-
Notifications
You must be signed in to change notification settings - Fork 12
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
tmpfs support #58
Comments
Thank you for opening your first issue in this project! Engagement like this is essential for open source projects! 🤗 |
What filesystem are you using? That information is very relevant for this project, but I neglected to add it to the issue template. 😅 |
|
I'm running tests on tmpfs. |
Yup, LocalFileIdManager so far has only be tested on the "mainstream" filesystems, namely: ext4, NTFS, HFS, and APFS. Depending on whether tmpfs support for LocalFileIdManager is possible, the solution here is to either skip the tests on tmpfs platforms or to implement tmpfs support. |
You will still be able to use the |
So I should basically be skipping these tests in the Gentoo ebuild for jupyter-server-fileid? I don't really want to guess whether user's filesystem is supported or not. |
For now, yes.
Unfortunately, we have not yet run a comprehensive test of LocalFileIdManager on every plausible filesystem. I hope you appreciate that this task would involve a considerable amount of effort. 🙏 We had been planning on doing so, but this has been assigned a lower priority after we made the decision to default to ArbitraryFileIdManager to ensure compatibility on all platforms. As a temporary workaround, perhaps I could somehow "separate" the Local and Arbitrary FIM tests for you? That way, on build platforms, you can just run the ArbitraryFIM tests exclusively instead, which all should pass. |
To be honest, I don't think that would ever be possible. If you're relying on anything non-portable, then sooner or later it'll break in unexpected ways (kernel changes, people using union filesystems…).
I'm sorry but I don't really know the package, we've ended up packaging it because it's needed by jupyter-server-ydoc. Do I understand correctly that the test suite right now is effectively run against "Local" FIM? If yes, then indeed I think it'd be beneficial to run against both FIMs. Perhaps using a parametrized fixture — that should be sufficient to make it easy for us to |
Yeah, I know. 😔 Which is why LocalFileIdManager will likely remain opt-in only.
There are tests for both FIMs in the test suite.
Is there no way to tell pytest to skip tests by name? The test names should be fairly stable, so your build process could just skip those 2 failing tests for now until we can allocate effort to work on this. |
I'm sorry, I've missed your reply. Do I understand correctly that basically What I was referring to is making sure that we skip these tests for "local FIM" without skipping the "arbitrary FIM" tests. |
@mgorny Yes. 👍 |
Description
When running the test suite, I'm getting inconsistent results, with 1-4 of the following tests failing:
Full output from one run (where two of them failed)
Full output from one run (where two of them failed)
Reproduce
python -m venv .venv
. venv/bin/activate
pip install .[test]
pytest
Expected behavior
Tests passing on every run ;-).
Context
Troubleshoot Output
The text was updated successfully, but these errors were encountered: