Closed quasicomputational closed 5 years ago
It looks like Stackage is already moving to Cabal 2.4: https://github.com/commercialhaskell/stackage/issues/3950
Yeah, the issue is that we can't support multiple Cabal versions at once. So, e.g., we could have the situation where one of dhall-to-cabal's dependencies can't be upgraded in Stackage because of our upper bounds, but upgrading dhall-to-cabal requires upgrading Cabal, but upgrading Cabal is hard because a whole bunch of packages need to support it. That's the logic for why this shouldn't be merged until Stackage is basically already there on Cabal 2.4.
Stackage's definitely moving on GHC 8.6 (and hence Cabal 2.4), so this ought to be safe to merge now.
This is mostly just updating the Dhall enumerations for things, fortunately.
Closes #128.
This was pretty simple: running
dhall-to-cabal --print-type $TYPE > dhall/$TYPE.dhall
for the affected enums worked like a charm. Should probably integrate that withdhall-to-cabal-meta
.Since Stackage won't include Cabal 2.4 for a little bit (not until GHC 8.6 is released, and probably not immediately), this probably shouldn't be merged yet; also because we've got to put
allow-newer
instack.yaml
since we're at the cutting edge right now.