Closed phadej closed 2 months ago
Is it intentional that haskell-ci silently stops generating CI matrix entries for GHC 7.* without any warning/error whatsoever, despite those being mentioned in tested-with
?
Is it intentional that haskell-ci silently stops generating CI matrix entries for GHC 7.* without any warning/error whatsoever, despite those being mentioned in tested-with?
Unfortunately it's how haskell-ci
works. We specify a version range for GHC versions; if today you write tested-with: GHC >=10
you won't get none, with GHC >=7.0
you'll get all (incl all minor versions, i.e. a lot).
OTOH, the
*INFO* Generating GitHub config for testing for GHC versions: 8.0.2 8.2.2 8.4.4 8.6.5 8.8.4 8.10.7 9.0.2 9.2.8 9.4.7 9.6.3 9.8.1
output is almos the only thing haskell-ci
outputs, so should be easy to spot if something as off; and obviously git diff
before you commit the changes.
So I don't see that "silence" as a major problem.
Oh, I didn't realise you can use anything else than ==
in tested-with
(it doesn't seem to be documented anywhere). Makes sense if you indeed can.
And yeah, as you may have noticed from the GitHub messages about referenced commits/PRs, I have implemented exactly the check you suggest: https://github.com/liskin/xmonad-maintenance/commit/50b9290d8e1a12c9de03b3100e65ef9956c78590
Seemed like a bit of a hack but good to know it's not 🙂
haskell-ci
is not meant nor designed to be run in non-supervised environments.
It creates a pull request, that's supervised enough for us 😏 Still, we want to automate as much as possible. Relying on humans tends to disappoint. Hence the safety check.
Resolves #719