Skip to content

Install matplotlib before test dependencies on Azure #19353

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

Closed
wants to merge 2 commits into from

Conversation

dstansby
Copy link
Member

Possibly fixes #19352. I think it might be better to try installing matplotlib first anyway, so if that fails there hasn't been a load of wasted time installing the test dependencies. Maybe I've missed something though.

@dstansby
Copy link
Member Author

Hmm, doesn't look like this has worked.

@dstansby dstansby closed this Jan 24, 2021
@dstansby dstansby deleted the install-order branch January 24, 2021 14:44
@dstansby dstansby restored the install-order branch January 24, 2021 16:50
@dstansby dstansby reopened this Jan 24, 2021
@dstansby
Copy link
Member Author

This seems to work now - the build is failing after 4 minutes instead of 1hr.

@jklymak jklymak requested a review from QuLogic January 24, 2021 22:31
@anntzer
Copy link
Contributor

anntzer commented Jan 25, 2021

I am confused by why CI is still marked as failed, though. There's the || [[ "$PYTHON_VERSION" = 'Pre' ]] which should make steps always succeed when using the prerelease version...

@dstansby
Copy link
Member Author

I am confused by why CI is still marked as failed, though. There's the || [[ "$PYTHON_VERSION" = 'Pre' ]] which should make steps always succeed when using the prerelease version...

I deliberately removed that at the install stages, so if installing fails the build fails.

I'm not too sure what the logic is for spending time on a build that always passes by construction, and therefore doesn't provide helpful feedback on whether anything is broken or not? If the motivation is not to block PRs on a failing python dev, wouldn't it be best to run it only on the master branch (not PRs) and let it fail properly?

@anntzer
Copy link
Contributor

anntzer commented Jan 25, 2021

Ah, I somehow completely missed the fact that you had removed it.
I have no strong opinion as to what to do with the Pre build. On the one hand it seems annoying to have CI marked as failing due to failures in unreleased versions; on the other hand I agree always passing it is not great either. Azure needs some kind of "canfail" jobs, like travis had...

@dstansby
Copy link
Member Author

Well, I think this is an improvement as it changes the python pre CI to failing after 60 minutes with a timeout to failing after 4 minutes. I'd say we should merge this, and discuss what to do long term with the python pre build on a new issue.

python -m pip install -ve . ||
[[ "$PYTHON_VERSION" = 'Pre' ]]
displayName: "Install self"
python -m pip install --upgrade pip
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The pip upgrade still needs to happen before us, I think.

@tacaswell
Copy link
Member

I'm inclined to merge #19396 ahead of this. It looks like we are currently failing on building numpy.

The cause of the current failure is

      _common.obj : error LNK2001: unresolved external symbol _PyGen_Send

which is due to a change in the internal cpython c-api's for handling generators interacting with code produced by cython. There is a fix to this in cython (cython/cython#3921) but not released yet. This seems a bit too far from our core-concerns to be worth testing?

@dstansby
Copy link
Member Author

dstansby commented Feb 4, 2021

Merging #19396 sounds 👍 to me

@dstansby dstansby closed this Feb 4, 2021
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

Windows python PRE build failing
4 participants