Skip to content
New issue

Have a question about this project? # for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “#”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? # to your account

Update mirror for nightlies #718

Merged
merged 6 commits into from
Oct 31, 2023
Merged

Update mirror for nightlies #718

merged 6 commits into from
Oct 31, 2023

Conversation

jtilly
Copy link
Member

@jtilly jtilly commented Oct 25, 2023

scientific-python-nightly-wheels is a lot more up to date than scipy-wheels-nightly.

CI failure is because of #714.

@MarcAntoineSchmidtQC
Copy link
Member

With the last commit we are now able to build the environment. The tests are failing because the new channel points to numpy 2.0. From what I can see, there's a simple fix (switch from np.Inf to np.inf), and there's one that I'm not sure yet (one about NEP50)


PRE_WHEELS="https://pypi.anaconda.org/scipy-wheels-nightly/simple"
PRE_WHEELS="https://pypi.anaconda.org/scientific-python-nightly-wheels/simple/"

Choose a reason for hiding this comment

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

Adding a review comment so that we do not merge this right now. We need to fix the errors caused by the newer versions.

@MarcAntoineSchmidtQC
Copy link
Member

Reading a bit on numpy 2.0. I'm worried that the switch to scientific-python-nightly-wheels will be problematic. The fact that it's more up to date also has a downside: the probability that those packages are not yet working together is much higher.

Currently, Pandas does not work with numpy 2.0: pandas-dev/pandas#55519.

@jtilly
Copy link
Member Author

jtilly commented Oct 27, 2023

Yep, got it. But that just means we shouldn't test against NumPy 2.0. I think it doesn't justify using an outdated mirror that won't get updated any longer 😉

@MarcAntoineSchmidtQC
Copy link
Member

I didn't know that the scipy-wheels-nightly was deprecated. I just thought that it was using older versions. So you are totally right, we should absolutely be switching.

I checked and pandas fixed the problem the same day we discussed it. I think our failures were mostly due to bad timing, so I'm willing to make the switch.

I made a PR to fix this. The nightly build is passing with numpy 2.0! #720

# for free to join this conversation on GitHub. Already have an account? # to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants