python@3.10 3.10.7

[branchvincent]

Created with brew bump-formula-pr.

resource blocks may require updates.

[carlocab]

This should be done in the next hour or so.

[iMichka]

Fix for speedtest-cli: #110021

[iMichka]

Looks like poetry did not like that update:

2022-09-07T01:36:26.2528322Z ##[group]Full test poetry output
2022-09-07T01:36:26.2529036Z �[32m==>�[0m �[1mTesting poetry�[0m
2022-09-07T01:36:26.2531618Z �[34m==>�[0m �[1m/home/linuxbrew/.linuxbrew/Cellar/poetry/1.2.0/bin/poetry --version�[0m
2022-09-07T01:36:26.2532183Z �[34m==>�[0m �[1m/home/linuxbrew/.linuxbrew/Cellar/poetry/1.2.0/bin/poetry new homebrew�[0m
2022-09-07T01:36:26.2532892Z �[34m==>�[0m �[1m/home/linuxbrew/.linuxbrew/Cellar/poetry/1.2.0/bin/poetry config virtualenvs.in-project true�[0m
2022-09-07T01:36:26.2533599Z �[34m==>�[0m �[1m/home/linuxbrew/.linuxbrew/Cellar/poetry/1.2.0/bin/poetry add requests�[0m
2022-09-07T01:36:26.2534260Z Creating virtualenv homebrew in /tmp/poetry-test-20220907-1029002-1fmn9m/homebrew/.venv
2022-09-07T01:36:26.2534481Z 
2022-09-07T01:36:26.2534581Z Failed to unlock the item!

[github-actions]

This pull request has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. To keep this pull request open, add a help wanted or in progress label.

[xxyzz]

Looks like it's keyring's error, it can be disabled by setting this PYTHON_KEYRING_BACKEND=keyring.backends.null.Keyring.

[carlocab]

poetry might have been fixed by #110433.

It would be good to rebase and re-run this. Not sure there's enough time for that before tomorrow's maintenance window though.

[xxyzz]

The poetry error is not fixed, maybe keyring is still need to be disabled. I wonder why the poetry tests only fail in this pull request.

[carlocab]

If we disable keyring in the test, does that mean users need to disable it on their own to use poetry?

If so, then that's not a good solution, because we are then shipping something effectively broken.

[xxyzz]

The test failed because Keychain can't be unlocked in CI, normal user probably won't encounter this error. But these test commands shouldn't need to find password in Keychain.

[carlocab]

The test failed because Keychain can't be unlocked in CI

This was always the case, though. Why didn't it fail previously, and why is it failing now with this version of Python?

[xxyzz]

Maybe because it's a new feature: python-poetry/poetry#1917 (comment)

I don't know why poetry wants to use keyring in this case but other people in that issue solved this error by disabling keyring.

[carlocab]

Maybe because it's a new feature: python-poetry/poetry#1917 (comment)

I don't know why poetry wants to use keyring in this case but other people in that issue solved this error by disabling keyring.

But the feature seems to be there already, but the failure appears to be triggered only by Python 3.10.7 and not Python 3.10.6. So I don't think the feature being new is enough of an explanation.

Hey @neersighted, apologies for the tag but we could use your expertise here. Does it make sense to disable keyring when testing poetry in CI here?

This makes sense to do if the keyring failure is peculiar to our CI environment. However, if it's an error a user will see when using it on their machine, then we'll want to do something else.

[neersighted]

Hi @carlocab -- keyring usage in Poetry is new, and the matrix of possible OS versions/keyring backends/local system configration is vast and well beyond the ability of any project to test. It's been pretty clear since this feature has made it out into the wild that Poetry needs more robust error handling/usage of keyring.

There should be improvements to keyring usage coming to future releases (including patch releases) of Poetry, but in the here and now, keyring should not be something you need in CI and can be safely disabled. It is a very desirable feature for end users (e.g. on macOS, storing credentials in the keychain instead of plaintext on disk), as PyPI credentials are incredibly valuable as supply chain attacks can be conducted using them.

I would suggest keeping an eye on the Poetry changelog for improvements to our handling of these errors, and in the mean time just setting the env var (or using the config file as described in python-poetry/poetry#1917) to disable keyring in CI.

[iMichka]

Let's disable the test then: #111744

[iMichka]

Fix for audio: #111745

[carlocab]

I've seen the macOS failures elsewhere so they're pre-existing. Having python3 built against gcc-5 is also causing failures elsewhere on Linux, so let's merge this.

We probably want to set CC=cc on Linux or something to avoid hardcoding the GCC version.

[BrewTestBot]

@carlocab has triggered a merge.