feat: get ci unstuck on broken flake

[JonathanLorimer]

No description provided.

[JonathanLorimer]

@mitchellwrosen I see that you are moving off nix for CI, but I figured in the interim I would fix the broken flake

[mitchellwrosen]

Thanks @JonathanLorimer, indeed there is some feeble attempt to use haskell-ci to generate a github actions file, but that PR (#19) still isn't passing tests due to a cryptic empty error coming out of tmp-postgres.

The current suspicion is that the tmp-postgres package code is poorly written and isn't fully capturing initdb's stdout and stderr streams in the event that initdb fails with non-zero exit code, so it's difficult (and unsatisfying) to debug.

We could consider moving away from the tmp-postgres dependency in favor of some more robust code. In the meantime, I see #22 as well, which is an improvement, so we're going to merge that, but don't hesitate to keep working on this and/or asking questions!

[JonathanLorimer]

closed in favour of #22