Is there a discussion of motivation for this action? The ecosystem uses the haskell/setup action pretty much universally. Should we switch from it to ghcup/setup for some alleged benefit?
This action is:
- simple
- fast
- predictable
It follows the unix principles and does only one thing: install ghcup.
It doesn’t try to map all sorts of things into yaml options that already exist in the properly mantained ghcup cli tool.
It doesn’t try to be smart and detect existing installations or invoke the package manager. You get ghcup and all other tools from the ghcup distribution. There are no surprises.
This action is also maintained by GHCup developers, so the installation logic should never break.
I guess haskell-actions/setup could theoretically use it.
I personally will be using only haskell/ghcup-setup in my own CIs.