Closed jneira closed 2 years ago
No objection to producing a 3.2.8, I guess forked from 3.2.7. I suspect supporting extra-1.7.10 is no big deal, and aeson-2.0.1 would be annoying but not fatal.
However, is there really a requirement to run the very latest HLS against ancient compilers? Would those developers be better off getting an HLS as of today, and just not upgrading it (like they've managed to not upgrade GHC for a very long period of time). By sticking to HLint 3.2 there are a bunch of bugs that won't be fixed, so all HLS users will have a worse HLint experience as a result, which is a bit sad.
No objection to producing a 3.2.8, I guess forked from 3.2.7. I suspect supporting extra-1.7.10 is no big deal, and aeson-2.0.1 would be annoying but not fatal.
Many thanks that will be very helpful.
However, is there really a requirement to run the very latest HLS against ancient compilers?
The motivation is support users who are still using ghc-8.6.5. I guess mainly hard to upgrade private code bases, people using ghcjs (it officially only supports 8.6.5), etc.
Our deprecation policy says ghc-8.6.5 will be deprecated after LTS and HLS full suppot for ghc-9.2
.
We dont have full support for ghc-9.0.1 yeat and stack lts in on ghc-8.10.7 so 🤷
Done. Can you confirm it matches your needs? None of my CI works with it, so I'm mostly just hoping...
@ndmitchell many thanks, i am trying hlint-3.2.8 in this branch: https://github.com/jneira/haskell-language-server/tree/brittany-9.0.1
I was able to upgrade extra to 1.7.10, but no luck with aeson-2.0, my main goal 😢
Brittany-0.14.0.0 only support aeson >= 2.0
and base >=4.15.0 && <4.16
so it doesnt allow to use aeson-2.0
for ghc < 9.0.1
. So i would need ask brittany maintainers to add aeson.2.0 for older versions of base. 🤦
Closing this as it seems the HLint piece is done, even if that wasn't as helpful as hoped
yeah, I hope I still could be useful to remove compat with aeson < 2 if Brittany maintainers release a version with support for both aeson -2 and GHC < 9