Skip to content
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

try to use pip instead of conda-build for develop install #2463

Closed
wants to merge 1 commit into from

Conversation

munkm
Copy link
Member

@munkm munkm commented Feb 26, 2020

PR Summary

A number of recent PRs are failing on appveyor because conda-build can't be installed in the conda environment and needs to be installed in the root environment. It looks like we use conda-build only to be able to do a development build of yt with conda develop -b. Maybe the appveyor builds can be fixed by using pip instead.

@munkm
Copy link
Member Author

munkm commented Feb 26, 2020

The Python 2.7 builds fail on appveyor, but this is due to the updated numpy support that we added from #2448.

#2450 removes the python 2 testing from appveyor, so I think these failures shouldn't block merging this PR.

@neutrinoceros
Copy link
Member

  • I tried to look up pip's documentation for a replacement of the -b flag (build extensions in place) in conda develop -b ., but couldn't find one.
  • Meanwhile, the build still fails on python 3 with a surprising (to say the least) error
    AttributeError: module 'numpy' has no attribute 'divmod'

Do you think there is a chance those two points are related ?

@neutrinoceros
Copy link
Member

update: seems like pip doesn't have an equivalent flag by design and relies on the environment manager to invoke python setup.py build_ext --inplace (see pypa/pip#7509)

@munkm
Copy link
Member Author

munkm commented Feb 26, 2020

@neutrinoceros the divmod errors are because the numpy version that's being pulled by appveyor is too old (it's 1.11) to be compatible with our changes that added 1.18 support. I'm not sure why appveyor is pulling such old versions of numpy for both Python versions right now; when the 1.18 support PR was merged it was testing against numpy 1.16.

@munkm
Copy link
Member Author

munkm commented Mar 10, 2020

Ok, it seems that the changes introduced by conda have been remedied and this isn't an issue anymore (thanks @rebeccabilbro for figuring it out!). I retriggered the appveyor tests and we can install yt from source with pip install rather than conda develop. The failing tests for Python 3.6 are for matplotlib 3.2 (which I'm working on in another PR).

If anybody feels strongly about using pip install instead of conda develop in our appveyor scripts I'm happy to reopen this PR. However, I don't think this is a change that needs to be merged.

@munkm munkm closed this Mar 10, 2020
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.

None yet

2 participants