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

ForceBalance 1.7.2 issues #1051

Open
mattwthompson opened this issue Jun 23, 2020 · 0 comments
Open

ForceBalance 1.7.2 issues #1051

mattwthompson opened this issue Jun 23, 2020 · 0 comments

Comments

@mattwthompson
Copy link

Here I am just documenting issues; they may (or may not) be resolved when 1.7.4 is cut so I don't think it warrants immediate action.

There is an issue with the ForceBalance 1.7.2 build on Linux with Python 3.6. Installing ForceBalance onto a Linux machine with Python 3.6 will not raise any errors until trying to import it, which will fail since the build is virtually empty. Note the size of the different tarballs:

image

Downloading and opening up the bad one shows that it lacks most of the contents it should have. This is localized to just that build; Linux on 3.7 is fine, and macOS on both versions is fine. (Also, other packages, i.e. OFFTK do not report similar errors.) It's unclear the source, but it seems to be related to NumPy in some way. The logs for 1.7.2 report not finding the right version of NumPy, but that did not cause the build to crash, just to ship something with missing files. I thought that leeping/forcebalance#181 would fix it, but the builds on 1.7.3 also fail (albeit on Python 3.7 instead) for reasons that are not clear to me: #1044

I have tried a handful of things in PRs here, but none seem to have worked properly:

  • Import package after build #1042 aims to add something like a python -c "import mypackage" at the end of a build to catch the edge case of the build succeeding, but producing an improper package
  • Update numpy defaults #1043 aims to change the default versions of the --numpy flag, although this did not alone work and it's not clear to me how much this flag matters; you can see in some logs that other (newer) versions are pulled when packages are actually built. Revert NumPy<1.18 pin on Python 3.6 #1041 is related, but somewhat independent and may be out of scope here.

Again, it's unclear to me what the source of the issue is, but it at least appears to be related to pulling in NumPy. Possibly, it may by more obvious to somebody with more experience with this build environment.

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

No branches or pull requests

1 participant