Skip to content

ASV py3.6 environment fails to install (numba/numpy conflict) #1439

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
roger-lcc opened this issue Apr 9, 2022 · 5 comments
Closed

ASV py3.6 environment fails to install (numba/numpy conflict) #1439

roger-lcc opened this issue Apr 9, 2022 · 5 comments
Milestone

Comments

@roger-lcc
Copy link
Contributor

Describe the bug
In asv.conf, the virsion of numba is conflictde with the virsion of numpy,so when run this code asv check,there is a bug.

To Reproduce
Steps to reproduce the behavior:

  1. Validate and build environments (may take a couple minutes to run): asv check
  2. See error

Expected behavior
modify the asv.conf

Screenshots
image

Versions:

  • pvlib.__version__:
  • pandas.__version__:
  • python:conda python 3.9

Additional context
And i change the virsion 0.36.1 of numba to the virsion 0.40.0 of numba ,it works.

@Naman-Priyadarshi
Copy link
Contributor

I tried to recreate the same behavior to get the error but my asv check ran successfully. When I published my result the graphs were empty. So, I think we should remove the python 3.6 environment as it has reached EOL and add another environment for python 3.9 or 3.10 and update all the versions to the latest for numpy, pandas, scipy, h5py, ephem, and numba.

@roger-lcc
Copy link
Contributor Author

I think the result of each benchmark is just one commit result because we just ran asv run.If we click into the graphs, we will see the results of benchmarks.
image
And I think if we run asv run v0.9.0——the virsion of pvlib, and we could get the results of all the commits. But the process of the benchmarks is too long, so I have not fininshed it.

@roger-lcc
Copy link
Contributor Author

And I tried again the default asv.conf, it still had the problem.And the test of benchmarks is upon the conda virtual environment which asv created itself.So I think maybe the problem is still about asv.conf.

@kandersolar kandersolar changed the title pvlib-python-asv.conf ASV py3.6 environment fails to install (numba/numpy conflict) Apr 11, 2022
@kandersolar
Copy link
Member

the result of each benchmark is just one commit result

Yep, each dot in the line chart corresponds to the results of a single commit on the pvlib master branch. Just running asv check won't actually store any benchmark results like asv run ... does.

the process of the benchmarks is too long

Yeah, the benchmarks are a bit slow :( it's a few minutes just to benchmark a single commit, so running benchmarks for a long history of commits does take quite a bit of time. Maybe we can improve that.

Anyway, this issue seems similar to pvlib/pvlib-benchmarks#1. The linux server we use to run the nightly benchmarks isn't affected by this problem, so maybe the incompatibility is related to the OS. In any case I agree that we should update the configuration file so that the instructions work on a typical PC. While I agree it makes sense to stop benchmarking on python 3.6 eventually, I don't mind keeping it around as long as we're still running the unit tests on 3.6. So maybe for now we should just increase the numba version and leave the python version update for later.

@roger-lcc
Copy link
Contributor Author

Thanks! @kanderso-nrel @Naman-Priyadarshi

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

No branches or pull requests

3 participants