Closed stijnvanhoey closed 2 years ago
I also added some github actions as the current drone setup does not support a matrix check as github actions actually does (testing multiple versions of python by a single specification; in drone we would need a separate specification for each version).
Still to add some additional unit tests and fix the 3.9 version issue
For the moment, I did drop the py3.9 in github actions. I can't reproduce the error locally an all test do pass in 3.9 on my local machine, need to check this further later on.
In order to have the github actions work, some more credentials need to be setup, e.g. the pypi and ghpages setup (if we move this from drone to github actions). I do think this switch make sense (e.g. keep things the same with pywaterinfo) and with the config in tox, things are easy to switch.
For the moment, I did drop the py3.9 in github actions. I can't reproduce the error locally an all test do pass in 3.9 on my local machine, need to check this further later on
The issue appeared to be the git lfs not loaded into github actions (which also explains the difference with the local setup). I included 3.8/3.9 again.
@Sachagobeyn with this PR, I do think you can setup the release 0.1.0 (cfr. https://github.com/cn-ws/rfactor/issues/34)! I added the release instructions to the docs, see https://github.com/cn-ws/rfactor/blob/update-scaffold/docs/contributing.rst#creating-a-new-release. I propose we do this the first time together, so I can introduce you these steps.
I would keep both the drone and github actions for the moment and just see how gh actions behaves The setup is similar to pywaterinfo:
Unit test coverage and docstring coverage are reported as part of the CI: Unit tests and docstrings
For the moment, I did drop the py3.9 in github actions. I can't reproduce the error locally an all test do pass in 3.9 on my local machine, need to check this further later on
The issue appeared to be the git lfs not loaded into github actions (which also explains the difference with the local setup). I included 3.8/3.9 again.
FYI, you can only use a limited number of LFS calls in github for free. We reached that limit, which may have caused tests to work first and break afterwards.
@johanvdw : let's ignore the git lfs for now ;)
I updated the cofiguration according to the latest pyscaffold template, did some review on the internal links of the sphinx docs, updated the pre-commit hooks and made sure it is run on all files, tackled #33 and made sure the analysis_flanders notebook is actually rendered in the sphinx documentation (although our pre-commit hook nbstripout would clean the notebook.
To achieve the latter, I added
"keep_output": true,
to the notebook metadata, by which it is conserved by nbstripout (cell counts are deleted, making the diff still easier to read). @johanvdw, maybe interesting for you as well and might be useful for other cnws package notebooks.