PR #24 added a somewhat complex step checking for problematic filenames.
I wonder: can we put this into a separate check_filenames.sh (or whatever name), and just call that from the YAML script? This has IMO several advantages:
- this script can potentially be reused by others (possibly even
ReleaseTools, although I have no intentions to doing that, but someone else might wish to take over development of it)
- we can run
shellcheck on it as part of our CI
- we can write dedicated tests for the script, i.e., to test whether it correctly identifies various naming issues, without having to make this part of a test of the entire
release-pkg workflow, which is IMHO far more work to pull off
The same would apply to some of the other more complex scripts (e.g. the date check or the Read data from package-info.json step). And of course also in some of our other actions.