Closed mfherbst closed 3 years ago
I think if we want to maintain full compatibility with Psi4, we should deploy all versions...
Sorry, I did not read this well enough 😄
dev
deployments should be limited to Linux and a single Python version IMHO.
Deployments to main
should be all 4 versions on macOS and Linux.
All right to that I can 100% agree. Because that's not what happens right now :smile:.
I don't know... deployments to dev
are using py38 only right now?! So if the label is set correctly, only py38 should be deployed to dev
for Linux & macOS (with the current configuration).
Ah yes you might be right. Might have been my mistake. Ok let's see. I let the CI run through on master. Sorry I might have confused myself over the many changes with the last PR :confused:.
It's not a problem, hope everything works as expected now. The best thing is that we succeeded to deploy ALL versions to conda main now, that's pretty amazing! Less than 1 hour for all 4 py versions on each platform 💪
Plus pip ... if you compare to the pain we had before ...
Everything seems to work as expected (only 3.8 deployed).
Is it really necessary to deploy all four python versions and all two OSes to Conda for each push to master? I would suggest it is more economical to stick only to the two latest python versions and maybe even drop macOS? What do you think @maxscheurer?