stacklock2nix
stacklock2nix copied to clipboard
Use all-cabal-hashes directly rather than extracting individual components
This gives us a significant speed-up, since all we need to do as part of the IFD is run cabal2nix.
This is slightly simpler than the some-cabal-hashes approach, but still speeds things up considerably for me. We are still doing a bunch of IFD to generate the cabal2nix files, but it goes pretty quickly. I still feel like something like the some-cabal-hashes thing is the right thing to do, but it's a bit more work than I have capacity for right now.
The downside is that now we require all-cabal-hashes to be an unpacked directory. I could add something to enforce this, eg, a derivation that unpacks it if it's a tarball (is there a utility to do this)?
What do you think?
Hey, sorry I haven't gotten to this yet. I think this looks interesting, but I haven't really be able to do an in-depth review.
I might not be able to really get to this for another month or so. Sorry to keep you waiting longer!
Don't worry about it at all! There's no rush