Open neutrinoceros opened 11 months ago
I finally reported the issue with requires_ds
upstream: https://github.com/yt-project/yt/issues/4692
After diving in I realised that requires_ds
is in fact pretty deeply rooted into yt's answer testing framework, which is not pytest compatible. However, I also reckon that yt_astro's tests seem only written around that framework out of convenience: the "answer objects" that are produced are very small (single floats, or very small numpy arrays), so it should be relatively easy to rewrite them and get rid of nose. I'll try to have a go at it tomorrow. Meanwhile, let's see if it's possible to add Python 3.12 to the test matrix here and get rid of #232
What's not figured out yet is how to run mpi tests: I migrated them to pytest but I couldn't actually run them yet (maybe I'm hitting a deadlock).
I guess splitting this work over two pull requests (migration to pytest then migration to GHA) would help with making sure I'm not breaking anything.
start migrating tests to GitHub Actions So far I couldn't get data-requiring tests to not skip, but I think this is symptom of a bigger issue with the
requires_ds
decorator that may not work at all with pytest. I'll try to make this more useful before I open for review