generated by matlabs blocproc function
Based on the suggested solution by Massimo, we should not compute the
remaining pixels per line but only at the end of the whole block, be it
tile or scanline.
I have not been able to find bw or palette examples that use load_separate
instead of load_contiguous, so that case is not tested. But, it also
doesn't make much sense to have planar when you have just one plane.
In a previous commit 1, 2 and 4-bit B/W images were converted to grayscale.
However, it seems that there is more of a use case for these images to be
handled as indexed, even though technically they can be considered
grayscale.
Also, the only way to export these images again in the same format, is to
have them as indexed.
So, let's change this back, so that these kind of images will be opened
as indexed. With a reminder that in the future we could add an option
at loading time where the user can choose whether they prefer it to be
loaded as indexed or grayscale.
We use grayscale mappings, that we moved in the previous commit, to
add a palette to these grayscale images.
In preparation of using some of them earlier, we move the variables
used for grayscale mapping of 1, 2 and 4-bit per pixel images to the
top of file-tiff-load.
I guess I missed this one as I was not building it locally.
Fixes:
> In file included from ../plug-ins/common/file-jp2-load.c:86:
> ../plug-ins/common/file-jp2-load.c: In function ‘jp2_class_init’:
> ../libgimp/stdplugins-intl.h:42:22: error: ‘set_i18n’ undeclared (first use in this function)
Hence avoiding the stderr messages. These are going to be localized with
centrally installed catalogs "gimp*-std-plugins", "gimp*-script-fu" and
"gimp*-python".
We now handle core plug-in localizations differently and in particular,
with kind of a reverse logic:
- We don't consider "gimp*-std-plugins" to be the default catalog
anymore. It made sense in the old world where we would consider the
core plug-ins to be the most important and numerous ones. But we want
to push a world where people are even more encouraged to develop their
own plug-ins. These won't use the standard catalog anymore (because
there are nearly no reasons that the strings are the same, it's only a
confusing logic). So let's explicitly set the standard catalogs with
DEFINE_STD_SET_I18N macro (which maps to a different catalog for
script-fu plug-ins).
- Doing something similar for Python plug-ins which have again their own
catalog.
- Getting rid of the INIT_I18N macro since now all the locale domain
binding is done automatically by libgimp when using the set_i18n()
method infrastructure.
- Set the "gimp30-python" Gettext domain and bind it to the proper
locale directory as installed by GIMP.
- Localize various strings with gettext.
- Remove calls to self.set_translation_domain() in
do_query_procedures(). This is technically wrong right now but I am
going to get rid of the menu item localization for plug-ins done by
the core.
We were using parameter iter in metadata_dialog_add_tag and
metadata_dialog_add_translated_tag.
However, iter is only ever set inside metadata_dialog_add_tag by calling
gtk_list_store_append. So, there is no need to pass this parameter around.
For this reason, let's remove parameter iter from the above two functions
and replace with a local variable.
Also fixed quotient to only take two arguments. Applied minor
optimization in execution of quotient, remainder, and modulo.
From revision 122 of the TinyScheme repository in SourceForge.
Instead, make outer script-fu-server register callbacks with inner scheme-wrapper.
Why? the inner scheme-wrapper should not know about the outer script-fu-server.
The change will allow a future, smaller scriptfu shared library,
that does not contain networking code.
We want a scriptfu library shared by separate script-fu-server,
future gimp-scheme-interpreter (!647), and script-fu executables.
This change does not alter observable functioning of the script-fu-server.
Except that I also changing the logging by script-fu-server,
so that I could test the changes.
I put a test plan in the comments.
Removal of the snprintf define was also part of the UCRT Windows patch
in MINGW. Although it builds fine without this change, there is no need
anymore to redefine snprintf. In addition, I seem to remember that
_snprintf had a shortcoming.
So let's remove the Windows specific code, since it works fine without it.
The UCRT environment is a successor to the MINGW environment on Windows
with a more modern C runtime library.
Building under UCRT we get this error:
../../gimp/plug-ins/file-raw/file-darktable.c:420:7: error:
function '__p__environ' is initialized like a variable"
To fix this we use the relevant part of patch 6 from:
https://github.com/msys2/MINGW-packages/tree/master/mingw-w64-gimp
when Offset and Dimensions Bigger than File Size.
If we didn't read the exact amount of bytes, the whole image would be
filled with white.
Let's change this, so we read as much bytes as we can, and fill the rest
with white.
Added GError parameters to all static file write functions, check for
failed writing and return an error if that is the case.
Additionally, at end of file make sure the last dummy byte is written to
make chunks an even size.
The animation playback now follows user preferences for checkboard
rendering.
Also fixing a small bug there, as the GIMP_CHECK_SIZE_SMALL_CHECKS was
the wrong type (it needed a check type, not a check size parameter).
Right now, the various Gimp*Preview classes have code initializing their
transparency colors correctly. Some plug-ins, such as tile-small are
directly using a GimpPreviewArea and need to be initialized directly by
the plug-in.
I also realized that several plug-ins (such as gradient-flare) are
rendering their own alpha checkboard, despite using a GimpPreviewArea
(maybe was it code from a time before this class had support for alpha
checkboard rendering?). So this would have to be improved later.
- Changed all static read functions to have a GError parameter, use a
parameter for the value read, and return a gboolean that will be FALSE
when reading from file failed.
- Check the return values of all read functions and set GError when
needed.
- Added more error checking, like comparing real filesize with what the
header tells us, check for valid speed and number of frames.
- Added some gdebug statements for easier debugging.
- Don't assume that all FLI/FLC writers followed the specs and wrote an
even number of bytes per chunk.
- Skip "frames" that do not have the FRAME type (in most cases this is
a PREFIX chunk).
Improvements to the FLI loading and exporting plug-in:
- I added a GError parameter to all public read/write functions.
- All public functions now return a gboolean result to show if they
succeeded or failed. Before most functions were void. For the non void
returns (the two fli_write_color functions) I added a gboolean
parameter.
- Do cleanup if we fail to read a frame. If at least one frame was read
successfully, we will keep the incomplete image/animation.
- Change name of layer to include the delay in ms.
Why: MR !389 changed the signature of PDB procedures to return GStrv instead of (int, GimpStringArray)
This changes handling of results from a call to such a changed signature, in a Scheme script.
The symptom was a cryptic TinyScheme "Error: car: argument 1 must be : pair"
Most changed procedures are named like "-list".
I did a cursory grep to look for other instances.
Updated porting guide doc.
The unsigned long type in this plug-in is always used for reading/writing
32-bit unsigned integers, so let's change it to guin32 to not get
confused.
Changed unsigned short to gushort, unsigned char to guchar.
Why: in v3 the signature for gimp-edit-foo PDB procedures changed to support multilayer selection.
This finishes the task for ScriptFu scripts in the GIMP repo.
This is not a complete list, since some calls were changed incidentally by prior commits.
You can grep for "edit-copy" etc. in script-fu/scripts to find instances.
This also makes incidental changes, to script calls to plug-in-tile,
which now also has a changed signature to support multilayer.
The changes are simple code transformations.
The changes to gimp-edit-paste calls are more complex,
a mixed bag of a few lines transformed to more lines.
I did not try to understand the larger logic of the changed plugins.
I did not test the changed plugins functionally.
I used a harness to call each changed plugin with improvised parameters,
only checking that the test plugin did not throw runtime errors,
not checking that they produced correct images.
As noted in the issue, this might be undone if the original signatures
for gimp-edit-foo are restored as convenience functions.
Why: this commit fixes PATTERN using code that commit 708f3228 fixed for BRUSH.
Importance:
There doesn't seem to be any issues it resolves.
No code in the GIMP repo seems to call this procedure.
I don't know whether third party plug-ins rely on this procedure.
Anyway, fix it so a brute force test works.
This change is part of multilayer select feature of v3.
Only fixes ScriptFu plugins, not Python or C.
Only fixes calls to file load/save procedures and other plugins already multilayer capable.
Any future PDB procedures changed to have multilayer signature will require changes to all callers.
The new code is verbose: "drawable" => "1 (vector drawable)"
Issue 5919 proposes a change to the PDB that would make this less verbose:
"drawable" => (vector drawable)" (without the number of drawables)
Tested that changed procedures will interpret without syntax errors, on one path.
Did not test each changed procedure for functional correctness.
We have some examples of DDS images that have two 16-bit channels that
we could not load.
This commit adds checks to see if we need 16-bit channels, and conversion
of the input to a correct 16-bit red, green, blue or alpha channel.
To be able to use certain variables we use in our DDS loader earlier,
we move the computation of these variables up.
We also add checks to see if the mask needs to be for 8 or 16-bit, since
there are DDS images with 16 bits per sample.
This was a long standing issue where the image was not printed at the
position as selected in our print preview.
The problem being that we do not take into account the dpi of the print
context we are printing to, but instead always assumed 72.
This is the fix that was suggested by Massimo.
Small text tweaks done too:
- Some question marks removed to be more consistent in the tooltip
style.
- The "save-transparent" only applies to *completely* transparent pixels
(not partially transparent).
Also changing "RGB->YUV" by using a real Unicode rightward arrow encoded
in UTF-8 since GTK functions (such as `gtk_widget_set_tooltip_text()`)
explicitly uses UTF-8 encoded strings as argument. This renders much
more nicely than an ASCII-made arrow.
As mentioned by Massimo in issue #6618, part of the problem there is an
integer overflow when using large size images when computing the offset
in pixels.
Let's fix our part of the problem by casting to guint64.
Besides that, also use casts to correctly compute progress for very
large images.
