meinardmueller/synctoolbox

Enable continous integration

Closed this issue · 7 comments

to solve installation issues it could help to include a number of continuous integration tests that run

  • runs the unit tests and installs the package
  • test the package installation in a conda environment using different python versions

openjournals/joss-reviews#3434

I created a workflow which runs the unit tests in a conda environment using python 3.8 version:
https://github.com/meinardmueller/synctoolbox/actions/runs/1070732431

Would it be necessary to have two more workflows for python 3.7 and 3.9 versions?

great!

Would it be necessary to have two more workflows for python 3.7 and 3.9 versions?

i usually test multiple python version for the pip tests and only a single one conda test because the conda workflows aren't that quick and do not cache that well..

See this example how run matrix builds with different python version using a single workflow

Thank you, the example has been very helpful.

I created two pull requests following your instructions:
https://github.com/meinardmueller/synctoolbox/pulls

I will merge them if you confirm it.

@yiitozer both look good to me.

Just noted: there are a bunch of deprecations shown in the unit tests when running with numpy 1.20 (See here). I would expect this package would break with a future numpy version soon, so better to fix this sooner rather than later

@faroit, thank you for pointing it out. I fixed it!

Can we close this issue?

yep. thanks!