Allow template-haskell-2.16

[sjakobi]

This uses -XDeriveLift to address the missing liftTyped definitions.

[sjakobi]

At this point, I can already successfully cabal test dhall:tasty dhall-json dhall-yaml with GHC 8.10 if I use this patch to work around bounds issues:

--- a/cabal.project
+++ b/cabal.project
@@ -1 +1,3 @@
-packages: ./dhall ./dhall-bash ./dhall-json ./dhall-yaml ./dhall-lsp-server ./dhall-nix
+packages: ./dhall ./dhall-bash ./dhall-json ./dhall-yaml
+
+allow-newer: serialise:ghc-prim, cborg-json:base, cborg:ghc-prim

@Gabriel439, maybe this would also be worth including in the release! (Not sure if it's still a minor release BTW)

[Gabriella439]

@sjakobi: Is it a breaking change to the API?

[sjakobi]

Is it a breaking change to the API?

I think the type changes in Dhall.Repl in #1717 probably were. Although I don't believe that a lot of library users will notice…

[Gabriella439]

@sjakobi: Oh yeah. Either way, that's fine. There's no harm in cutting a 1.32.0 release. I don't obsess too much over version numbers if it's necessary to bump them to push out fixes. In my view, the only mistake is not releasing frequently enough (i.e. piling up too many breaking changes in infrequent big bang releases)

[Gabriella439]

Also, looking at this more closely, I don't think the Dhall.Repl changes are breaking anyway because we don't expose those in the public API. See: https://hackage.haskell.org/package/dhall-1.31.0/docs/Dhall-Repl.html

[sjakobi]

Also, looking at this more closely, I don't think the Dhall.Repl changes are breaking anyway because we don't expose those in the public API

All the better! :)