testing/ruff: enable package for all archs

Hello,

Couldn't we make this package available for all archs ? I see some archs have been disabled because of failing tests. Couldn't we just disable tests on those archs ? I can see that the GitHub release includes those archs so it should be possible, right?

I have a package (testing/py3-xsdata) which added ruff as a dependency (!58432 (closed)) and it would be annoying to drop some arch compatibility.

mentioned in merge request !58433 (closed)

Couldn't we just disable tests on those archs ?

Depends on the kind of test failures, but that the tests are failing does indicate something is not working as expected. Just disabling test failures defeats the purpose of having and running those tests in the first place.

I understand. Anyway there should be a way to make those tests successfull, given that official releases from GitHub includes those archs.

assigned to @WhyNotHugo

mentioned in merge request !58432 (closed)

!59017 (merged) solved ppc64le.

!59132 (merged) is ready to merge and solves riscv64

Thank you @WhyNotHugo. Do you know what is blocking on the remaining archs?

Tests fail on the 32bit architectures, I have no idea why.

It might be worth asking upstream.

Ok then. Should I do it or do you want to take care of it?

Please do if you can. Honestly, I'd prefer a way to locally reproduce the failure locally (qemu in 32bit mode?). I suspect that upstream would appreciate this too for testing/debugging.

We can provide an LXC container if interested.

@kdaudt yes please, could you provide a LXC container?

You should be able to ssh into root@172.16.26.26, assuming you have connection to our vpn. I've used your gitlab ssh keys.

Ok so the failing test is display_default_settings. I'll open an issue upstream.

Should be fixed now, let's wait next ruff release

Ruff has issued a new release. @WhyNotHugo could you update the package and check if it work on all archs please?

I've pushed a draft MR to confirm that CI passes for the new architectures: !62439 (merged)

CI still fails on 32bit architectures: https://gitlab.alpinelinux.org/WhyNotHugo/aports/-/jobs/1309514

Identified the problem, opened a new issue on upstream.

OK, seems fixed now. This time I actually tested it on my x86 container before saying that.

Next release (in a few days) should be good to go.

Most other architectures now build fine: !62439 (merged)

Only s390x remains disabled.

Have you identified why not on s390x? I don't understand the comment "nix crate".

@kdaudt is it possible to get a s390x container please?

https://crates.io/crates/nix

The nix crate (a dependency) doesn't build on s390x.

From the project page it seems it does. s390x is listed in their tier 2 support. I'll need to check what's going on.

nix has built on s390x for years. only some function usage on musl didn't work, but a few separate instances of that were also fixed in the latest releases. ruff still uses 0.26.4

porting the dep tree to 0.28 will probably fix that

So nix is used by dependency clearscreen. Here is the issue to upgrade this dependency.

dependancies have been bumped for clearscreen and then for ruff. Let's wait for next ruff release and we should be good to go.

Don't mean to steal the earned last part of this, ruff 0.3.7 is now released: https://github.com/astral-sh/ruff/commit/2e37cf6b3b4b873ad9ffe9728bccdf134862b768

Great work @raspbeguy, thanks!

!64043 (merged) should work, but the python 3.12 rebuild is on-going. I'll retry CI when the rebuild is complete.

s390x is now building properly! Thanks for finding and reporting all the issues upstream.

mentioned in commit WhyNotHugo/aports@c9806dee

mentioned in merge request !62439 (merged)

mentioned in commit WhyNotHugo/aports@c6b9d9cb

mentioned in commit WhyNotHugo/aports@bea4fdb0

mentioned in commit WhyNotHugo/aports@9715dab2

closed with merge request !62439 (merged)

reopened

mentioned in merge request !63007 (closed)

mentioned in merge request !63125 (merged)

mentioned in merge request !63127 (merged)

mentioned in commit WhyNotHugo/aports@3622eca9

mentioned in merge request !64043 (merged)

mentioned in commit WhyNotHugo/aports@8746aad6

closed with merge request !64043 (merged)