GHC 9.0 & 9.2 support in new Hackage release
Currently, base <4.15.
https://github.com/protolude/protolude/pull/131 Supports 9.0.1
Is there a reason it's not released on Hackage yet?
I don't mean to nag (actually I do), but could you add 9.2.2 support as well? As far as I can tell it's #131 that causes breakage with base 4.16
9.2.2 support required for postgrest.
We're (Artificial) also excited to get a Hackage release of 0.3.1 for the GHC 9.2 support.
I'll be watching this issue for updates.
It's done! https://hackage.haskell.org/package/protolude-0.3.1. This issue can be closed.
Hurrah! 👏 Thanks
The exports executable in the cabal still requires base <4.15. Can that be bumped, too or are there reasons not to?
Also 0.3.1 on Hackage still seems to require base<4.15, which excludes GHC 9.0.x (base 4.15.x) as well as GHC 9.2.x (base 4.16.x). Maybe there’s an upstream dependency which either hasn’t added support for base>=4.15, or has and needs its version range bumped?
The executable can be disabled by setting the dev flag to false (this is the default, but I think stack doesn't handle this defaulting correctly). But for some reason I could not get this to work when building stackage nightly.
I tried bumping the base dependency for the executable but it doesn't compile.
I'm not sure what the purpose of this executable is. Some options:
- Fix it
- Remove it
- Keep it, but remove from the .cabal file (or add it to extra-source-files)
The purpose of that executable is to create an export list over all of protolude. However, it's using the ghc api and no longer stable, so my intension is to remove it, and if we ever need it we can resurrect it back from the commit history.
Thus, I'll be adding a new PR with the following changes, and cut a new release for it:
- Add support for 9.2.2
- Drop the executable.
https://hackage.haskell.org/package/protolude-0.3.2
9.2.6 support is blocked on an ostensibly unnecessary < 0.11.4 constraint on bytestring.
9.2.6 support is blocked on an ostensibly unnecessary
< 0.11.4constraint on bytestring.
It builds fine here with ghc-9.2.6 (against bytestring-0.11.3.1), could you elaborate as to how it's blocked? (I suppose there might be trouble when also depending on some other ghc-bundled packages that require bytestring 0.11.4?)
Indeed, sidestepping the GHC provided version of core packages is often troublesome, so I'd expect e.g. Stackage LTS-20 to jump to that version, for example. nixpkgs will do so soon in any case which is how it is blocked for us – for better or worse, we use the core packages where available.
The tight upper bound on bytestring does look accidental -- I've update https://github.com/protolude/protolude/pull/141 to relax that.