* Move common things to the new .default section extensible for all jobs
* Reorganize order of job elements to be way more consistent and rational:
from general (pipeline/rules) to specific (job artifacts) in a liner way
* The order of rules was changed to be a bit more easier to understand
- Drops 'dependencies' which is redudant, and 'needs' have more options
- Drop apt-cache which lost use and it is sometimes troublesome to cache
+ First try to boost the use of ccache in all build jobs, when possible
+ Fix the problem of non-failing jobs introduced by GNOME/gimp!1171
Build for macOS on CI breaks with:
> ../gimp3-2.99.19/libgimpwidgets/gimppickbutton-quartz.c:210:45: error: use of undeclared identifier 'monitor_profile'
> space = gimp_color_profile_get_space (monitor_profile,
In case there are no fonts loaded (which may happen in particular when starting
with --no-fonts), we need to make sure that font is initialized to NULL so that
we can return the standard GimpFont object and not a broken pointer.
Thanks to Idriss for the additional diagnosis, which would hopefully complete
commit 5dc86277e6 and take care of #11219.
This commit uses definitions and references to reduce repetition of CI code.
Some people (like the GitLab devs?) prefer (almost imposes) this to be done in
multiple .yml's using 'worfklow:' but this makes the CI way more complicated
and very slow to read. So, let's just keep simple in one human readable file.
---
Other changes are:
- The custom native arch pipelines now they always generate a .zip package.
This helps to ensure that everything is in place and facilitate testing;
- The pacman update (-Syyuu) was removed since it is auto done by runners;
- The ccache config was moved to .yml. This does NOT affect local builds
since meson saves ccache by default for user in '%LocalAppData%/ccache'.
Previous code would only set the class member when we were actually
loading fonts, so we ended with code paths where the member was NULL.
Fixes:
> GIMP-CRITICAL: gimp_container_get_n_children: assertion 'GIMP_IS_CONTAINER (container)' failed
It should hopefully also fix#11219 (though I couldn't reproduce this
failure).
Resolves#11256
A GError variable was created but not passed to
file_open_with_display () when opening recent files.
As a result, if an error did occur the variable would be
used while it was still NULL. This patch fixes this
by passing a reference to error as the last parameter
of file_open_with_display ().
We compute the size of the mipmapped layers wrong. We assume that
width, height are always exactly a power of 2, which is the normal case
for dds images. See issue #11256.
Since other sizes are possible, we adjust this by computing the width
times height times bytes per pixel of the current mimmap level.
Resolves#11075
In the DrawableChooser pop-up, choosing an image without
also picking a layer or channel causes the pop-up not to
reappear on subsequent attempts to click it. This is because
the callback is only recreated when the drawable is updated,
and choosing just an image does not update the drawable.
For 3.0, we'll set the Okay button to insensitive unless a layer
or channel is picked.
I added a few improvements to how we load the Photoshop layers:
- When included from TIFF, we forgot to decrease the data_length for
the parts we already read.
- Also when included from TIFF, we need to skip the global layer mask
info.
- The block length for the layer data needs to be a multiple of 4.
- Catch an invalid data condition where block_len is greater than
total_len. This could cause a change total_len to a very large
value, so protect against that.
When loading Photoshop metadata from TIFF images, the size of the data
in the temp file was often incorrect.
This happens because run_mode is usually GIMP_RUN_INTERACTIVE when
run from within GIMP. In interactive mode, the last saved config values
take precedence over the ones set in the function call, which is not
what we want in this case.
We would prefer to use interactive here, because then we can show a
dialog with a list of unsupported Photoshop features that were found
inside the image.
For now, we will use GIMP_RUN_NONINTERACTIVE and accept that we can't
do that at this time. This way we will at least have the correct values
when processing the Photoshop data.
The clipping path values are all 2 bytes, but we were only reading in
1 byte for each value. In addition to that, the path_flatness_fixed
value was not converted to the correct Endian representation.
Since these values are usually 0 and get clamped to valid values, this
went unnoticed until now.
Fixes 2 such warnings:
> warning: format ‘%i’ expects argument of type ‘int’, but argument 5 has type ‘size_t’ {aka ‘long unsigned int’} [-Wformat=]
As well as:
> warning: conversion lacks type at end of format [-Wformat=]
And:
> warning: no previous declaration for ‘reset_output_port_struct’ [-Wmissing-declarations]
Fix unreported issue that new-style dialog using GimpProcedureDialog
does not display progress.
Change progress reporting to use app's status bar, for all scripts,
old and new style.
As before, report progress whenever a script calls a PDB procedure,
by putting the procedure name in the status bar.
Eliminate progress bar widget from old-style script dialog.
Since it was failing on Wayland, and since it is inconsitent
with new style dialog using GimpProcedureDialog.
Resolves#11246
On Windows, a CRITICAL was being repeatedly thrown if you
didn't have a monitor profile set. This is because we were trying to
pass a NULL value to g_utf16_to_utf8 () whenever we needed to
convert a GeglColor.
This patch fixes it by checking if the monitor profile filename is NULL
before trying to convert it.
Resolves#11244
GtkRadioButtons can also have the .text-button CSS class set. This class was overwriting
the radiobutton background color, making it appear dark and box-like.
This adds a specific radiobutton.text-button definition to ensure the background color
is consistent (as we had to do with checkbutton.text-button)
It is broken and throws critical, see #10596.
Expedient to hide it instead of fixing it, to not block 3.0rc1.
As discussed in #10652, a user can simply restart GIMP instead.
"Refresh Scripts" is a poor design, a bandaid.
There are plans for better alternatives to install/remove scripts.
The MR for #10652 shows the reason it throws critical,
and could fix it,
but completing the MR requires time we don't have.
Note the MR is not on these files, but on core plugin machinery.
The change is to extract functions to a separate file,
and still build them, but not call them or link them.
They also are not translated while not being called.
Eventually, the extracted files can be deleted (or restored when they work.)
There is other related cruft that needs deletion in scheme-wrapper.c.
Add canonical/idiomatic functions call-with-output-string,
call-with-input-string, any->string to the library init.scm.
Since they are useful, especially in test scripts and frameworks.
Remove their definitions from test scripts and test frameworks.
Also comment out failing test in test script string-escape.scm.
Redefine call-with-output-string etc. in the test script: test9.scm
Other fixes to the test script.
Improve logging of no memory condition for string-ports
Fix bug freeing ports that have been closed.
When ports are closed, string-port code was treating them as string-ports.
Fix#11040 and #9495 bugs with unicode in string-ports
Fix other unreported bugs with allocations and garbage collection of string-ports.
Obsolete open-input-output-string, a string port of kind input-ouput is no longer supported
and the function actually opens a string-port of kind input which cannot be written.
Comments and Gimp style in the original TinyScheme code, now in string-port.c
Revise tests of string-port.
Update TinyScheme README: Discuss more thoroughly the ancestry, and changes from upstream.
Extract pseudo class string-port methods to separate file.
No functional changes intended.
Style changes not complete: anticipate further bug fixes and styling.
This is not seen on all distributions (mostly on Debian-based ones in my
experience), but some distributions install libraries in a second-level
directory under prefix (e.g. lib/x86_64-linux-gnu/ instead of lib/ or lib64/)
whereas our prefix-guessing code for relocatable builds harcoded moving up from
1 level.
This new heuristic will assume that if the leaf directory is neither bin/ nor
starting with `lib`, then it's likely a multiarch folder and we must move up
once more to find the prefix folder.
This should also fix the problem encountered by Bruno for the current work on a
potential official AppImage.
We dereference "file" twice in gimp_prop_file_chooser_button_callback(),
once in the GIMP_IS_PARAM_SPEC_CONFIG_PATH () condition and then
at the end of the function. This is likely due to a similar structure in 2.10,
which did not have a second condition.
As a result, any Gegl filter which had a file property would crash if set from
a GUI. The first g_object_unref() is removed so we only dereference "file"
once.
Resolves#9795
(Solution provided by Andreas G, @uriesk)
Some system themes provide errant values for tooltip
box-shadows which cause flickering effects.
This patch defines the tooltip box-shadow to 0px to
prevent the issue.
Resolves#11054
NDE Filters clip to the size of the layer when
applied. Group layers get their dimensions from
the largest layers under them, so an empty
layer has dimensions of 0 by 0. This means a
filter applied when the layer group is empty will
be clipped to 0 by 0 as well.
This patch adds code to refresh the filter's crop
whenever the group layer is resized by a
layer being added or removed.
These commits introduced sensible drawbacks in our CI file organization.
Some of them I already commented in the MR but was "ignored" back then
probably due to the complexity of the MR. So, the let's revert:
- Unecessary repetition of libomp installation (4 times, now 1)
- Unecessary repetition of submodule initialization (6 times, now 4)
- Creation of new variables in the hope of more clarity. But,
they aren't consistent with the gimp-web-devel instructions
nor with other systems variables. This is terrible to undestand.
Please, let's not reinvent the wheel since this is a bad example to
packagers, since they tends to isolation and overly-engineering.
Also, made some little adjustments that I should have done in previous MR.
---
This MR was splitted. Even more simplification will be done in the future.
...to accept GeglColor paramaters instead
of GimpRGB. This function is used to draw
the layer/channel color tags.
Note that a temporary GimpRGB was left
to use with gimp_rgb_composite ().
It will be replaced once that function is
also converted to use GeglColor.
