For library testing, I understand the benefit of CI. I saw a post where @jeremy proposed hosting a small amount data on AWS for testing code.
However, some notebooks churn through a lot of data and it may not be desirable (or affordable) to have GitHub Actions download and run through the data for every commit. However, it’s still beneficial to nbdevify these repos, for all the other reasons that make nbdev
so great—namely nbdev_test
, git
diffs, and to take advantage of Pages documentation deployment.
Presumably this is handled by having library repos do CI and non-library notebook collections to not do CI. Is this correct? If so, what is the preferred way to inactivate CI? Just delete the test.yaml
file?