Regression in commit 0af126fec7. We added
support for some "test cases/" stuff that actually relies on test files
being a symlink, but when testing the image builder, we copied the meson
repository contents into the docker container without telling python
that it is in fact super important to copy symlinks as symlinks.
As a result, the tests themselves ran fine when merging, but caused the
image builder to thereafter fail.
They do not appear to have 20 in their repos anymore, and no traces can
be found of it in the history, as usual. They do have 11, 17, and 21.
Last time we chose one randomly and hoped it doesn't keep changing
value. This dismally failed: 20 was upgraded to 21. It looks like 17 may
have some staying power.
They do not appear to have 15 in their repos anymore, and no traces can
be found of it in the history, as usual. They do have 11, 17, and 20, so
choose one randomly and hope it doesn't keep changing value.
In this case, PEP 668 was created to allow a thing that Debian wanted,
which is for `pip install foobar` to not break the system python. This
despite the fact that the system python is fine, unless you use sudo pip
which is discouraged for separate reasons, and it is in fact quite
natural to install additional packages to the user site-packages.
It isn't even the job of the operating system to decide whether the user
site-packages is broken, whether the operating system gets the answer
correct or not -- it is the job of the operating system to decide
whether the operating system is broken, and that can be solved by e.g.
enforcing a shebang policy for distribution-packaged software, which
distros like Fedora do, and mandating not only that python shebangs do
not contain `/usr/bin/env`, but that they *do* contain -s.
Anyway, this entire kerfuffle is mostly just a bit of pointless
interactive churn, but it bites pretty hard for our use case, which is a
container image which is fortunately tested before deployment, so
instead of failing to deploy because of theoretical conflicts with the
base system (we specifically need base system integration...) we fail to
deploy because 5 minutes into pulling apt updates at the very beginning,
pip refuses point-blank to work. I especially do not know why it is the
job of the operating system to throw errors intended for interactive
users at people baking "appliance" containers who cannot "break" the
system python anyway.
Fix this by doing what Debian and Ubuntu should both have done from the
beginning, and opting containers out of this questionable feature
entirely.
Note that CI images may still not actually complete their build/test
cycle and be updated, because e.g. LLVM 16 issues tracked by #11642 or
glib ASAN issues tracked by #11754.
The bionic image is really old and mainly exists to test that Meson
itself still works on really old distros (and really old python).
Ideally we'd avoid depending too much on it.
We can get a very modern pypy3 automatically this way, and potentially
use it for more stuff too.
It's already run on other distros. This one fails though, due to missing
debug info.
```
valgrind: Possible fixes: (1, short term): install glibc's debuginfo
valgrind: package on this machine. (2, longer term): ask the packagers
valgrind: for your Linux distribution to please in future ship a non-
valgrind: stripped ld.so (or whatever the dynamic linker .so is called)
```
It doesn't seem possible to have this work out of the box. The debuginfo
packages aren't reliably available, and debuginfod servers -- even if
they worked, which they apparently don't -- would not help anyway since
old version pruning can result in symbols disappearing before the image
is rebuilt, and thereby causing failure.
It's not really critical to test this, since as mentioned we already
have coverage of Meson's side in other distro ciimages.
From the Zen of Python: "Explicit is better than implicit."
As it turns out, it's no longer a safe assumption that pip uses
setuptools??? Well, anyway, install it properly regardless.
The coverage report was always the final section of the main test run.
This made it hard to scroll around and find exactly what went wrong --
particularly as not everyone realizes that coverage isn't part of the
test run, but also because the output from coverage is... excessively
long.
This mirrors what we do in our other workflows.
After a recent CI image builder update successfully ran the tests, but
didn't run the cross tests, it updated the image that then got used by
the regular CI cross tests. Somehow this resulted in a bunch of tests
now failing because zlib could not be picked up. We probably dropped a
transitive dependency somewhere. Anyway, it's correct to explicitly
specify it if we need it.
We've never used it for anything, it was originally added for #3776 but
that never got finished so it's just a waste.
This also prevents successful regeneration of the build image, because
nim is not available for Ubuntu rolling. It's available in 20.04 and
22.10, but vanished in between for reasons best known to Ubuntu.
This enables the fortran tests for Azure.
We only test on x64, because:
- ifort isn't arm64 compatible
- x86 may in theory exist, but Meson reports it cannot compile
executables
The pip package is for python 3.6, but installs pip for all versions of
python. Apparently. Including python 3.7.
So do all other packages, especially the ones where it doesn't work but
pip thinks it is installed anyway. Force a reinstall.
This has been removed as an explicit package in impish. It seems that
having pkg-config installed and adding arm as an arch will cause it to
be generated automatically
Set MESON_CI_JOBNAME for all CI jobs which run project tests.
(Note that ${{ github.job }} is the literal job.id used in the yaml, not
any name given to the job with job.id.name, and so is the same for all
matrix entries, and thus not suitable for our purposes there).
Remove hard-coded framework test skip logic in skippable(), instead
annotate test.json with environments in which skip is expected.
(Mainly this is done with by testing the value of MESON_CI_JOBNAME now
set for linux jobs)
There hasn't been any such package since the original addition of a
gtk3 version of wxgtk... back in 2017. The "new" wxgtk2 package provided
a virtual provides ever since, so people still depending on "wxgtk"
would get the old gtk2 version. This virtual provides got dropped today,
resulting in the package being uninstallable.
Resolve the provides to its canonical name, thus making it installable
again.
Some changes:
* Set HOME to /root, since github mounts its own HOME and 'wine'
(because of permissions) and 'dub' (can't find packages) don't
like that.
* Remove the seccomp option, doesn't seem to be needed.
Eli Schwartz
Xavier Claessens
Eli Schwartz
Ralf Gommers
Nirbheek Chauhan
Josh Soref
mattip
Mark Bolhuis
Matthias Klumpp
Dylan Baker
Christian Clauss
Daniel Mensinger
Jon Turney
Luca Weiss
Christoph Reiter