simon@gimp.org>
* app/core/gimpundo.[ch]: Make it possible to refresh the preview
of an undo step.
* app/tools/gimpeditselectiontool.c
* app/widgets/gimpitemtreeview.c
* app/widgets/gimplayertreeview.c: refresh the preview when
compressing undos. This ensures that the last preview in the undo
history always reflects the current state of the image.
2004-03-20 Sven Neumann <sven@gimp.org>
* app/core/gimpimage-undo-push.c (undo_pop_text_layer): don't
exchange the text_layer's text object but sync it with the text
object from the undo step.
* app/text/gimptextlayer.c (gimp_text_layer_set): in case the
layer has a mask, push an undo group around the text modifications.
* app/tools/gimptexttool.c (gimp_text_tool_idle_apply): push a
text layer undo before applying the text changes.
2004-03-17 Michael Natterer <mitch@gimp.org>
* app/core/gimpdrawable-transform.c
(gimp_drawable_transform_paste): push the undo group around all
operations, not just around some of them. Pulled common code
out of the if() { } else { } construct. Cleanup.
2004-03-17 Simon Budig <simon@gimp.org>
* app/gui/plug-in-menus.c: Sort the plugin menu entries with
the mnemonics stripped. Avoids weird ordering in the "C" and
"POSIX" locales.
* app/widgets/gimpitemtreeview.c: make a simple click on the
"New" Button use defaults and use shift-click for the new-dialog
invocation.
Some more useless button cleanup:
* app/widgets/gimpdatafactoryview.c: only create an Edit button
when the edit_function is set.
* app/core/gimp.c: don't set an edit func for the patterns.
* app/gui/patterns-menu.c: Don't create the "New", "Edit" and
"Duplicate" Menu entries for the patterns.
* app/widgets/Makefile.am
* app/widgets/widgets-types.h
* app/widgets/gimppatternfactoryview.[ch]: New widget:
gimp_pattern_factory_view. Necessary to be able to hide the
"duplicate" button...
* app/gui/dialogs-constructors.c: Use it.
2004-03-17 Sven Neumann <sven@gimp.org>
Changes for help i18n in the core, the rest will take place in the
help plug-in:
* app/text/gimptext.[ch]: removed gimp_text_get_default_language()
* app/core/gimp-utils.[ch]: ... and added it here as
gimp_get_default_language().
* app/config/gimprc-blurbs.h
* app/config/gimpdisplayconfig.[ch]: added property "help-locales".
* app/widgets/gimphelp.c: use the new property and pass it to the
help plug-in.
* app/core/gimpselection.c (gimp_selection_invalidate_boundary):
removed unused variable.
2004-03-16 Michael Natterer <mitch@gimp.org>
* app/core/gimplist.c (gimp_list_dispose): removed this function.
It was removing all items without freezing the container first,
which caused excessive signal handler activity in GimpContext.
Cleaned up the whole file a bit.
* app/core/gimpcontainer.c (gimp_container_dispose): call
gimp_container_clear() which does the right thing and e.g. speeds
up quitting significantly when gimp->documents is huge.
Reported by Jimmac.
2004-03-15 Michael Natterer <mitch@gimp.org>
* app/core/core-enums.[ch] (enum GimpUndoType): replaced
GIMP_UNDO_LAYER_MOD and GIMP_UNDO_CHANNEL_MOD by
GIMP_UNDO_DRAWABLE_MOD.
* app/core/gimpimage-undo-push.[ch]: ditto: replaced
gimp_image_undo_push_layer_mod() and
gimp_image_undo_push_channel_mod() by
gimp_image_undo_push_drawable_mod().
* app/core/gimpdrawable.[ch]: added undo_desc strings for "resize"
and "scale" to the GimpDrawableClass struct.
(gimp_drawable_scale)
(gimp_drawable_resize): pass push_undo = TRUE to
gimp_drawable_set_tiles_full() and use the undo_desc from the
class.
(gimp_drawable_real_set_tiles): push a GIMP_UNDO_DRAWABLE_MOD here...
* app/core/gimpchannel.c
* app/core/gimplayer.c: ...and don't push undos in
GimpItem::scale(), GimpItem::resize(), GimpDrawable::set_tiles().
* app/core/gimpchannel.c: Removed even more bounds_known = FALSE
assignments from functions which already call
gimp_drawable_set_tiles().
2004-03-15 Michael Natterer <mitch@gimp.org>
* app/core/gimpdrawable.c (gimp_drawable_set_tiles_full): emit
"update" signals from the drawable before and after setting tiles
and offsets.
* app/core/gimpdrawable-offset.c (gimp_drawable_offset)
* app/core/gimpdrawable-transform.c (gimp_drawable_transform_paste)
* app/core/gimpimage-undo-push.c (undo_pop_layer_mod, _channel_mod)
* app/text/gimptextlayer.c (gimp_text_layer_render)
* app/tools/gimptransformtool.c (gimp_transform_tool_doit):
removed calls to gimp_drawable_update().
* app/core/gimpdrawable-offset.c (gimp_drawable_offset): don't
push an undo step before calling gimp_drawable_set_tiles()
but simply pass push_undo == TRUE and the undo_desc.
2004-03-15 Michael Natterer <mitch@gimp.org>
* app/core/gimpdrawable.[ch]: added "offset_x" and "offset_y"
parameters to GimpDrawable::set_tiles().
(gimp_drawable_set_tiles): removed the "GimpImageType" parameter.
(gimp_drawable_set_tiles_full): new function adding type, offset_x
and offset_y parameters.
(gimp_drawable_real_set_tiles): set the drawable's offsets from
the offset parameters and its size from the passed TileManager's
size. Emit "size_changed" accordingly.
* app/core/gimpchannel.c
* app/core/gimpdrawable-offset.c
* app/core/gimpdrawable-transform.c
* app/core/gimpimage-convert.c
* app/core/gimpimage-undo-push.c
* app/core/gimplayer.c
* app/text/gimptextlayer.c
* app/tools/gimptransformtool.c: changed accordingly: removed
calls to gimp_viewable_size_changed() and all sorts of hackish
assignments of the drawable's width/height/offset_x/offset_y
properties.
2004-03-15 Michael Natterer <mitch@gimp.org>
* app/text/gimptextlayer.c (gimp_text_layer_render): don't call
gimp_image_flush().
* app/tools/gimpxttool.c (gimp_text_tool_apply): call it here
instead.
Now that we have a common place that exchanges drawable->tiles,
we can abstract away boundary invalidation for this operation:
* app/core/gimpdrawable.c (gimp_drawable_real_set_tiles):
call gimp_drawable_invalidate_boundary() before setting
the new tiles.
* app/core/gimpchannel.c (gimp_channel_set_tiles)
* app/core/gimpdrawable-transform.c (gimp_drawable_transform_paste)
* app/core/gimpimage-undo-push.c (undo_pop_layer_mod)
* app/core/gimplayer.c (gimp_layer_scale) (gimp_layer_resize)
(gimp_layer_flip) (gimp_layer_rotate) (gimp_layer_transform)
* app/text/gimptextlayer.c (gimp_text_layer_render): removed
calls to gimp_drawable_invalidate_boundary() from all functions
which finally call gimp_drawable_real_set_tiles().
* app/tools/gimptransformtool.c (gimp_transform_tool_doit): no
need to set channel->bounds_known to FALSE, because
gimp_drawable_set_tiles() already did this.
2004-03-15 Michael Natterer <mitch@gimp.org>
* app/core/gimpimage.[ch]: added "gboolean alpha_changed" to
GimpImageFlushAccumulator. Install an "alpha_changed" handler
on gimage->layers and set flush_accum.alpha_changed = TRUE
whenever the image's *only* layer changed its alpha.
* app/core/gimpimage-undo-push.c (undo_pop_layer_mod)
* app/core/gimplayer.c (gimp_layer_add_alpha): removed
explicit calls to gimp_image_alpha_changed().
2004-03-15 Michael Natterer <mitch@gimp.org>
* app/core/gimpdrawable.c (gimp_drawable_real_set_tiles): emit
"alpha_changed" if the drawable got/lost an alpha channel by
setting the new tiles.
* app/core/gimpimage-undo-push.c (undo_pop_layer_mod)
* app/core/gimplayer.c (gimp_layer_add_alpha): removed explicit
calls to gimp_drawable_alpha_changed().
2004-03-15 Michael Natterer <mitch@gimp.org>
Closer to text layer undo:
* app/core/gimpchannel.c (gimp_channel_set_tiles): invalidate the
channel's boundary and bounds.
* app/text/gimptextlayer.c: implement GimpDrawable::swap_pixels()
and set text_layer->modified = TRUE after upchaining.
(gimp_text_layer_render): use gimp_drawable_set_tiles() and
set text_layer->modified = FALSE afterwards.
* app/core/gimpimage-undo-push.c: cleaned up variable declarations
and initializations.
(undo_pop_layer_mod)
(undo_pop_channel_mod): use gimp_channel_set_tiles() instead of
touching drawable->tiles manually. Now all pixel manipulation
(at least on layers) should be virtualized and can be detected by
the text layer.
2004-03-15 Michael Natterer <mitch@gimp.org>
Prepare the undo system for proper text layer undo:
* app/core/core-enums.[ch] (enum GimpUndoType): replaced
GIMP_UNDO_IMAGE and GIMP_UNDO_IMAGE_MOD by GIMP_UNDO_DRAWABLE.
* app/core/gimpimage-undo-push.[ch]: ditto: replaced
gimp_image_undo_push_image() and gimp_image_undo_push_image_mod()
by gimp_image_undo_push_drawable() which *always* expects to get a
TileManager passed. Also added g_return_if_fail()s to check if the
passed in tile manager follows the semantics of the "sparse"
boolean.
(undo_pop_drawable): removed all code and call the new
gimp_drawable_swap_pixels() instead (see below).
* app/core/gimpdrawable.[ch] (gimp_drawable_push_undo): if tiles
are NULL, create a copy of the area here and always pass tiles to
gimp_image_undo_push_drawable(). Added lots of g_return_if_fail()
here too.
Added new vitrual function GimpDrawable::swap_pixels() which
does what undo_pop_drawable() did.
* app/core/gimpchannel.c: implement swap_pixels() and invalidate
the channel's bounds and boundary.
2004-03-15 Simon Budig <simon@gimp.org>
* app/core/gimpimage.[ch]
* app/core/gimpimage-undo.c: renamed gimage->freeze_count
to gimage->undo_freeze_count (more explicit).
* app/file/file-open.c: Ensure that the undo system is enabled
after loading an image (to protect against broken loaders).
2004-03-14 Simon Budig <simon@gimp.org>
* app/core/gimpimage.[ch]
* app/core/gimpimage-undo.c: introduced a freeze counter, so
that undo_freeze(); undo_freeze(); undo_thaw(); no longer
results in an thawed undo stack.
Fixes Bug #124176.
This might reveal other places in the code where
gimp_image_undo_freeze/thaw or gimp_image_undo_disable/enable
were used in an inconsistent (not-paired) manner.
* app/file/file-open.c: e.g. here, where _enable was called on
the new image. The intention was to free a potential undo stack,
so now call gimp_image_undo_free() instead.
2004-03-14 Michael Natterer <mitch@gimp.org>
* app/core/gimpdrawable-offset.c (gimp_drawable_offset)
* app/core/gimpdrawable.c (gimp_drawable_scale) (gimp_drawable_resize)
* app/core/gimplayer.c (gimp_layer_convert) (gimp_layer_add_alpha):
use gimp_drawable_set_tiles() instead of fiddling with
drawable->tiles, ->bytes etc. manually.
2004-03-13 Sven Neumann <sven@gimp.org>
* app/core/gimpdrawable-blend.c (gradient_calc_shapeburst_*):
applied a patch from Pedro Gimeno that fixes the crash reported in
bug #136219.
2004-03-13 Michael Natterer <mitch@gimp.org>
Completed the fix for bug #136702:
* app/core/gimpitem.[ch]: added "gboolean supersample" and
"gint recursion_level" to GimpItem::transform().
* app/core/gimpitem-linked.[ch] (gimp_item_linked_transform): ditto.
* app/core/gimpdrawable-transform.[ch]: added "recursion_level"
parameters and removed the RECURSION_LEVEL #define.
* app/core/gimpchannel.c
* app/core/gimpdrawable.c
* app/core/gimplayer.c
* app/vectors/gimpvectors.c: changed accordingly.
* app/tools/gimptransformoptions.[ch]: added new property
"recursion_level" which is not serializable and has no GUI. Pretty
useless, but it's IMHO better to hardcode the default value here
than in gimpdrawable-transform.c
* app/tools/gimptransformtool.c: changed accordingly.
* tools/pdbgen/pdb/transform_tools.pdb: hardcode "recursion_level"
to 3.
* app/pdb/transform_tools_cmds.c: regenerated.
2004-03-13 Sven Neumann <sven@gimp.org>
* app/core/gimpdrawable.[ch]: added new virtual function
GimpDrawable::set_tiles().
* app/core/gimpchannel.c
* app/core/gimplayer.c: push an undo before chaining up in
set_tiles().
* app/core/gimpdrawable-transform.c
* app/core/gimpimage-convert.c
* app/tools/gimptransformtool.c: use gimp_drawable_set_tiles()
instead of fiddling with the drawable's tile manager directly.
2004-03-13 Raphael Quinet <quinet@gamers.org>
* app/tools/gimptransformoptions.[ch]: added new "supersample"
property to GimpTransformOptions and added corresponding check
button in the option dialog for the transform tools.
* app/core/gimpdrawable-transform.[ch],
* app/core/gimpdrawable.c,
* app/tools/gimptransformtool.c: new "gboolean supersample"
parameter added to gimp_drawable_transform_tiles_affine() and
gimp_drawable_transform_affine().
* tools/pdbgen/pdb/transform_tools.pdb: ditto. For the PDB calls,
the supersample parameter is set to FALSE for "rotate" and "shear"
and set to TRUE for "perspective", "scale" and "transform_2d".
* app/pdb/transform_tools_cmds.c: regenerated.
The new "supersample" option lets the user decide if the
transformations should use supersampling (RECURSION_LEVEL 3) or
not. This fixes both bug #136702 and bug #109817. Hopefully for
good, this time.
2004-03-12 Raphael Quinet <quinet@gamers.org>
* app/core/gimpdrawable-transform.c (RECURSION_LEVEL): Set to 0 in
order to avoid nasty blurring effects. Fixes bug #136702, but
re-opens bug #109817. This is the best compromise for 2.0.
Later, we will have to make the supersampling optional as
suggested in the original patch to bug #109817.
2004-03-11 Sven Neumann <sven@gimp.org>
* app/config/gimpconfig-utils.c: made gimp_config_sync() and
gimp_config_connect() also work on objects of different types.
Properties with the same name and the same type are synced /
connected.
* app/core/gimpcontext.[ch]: added convenience functions to get/set
the font by name.
* app/tools/gimptextoptions.[ch]: don't hold a GimpText object
that duplicates properties like font and color which are in
GimpContext already. Instead added all text properties that are
controlled from the text tool options. Handling of the foreground
color is somewhat broken and needs a GimpContext wizard (Mitch!).
* app/text/gimptext.c: blurbs are not any longer needed now that
the property widgets are created from the GimpTextOptions.
* app/tools/gimptexttool.c: changed accordingly.
* app/widgets/gimptexteditor.[ch]: use an internal GtkTextBuffer
and emit "text-changed" when it changes.
2004-03-07 Michael Natterer <mitch@gimp.org>
* app/core/gimp-edit.c (gimp_edit_extract): fixed bug added when
factoring out gimp_edit_extract(): check if tile_manager_crop()
cropped away the whole TileManager before making a GimpBuffer from
it.
2004-03-07 Sven Neumann <sven@gimp.org>
Made size of undo previews configurable. Not dynamic for now, but
at least not hardcoded any longer. Fixes bug #119905:
* app/config/gimpcoreconfig.[ch]
* app/config/gimprc-blurbs.h: added "undo-preview-size" property.
* app/core/gimpundo.[ch]: use the new property instead of a
hardcoded value.
* app/widgets/gimpundoeditor.[ch]: added a "preview-size"
construct property.
* app/gui/dialogs-constructors.c: changed accordingly.
* app/gui/preferences-dialog.c: added a widget to control the undo
preview size.
2004-03-03 Sven Neumann <sven@gimp.org>
* app/core/gimpimagefile.c (gimp_imagefile_get_new_preview):
let GdkPixbuf scale the thumbnail. It does a nicer job than
temp_buf_scale().
2004-03-03 Michael Natterer <mitch@gimp.org>
* libgimpthumb/gimpthumb-utils.[ch] (gimp_thumb_file_test):
return the "errno" of a failed stat() call.
* libgimpthumb/gimpthumbnail.[ch]: added member
"image_not_found_errno" which contains the errno as returned by
gimp_thumb_file_test() if thumbnail->image_state is
GIMP_THUMB_STATE_NOT_FOUND.
* app/core/gimpimagefile.c (gimp_imagefile_get_desc_string): display
the real error as specified by thumbnail->image_not_found_errno
instead of just "Could not open".
2004-02-28 Michael Natterer <mitch@gimp.org>
* app/core/gimplayer.c (gimp_layer_apply_mask): when deleting the
mask, use "Delete Layer Mask" instead of "Apply Layer Mask" as
undo description.
2004-02-27 Sven Neumann <sven@gimp.org>
* app/core/gimpimagefile.c (gimp_imagefile_save_thumb): peek the
image to make sure that mtime and filesize are written correctly.
2004-02-25 Michael Natterer <mitch@gimp.org>
Changed libgimpthumb API in a way that will make GtkFileChooser
integration possible without doing major hacks:
* libgimpthumb/gimpthumb-enums.h: added GimpThumbFileType enum
and extended GimpThumbState enum.
* libgimpthumb/gimpthumb-utils.[ch]: return GimpThumbFileType
instead of gboolean from gimp_thumb_file_test().
* libgimpthumb/gimpthumbnail.c
* app/core/gimpimagefile.c: changed accordingly (the icons
used for REMOTE, FOLDER and SPECIAL need to be changed).
2004-02-23 Michael Natterer <mitch@gimp.org>
* app/core/core-types.h: added #defines for GimpCoords' default
pressure, tilt and wheel values.
* app/display/gimpdisplayshell-callbacks.c
* app/paint/gimppaintcore-stroke.c
* app/text/gimptext-vectors.c
* tools/pdbgen/pdb/paint_tools.pdb: use them. Fixes lots more
buggy default values for tilt (in fact all of them were wrong).
* app/pdb/paint_tools_cmds.c: regenerated.
2004-02-23 Michael Natterer <mitch@gimp.org>
* app/core/gimpimage-snap.[ch]: return the snapped coordinates as
gdouble, not gint because we must not round the unsnapped axis
if only one axis snapped.
* app/display/gimpdisplayshell.[ch] (gimp_display_shell_snap_coords):
changed accordingly. Added boolean return value indicating "snapped".
* app/display/gimpdisplayshell-cursor.[ch]
(gimp_display_shell_update_cursor): take both display_x and _x and
(possibly snapped) image_x and _y. Use the snapped values for the
statusbar and the unsnapped ones for updating the info_window.
* app/display/gimpdisplayshell-callbacks.c
(gimp_display_shell_canvas_tool_events): applied modified patch
from Simon Budig which moves guide/grid snapping out of the
button_press/motion/button_release handlers and *always* snaps
the image_coords if the active_tool wants them to be snapped.
Makes sure that e.g. the brush preview is drawn at the right
place and fixes bug #134284.
Also pass both display_coords and image_coords to
gimp_display_shell_update_cursor().
2004-02-19 Michael Natterer <mitch@gimp.org>
* app/core/gimpimage-qmask.c (gimp_image_get_qmask): new function
which returns the image's QMask.
* app/core/gimpimage.c (gimp_image_channel_name_changed)
* app/gui/qmask-commands.c (qmask_query_response): use it.
2004-02-19 Michael Natterer <mitch@gimp.org>
* app/core/gimp.[ch]: added new signal "buffer_changed" and new
function gimp_set_global_buffer() which emits it when the global
buffer changes.
* app/core/gimp-edit.c (gimp_edit_extract): use it instead
of fiddling with gimp->global_buffer manually.
* app/gui/image-menu.c: connect to "buffer_changed" and set the
"Paste" menu entries sensitive. Fixes bug #134752.
(image_menu_update): changed accordingly. Also changed a lot of
buggy SET_SENSITIVE() lines which made menu items which work fine
on any drawable insensitive when there were no layers (instead
of no drawable).
* app/gui/edit-commands.c: added new macro return_if_new_drawable()
and use it instead of return_if_no_image() so we don't run
into assertions if there is no active drawable.
(cut,copy_named_buffer_callback): unfortunately had to introduce
two new translated messages about not being able to cut/copy when
there is no active drawable.
2004-02-18 Sven Neumann <sven@gimp.org>
* tile-cache.c
* tile-private.h
* tile.[ch]: removed trailing whitespace, added some newlines,
let tile_is_valid() return a gboolean instead of a gint.
* app/core/gimpimage-projection.c
* app/core/gimpimage-undo-push.c
* app/paint/gimppaintcore.c
* app/tools/gimpinktool.c: use the return value from tile_is_valid()
as a boolean.
2004-02-18 Michael Natterer <mitch@gimp.org>
Some code review:
* app/config/gimpconfig-deserialize.c
* app/config/gimpconfig-serialize.c: removed obsolete code which
made sure serialize_property()/deserialize_property() are only
called for properties of the correct class. We do it the right way
for quite a while now and clear the inherited function pointers in
gimp_config_iface_base_init().
* app/config/gimprc.c (gimp_rc_dispose): don't forget to chain up.
* app/base/gimplut.c
* app/base/lut-funcs.c
* app/config/gimpconfig-deserialize.c
* app/config/gimpconfig-serialize.c
* app/config/gimpconfig-types.c
* app/config/gimprc.c
* app/core/gimp-modules.c
* app/core/gimpbrush.c
* app/core/gimpcontainer.c
* app/core/gimpdocumentlist.c
* app/core/gimpitem.c
* app/gui/about-dialog.c
* app/gui/color-notebook.c
* app/gui/gui.c
* app/gui/tips-dialog.c
* app/paint/gimppaintcore.c
* app/paint-funcs/paint-funcs.c
* app/widgets/gimpcolorpanel.c
* app/widgets/gimpcontainergridview.c
* app/widgets/gimpcontainermenuimpl.c
* app/widgets/gimpselectioneditor.c: removed I/O includes from
files which don't use lowlevel APIs any more. Also removes
a whole bunch of G_OS_WIN32 special casing. Removed trailing
whitespace. Misc tiny cleanups.
2004-02-17 Michael Natterer <mitch@gimp.org>
* app/core/gimpviewable.c (gimp_viewable_calc_preview_size): argh,
take the resolution into account if "dot_for_dot" is FALSE, not
TRUE.
2004-02-17 Sven Neumann <sven@gimp.org>
* app/core/gimpimage-scale.c (gimp_image_scale_check): exclude the
size of the undo stacks so that the memory size is comparable with
the value displayed in the "New Image" dialog. Fixes bug #134512.
