Closed Mikejmnez closed 1 year ago
You might need my approval to issue a new release on pypi. I'm pretty sure the releasing procedure we are using is quite old. If you can wait until Monday (it's a bank holiday in Italy) I'll give you the permissions to release and I will also update the procedure (in the future, you will only have to make a release on github and both pypi and conda forge will also issue a release with the right version).
Great! It is quite old. bump2version
hasn't been updated in like two years. I'll wait until next week then.
Looks like issues with xesmf showed up when #288 was merged into main. Not sure why, it was ok a few hours ago: https://github.com/hainegroup/oceanspy/actions/runs/3675826523
Do you guys have time to take a look?
Sure, I'll take a look. I also noticed that the documentation hasn't updated to show any changes you did in your PR.
Good catch! RTD should be OK now, it was still pointing to "master"
@malmans2 I still don't see the new updated documentation. There was a build based on your PR #288 but it appears to have created a new version based on that PR. See the two cases below (screenshot from my laptop) :
This is the screenshot when I go directly to the oceanspy documentation. Here, the Contributing
(and other sections like People
) remain as before, unchanged.
In this version of the documentation, To see this particular version, I have to click on the link you supply on your comment in PR #288 . The link is: https://oceanspy--288.org.readthedocs.build/en/288/contributing.html#deploying
I synced my fork and noticed there are two untracked files resulting from that PR #288. Here's the snapshot from my command line:
These were not there before in my local repository. I will stage them, create a new PR and maybe this will help update the documentation. Particularly resolve #253
This looks good to me (I only edited the Deploying section at the bottom): https://oceanspy.readthedocs.io/en/latest/contributing.html
Maybe you have to refresh or clear your browser cache?
Ok, that could be it...
Great, it did! I did have to refresh a couple of times. And actually, the binder is the correct one so #250 can now be closed! I am a little annoyed though...
@malmans2 what do you suggest to do with all the untracked documentation files (docs/datasets.rst
and docs/generated/
)?
I think those dir and file should go back in the .gitignore as they are generated each time we compile the documentation. I updated the .gitignore and I inadvertently removed them.
mmm, https://anaconda.org/conda-forge/esmpy was released on conda-forge a few hours ago. Must be related to the issues we are seeing (BTW, if pytest fails the workflow will not proceed with the release on PyPI)
OK. So we need to fix this pytest fail before the release. I'll work on this test fail today then. After fixing it, will issue a new release
Fixed failing tests by pining esmf, see PR #292 and issue #291. I am releasing a new version of oceanspy and thus will close this issue.
The release on PyPI did not go through yesterday because you forgot to delete the tag when you deleted the release. There were also a couple of typos in the workflow.
All fixed now and v0.3.0 is on PyPI: https://pypi.org/project/oceanspy/#history
conda-forge
should also open a PR soon
From now on releasing is just a matter of creating a new release on GH.
Great! I didn't realize tags needed to be deleted but is all learned now! Thanks a lot @malmans2 !
There's just one thing left: the conda-forge release. Can you handle it please so you'll be able to do it in the future? Here is the PR: https://github.com/conda-forge/oceanspy-feedstock/pull/9
If you need support to merge the PR, Filipe Fernandes is the conda-forge guru (github tag is @ocefpaf
)
Sure!
I will be working on a new release, long overdue, in the next couple of days. This will make it easier to use the new oceanspy features (specially PRs #281 #284 and #285) say on Sciserver, without needing a pip install from the GH repo.