...to fully use and return GeglColor.
Also, fix gimptext-parasite sending a
GimpRGB to create a GimpText instead of
the now required GeglColor, and update
documentation in gimp_checks_get_colors
to reference GeglColor instead of GimpRGB.
Resolves#11092
Same fix as in 2b3b2af2, since Blinds and Jigsaw
were the two plug-ins which used GimpAspectPreview
before the port to GimpProcedureDialog.
After the port to gimp_procedure_dialog_get_drawable_preview (),
the preview widget was being sized to the original drawable's dimensions
while the drawable was being redrawn with only its thumbnail dimensions.
This created a scrollbar that showed unallocated/overwritten buffers.
To fix, we update the preview size and boundaries whenever the preview
is changed. This removes the scrollbar and returns to 2.10 behavior.
In addition to the port, mnemonics were added to both file-pdf-save-multi
and file-pdf-save's GUI parameters.
Some runtime warnings were also fixed, such as trying to use a NULL string
parameters to make a file and trying to access properties from file-pdf-save
that don't exist in file-pdf-save-multi.
As shown in a video for issue 11087, some system themes
may add a white box-shadow on the side of unselected
tabs. This patch forces box-shadow to "none" to prevent
this.
Resolves#11057
Per Idriss Fekir's research, data from GBytes
should be written before the GByte is freed.
Perhaps due to luck, the data persisted on
Windows but was freed too early on Linux.
This patch should resolve the issue.
Resolves#11083
Per bootchk, changes in function calls to
palette_entry_get_color () and
gradient_segment_get_left_color () were
broken in Python scripts due to GelColor
being used post Color Space Invasion.
This patch fixes the calls and returns.
More descriptive name for script-fu-interface-dialog
Delete SF flag for gimp_ui_init already called: gimp_ui_init can be called often.
FUTURE: might be faster to call just gegl_init when mode is non-interactive.
After the color space invasion, the
Clip Warning filter no longer had color
areas for its shadow, highlights, and
bogus color properties. This patch ports
them to GeglColor so the widget can be
created correctly.
GimpRGB structs are also converted to
GeglColor in GimpPanedBox and two
plug-ins.
Otherwise we are not in a valid position anymore and further loading the XCF
fails. This may happen when updating a GEGL operation (in particular removing
effect properties, or renaming them).
The two threshold color pick buttons needed to have
a GeglColor parameter for their GimpPickerCallback
rather than a const GimpRGB.
This also fixes a crash caused by trying to free the GeglColor
rather than unreferencing it.
Without this, there is no easy way to quietly check for existence of a resource by name.
Needed when plugin authors can name defaults for resource args by name.
Add tests to script-fu/test/tests
Mainly fixing GimpRGB comments and
parameters that are unused (or in unused
functions).
GimpCircle and GimpGradientChooser
have color conversions ported to use
GeglColor exclusively.
Since commit 10b96b1025, we had buffer leaks as soon as we ran filters.
gimp_drawable_get_buffer_with_effects() was either creating a new buffer (when a
drawable had filters) or not. And calling code couldn't know which is which.
Code needs to be consistent. Either we return a new reference to take ownership
to, or not. So let's change to add a new reference to the drawable' buffers so
that it always returns a new reference to free.
So far, we did not consider that group layers could also be part of
a set of layers that have clipping set. This means that clipping groups
can be nested.
To support this we have completely rewritten the layer parser to detect
all clipping groups and mark them in a first pass. We now go through
the layers from the last (top) layer to the first (bottom) in this
first pass, because that makes it easier to detect the start/end
pairs of a clipping group.
In the second pass we handle the clipping groups first before adding
layers, except when a regular group is also the start of a clipping
group. In that case we need to handle the clipping after creating the
regular group, or we get an incorrect order.
Unsupported psd layers were dropped when loading psd images. This
causes problems when handling psd clipping groups, see issue #8453.
PSP also does not drop unsupported layers, so let's keep the layers
and accept that they will usually show up empty. This will help in
fixing clipping groups in a next step.
While we are at it, also remove the warnings that are only shown when
define CONVERSION_WARNINGS is TRUE (defaults to FALSE). We already
have a better way to show unsupported features, so at least for the
unsupported layers it is not needed anymore.
There are still some CONVERSION_WARNINGS warnings left over in the
blend modes conversion, which do not have a similar warning in our
unsupported features dialog.
* Makes the wording more concise and "first contributor-friendly"
+ Adds info about submitting patches
* Makes the explanation of use cases implicitly mandatory
+ Adds discuss info to facilitate the redation of use cases
