-
Notifications
You must be signed in to change notification settings - Fork 703
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
{devel}[SYSTEM] Reframe v4.6.2 #21307
Conversation
…above, including 3.12
…asyconfigs into reframe_462_system
Test report by @bedroge |
@boegelbot please test @ generoso |
@casparvl: Request for testing this PR well received on login1 PR test command '
Test results coming soon (I hope)... - notification for comment with ID 2326583819 processed Message to humans: this is just bookkeeping information for me, |
Test report by @boegelbot |
Ah, yes, we already found out in the other PR that both our test clusters are missing some os dependencies. So this failure should not be considered a problem, it is correct behavior of the EasyConfig. |
Test report by @casparvl |
Test report by @casparvl |
The failure of
I would have expected easybuild to just use the info it obtained here:
to conclude the python version isn't up to the specified requirement. Oh well, that's not an issue of this EasyConfig PR. |
Ok, I investigated a bit more with
Why is https://github.com/easybuilders/easybuild-easyblocks/blob/cc5b95dab1aabae2a64360c492a4f068f8eab267/easybuild/easyblocks/generic/pythonpackage.py#L158 only writing to the debug log and continuing? I would assume that if a certain requirement is set in the EasyConfig file, and if that isn't met, EasyBuild would shout at me, and then exit, or at least, I'd think that'd be the appropriate behavior :\ |
Ok, easybuilders/easybuild-easyblocks#3430 should fix that. Let's test... |
Test report by @casparvl |
Ok, that failure was actually to prove the EasyBlock change in easybuilders/easybuild-easyblocks#3430 :) It is an expected failure, since the OS python there doesn't meet the requirement set in this EasyConfig. |
Test report by @casparvl |
Test report by @casparvl |
Test report by @casparvl |
Ugh, system python has a |
… a maximum version
Test report by @casparvl |
Test report by @casparvl |
I'm not sure what kind of funny stuff is happening on the Vega login nodes, but on the batch nodes, this seems to work fine (see this test report). I'm not going to spend my time debugging that behavior on the login node, I think we have enough successes to show that this set of EasyConfigs is sane. There is always some degree of unexpected behavior possible for EasyConfigs at the system toolchain level... |
Test report by @bedroge Expected failure when using Python 3.12 while only versions up to 3.11 are specified as supported. |
Test report by @bedroge |
…asyconfigs into reframe_462_system
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Lgtm
Going in, thanks @casparvl! |
Split off from #21269 since this requires a bit more work at the system toolchain level...
As discussed with @bedroge on chat, the problem here is that Python 3.12 would need a newer setuptools (at least 66.1.0) to function. To keep at least python 3.7 support, it could be at most 68.0.0. We can not make something that works for both Python 3.6 and Python 3.12.
The only approach here would be to make two EasyConfigs: one that would work on Python 3.6-3.11 (based on
setuptools 59.6.0
, and one that would work on Python 3.7-3.12 (based onsetuptools 68.0.0
, and probably also a bumped version forwheel
). I'll implement that here, though we should think of a relevant suffix... Maybe I'll just do one without suffix, and one with-python-3.6
. I.e. the one that uses the newer setuptools would be considered the 'default' one, and only if you really still have a system python 3.6 do you use the older one.WARNING: this PR requires easybuilders/easybuild-easyblocks#3431 in order to support the
max_py_majver
andmax_py_minver
keywords.