CI issue

[sunt05]

In terms of fixing the CI, I can't even remember what the issue was, but if you find out what that was feel free to ask me for help if you need it. My email right now is [email protected].

Originally posted by @mcgibbon in #1 (comment)

Just wondering as GH Actions has been out for a while, shall we consider using that instead of fixing the travis-based CI?

[mcgibbon]

Github Actions (from my quick reading) still needs somewhere to execute the tests. The issue wasn't really anything to do with travis per se, IIRC it was because I was working on a transition towards using udunits for unit handling (optionally), and udunits or any non-standard binary package is really hard to install on CI. A suitable fix would be to roll back the udunits changes entirely and go back to only supporting pint-based unit conversions.

[mcgibbon]

Ideally the CI (whether travis or otherwise) should be able to install with only pip install steps, and no need for binary packages. I'm willing to lose the related functionality (like udunits support, which was never working anyways) to do this, as the core functionality which people seem to use (thermodynamic calculations) doesn't depend on it.

[sunt05]

+1 for rolling back the pint-based support for unit.

I brought up the GH Actions just for better consistency: so all code and testing reside in one place; while I fully understand the issue per se has nothing to do with travis.

[mcgibbon]

If GH Actions is free and provides a free server to run the tests on, I'm not opposed to it. My understanding is that you use it to trigger a job on a server somewhere else, which Github doesn't run, but I could be wrong.