It was only used for the gimp-win-a64 job and was coming from MSYS2 repository
which already dropped it:
a98352b2ba
The first patch is still needed as the upstream fix is meant to appear in clang
18 according to bug report, yet our CI still uses clang 17.0.6. See:
https://github.com/llvm/llvm-project/issues/51286
Since !1171 we can use the CI scripts locally, but the initial implementation,
after some commits, revealed to be troublesome. So, now reworked to be simple.
As noted by Jehan, the way of defining the (many) GIMP versions can change.
So, let's take the GIMP version directly from the generated config.h file.
Also, changes the declaration of some variables to conform with "best pratices".
This list is referenced in the gimp-web-devel site, and my initial
implementation, in terms of ease to use, was a throwback.
Also, make the CI code more simple without changing the list file.
The rest of the time, we want to use HEAD of their main branches.
The last ctx update seem to have broken the CI build on Windows. While we'll
have to fix this, let's do this after we release GIMP.
Anyway we obviously don't want to use unstable code for actual point releases,
so it's a good thing that this made me think of adding such rules.
Here, the instructions was hard to find in the middle of maintainer stuff.
Also, be able to build GIMP without dev-oriented distros is something very
desirable, and the gimp-web-devel is the suited place fot it.
Aaaargh! I think this was the main issue which would explain why the problem was
mostly happening on the packaging job. In this job, we were still consistently
calling `pacman --noconfirm -Suy` (i.e.g "Synchronizing package databases")
which we had stopped to do for the CI builds as that was highly increasing
chances of locking pacman's process.
Fixing a case where the script continued after a failed `ninja`. After further
testing on how `set -e` works, it looks like a toplevel `false && true` command
will fail the script (which is what we want), but it won't when inside an `if`
block!
Breaking it down into 2 separate calls fail the script correctly though. So
let's do this.
MSYS2 pacman gets randomly stuck on Windows/Aarch64. The actual issue is still
being investigated by upstream projects, though anyway it's bad for us right
now, to the point that there are discussions to remove Aarch64 support from the
Windows installer (whereas it just got added recently!) in #10729.
This is an attempt to a workaround. Instead of getting stuck forever and waiting
until the whole job times out (per Gitlab CI settings), I time-out (after 3
minutes) the pacman command within our script and try again, up to 2 more times.
Hopefully one of the calls would succeed.
I also send a SIGKILL through the timeout (though I have no idea how signals
translate to Windows processes) and run again taskkill after this, which may
seem overkill. Interestingly I get output for both, which seems to indicate that
the kill succeeds in both cases (because of several processes?).
Anyway clearly it's a bit of random code not completely understood, but the
inability to test this all locally clearly doesn't help so it's good enough for
the time being.
See: https://github.com/msys2/MSYS2-packages/issues/4340
While this is the job of `set -e`, putting parts of the script in shell
functions hid error returns in the build-deps-msys2 script.
Cf. #10729.
For the build-gimp-crossroad script, let's just add `set -e`.
Cf. #10725.
The list contains, in addition to the GIMP deps, all babl and GEGL deps.
So, let's use it.
Also, nothing new under the sun. We already use a unique list on the
Windows build instruction page (for stable branch) and do the same to
Debian image job.
As suggested by Jehan, the manifest will be configured by meson.
Of course, for CI compliance, this pushes the job to a further stage:
'packaging', which is what Flatpak is about, after all.
The distribution job name has also been changed, because when Store job
is merged, this will be the new nomenclature of the distribution jobs,
which was not changed since !1171 by lack of time.
There is no point in keeping .patches in our source since we aren't applying
them and there are "backup" MRs in GTK: GNOME/gtk!3275 and GNOME/gtk!4432.
Now, the script reads the major.minor version automatically from the main
meson.build file, reducing maintenance work, which is already especially
high regarding flatpak because of the natural hashes updates etc.
Also, a small organization made to make the code more humanly readable.
As consensus on Windows contributors, it is more useful for most users
the generation of the desktop shortcut by default.
Now, users who don't want it should uncheck the option in custom install
or manually delete the shortcut after a full install.
The 'jhbuild' build of GIMP (Windows only) was added in 2013 and then maintained
by only one contributor (it even offered unnoficial nightly builds) until 2017,
when it never received a new jhbuild-specific commit again.
After so long, update it to keep up with current development status is quite hard,
so it is reasonable to remove this already dead version of GIMP.
* Fixes an ARM .patch that stop to working after !1171, causing job fail
* Changes the logic of system upgrade (pacman -Syyuu) to be less redundant
This also improves consistency with the Debian jobs
- Fix "no icons" errors generating loaders.cache with .cmd (CI-Cross).
! This is a sub-optimal fix, but it's better than a useless build.
- Fix "no interpreter" errors generating .interp with Meson (CI-Native)
and generating .interp and copying .typelib with .cmd (Local-Native).
! This is a sub-optimal fix to Local-Native, but plug-ins will work.
- Fix "no iso" error copying iso_639.xml with Meson (Local-Native).
The Inno installer scripts contents (only 3 files: files, gimp3264 and
32on64) and filenames have been organized, making them much easier to
read, and slightly less hardcoded so less prone to being misunderstood
and pervasively receiving packaging stuff.
Just to be clear, one more time: the Inno installer (or future MSIX)
scripts never should be the center of attention. This "installcentrism"
caused a domino effect of partially "abandoning" the packaging, build.sh
and the meson scripts, which explains the existence of this MR...
(Some things still hardcoded since wildcards in Inno are very limited.
Also, the rational ordering principles of this MR were not applied since
these scripts are heavily based on the x86 .zip package and changing the
order of things here, according to my tests, breaks things quite easily)
The CI crossbuild job now use the same main script (and the same
sub-scripts of linking and debug generation) of the native CI for
packaging. These unified scripts greatly facilitates maintenance.
The crossbuild deps script is now more consistent in relation to the
native one. As this is polished, the cross one is now polished too.
A crossroad gimp build script was created with a more clear code.
Also, finally make the script for packaging only, removing build step
stuff that shoudn't be here (glib-schemas generation) since this causes
disparity with the Local native build; and adding some packaging
decisions that shoudn't be in the installer scripts (eg. specific
folders of ghostscript, glib; no share/themes), which also facillitates
INNO (and future MSIX) maintenance.
The local builds now can use the same script of CI for sake of time.
The compatibility layer is very simple and makes the script more
used/tested, therefore more reliable.
The local builds now can use the same script of CI for sake of time.
The compatibility layer is very simple and makes the script more
used/tested, therefore more reliable.
Also, remove some redundacy on the code for better maintenance.
Now, we have only ONE list applicable for build-deps-crossroad.sh,
build-gimp-msys2.sh and package-gimp-msys2.sh, instead of hardcoding
3 times. This unified list greatly facilitates maintenance.
Debian changes:
- Since autotools has gone, we don't need to specify 'meson' in the
debian job and others.
- The "INSTALL_PREFIX" was renamed for the more usual "GIMP_PREFIX" and
the meson sintax of Debian jobs was also updated.
Windows changes:
- Then, clarify that the win64-nightly and win32-nightly jobs are, in
fact, in the 'packaging' step, since we don't really "distribute" GIMP
in .zip and the commands are almost the same of the packaging .SH
script, without scripted optimizations for Inno Installer (or future
.MSIX), crucial for distribution.
- We don't need to specify "native" sufix in any build since they are
the rule and cross builds are the exception.
General changes:
- The job names was changed to be more consistent and in accordance
with the folders present in the artifacts.
- The 'nightly' sufix was removed from the Inno Windows Installer job
and others, since this doesn't reflect the real build frequency.
- The scripts filenames are altered to stay "in order". This is not
essential but ultra convenient since it is easy to view and search.
(The -uni suffix is explained in a further commit)
- All artifacts names now have the commit to avoid apparently duplicate
files when downloading same step artifacts from different projects.
- Finally, rearrange the order of jobs rationally: first the OSes and
archs (from the most free and modern to the most closed and legacy),
then the stages (from 'prepare' to 'analysis'), ending with the
frequency of jobs (from the most frequent, called at each push, to
the least/weekly).
Overall, this changes, although difficult to review at the first
sight, will avoid in the future quite "dumb" issues like:
GNOME/gimp#10195
This is in part a port of commit 6f921b27bb from gimp-2-10 branch, except that I
could not easily cherry-pick because too much had changed already in the master
branch.
There are also a bunch of additional changes in some other glue scripts.
Also some rules which were in the 2.10 branch don't apply to the main one, such
as Python being copied from the x86 binaries, since clang-aarch64 has Python3
(unlike Python2).
Jehan