csdms / pymt

A Python toolkit for running and coupling Earth surface models
https://pymt.readthedocs.io
MIT License
52 stars 20 forks source link

Use relative_files configuration for coveralls #146

Closed mdpiper closed 3 years ago

mdpiper commented 3 years ago

This PR attempts to fix the successful, but 0%, coverage score reported by Coveralls through the coveralls-python Action.

mcflugen commented 3 years ago

@mdpiper I've forgotten how exactly we test our coverage. It looks like it is part of our ubuntu and Python 3.9 Test action but yet it also shows up as two separate tests (one of which is failing).

mdpiper commented 3 years ago

@mcflugen The failure is only a nominal sign error (-0.0%). I've seen this before with Coveralls, plus the only change from the successful commit prior to this was an update to the changelog.

mcflugen commented 3 years ago

@mdpiper I guess my real question is, why do those two coveralls checks show up as separate entries at all? Should they not be part of the Test / build-and-test (ubuntu-latest, 3.9) check? Are we running extra coveralls checks that we don't need to?

mdpiper commented 3 years ago

@mcflugen I'm not going to guess at what Coveralls or the coveralls-python Action are doing, but AFAIK we're following the directions.

mdpiper commented 3 years ago

@mcflugen On Coveralls, there's one job listed under a build (these are the two links we see in the test results). My hunch is that since coveralls-python is built for parallel testing, the build is a container for holding multiple testing jobs in a commit, while the one job is our test, run in serial. This also may explain the label "false", since

with:
  parallel: false

is assumed by coveralls-python since we didn't set it in the action. It seems likely that Coveralls is only called once.

mcflugen commented 3 years ago

@mdpiper sounds good. 👍