| 5 Jun 2021 |
 sterni (he/him) | In reply to @maralorn:maralorn.de
If so I imagine a broken = forDefaultCompiler would work better? yeah well naming is the other thing always :) but that would be the idea indeed | 12:40:39 |
 maralorn | In reply to @Las:matrix.org
Couldn't just haskell language server stay on GHC 8.10? I mean we could still deliver an 8.10 hls to the users. But they wouldn‘t be able to use it on their haskellPackages packages. | 12:41:39 |
 Las | makes sense | 12:41:48 |
| maralorn | Maybe that’s something we need to swallow. But it would completely break my use case for nixpkgs and I would likely quickly loose interest in maintaining it.^^ | 12:42:40 |
| Las | Is there some overview of why it's broken on GHC 9? I couldn't find anything on their GitHub | 12:42:49 |
 joe (he/him) | isn't it one of the pinned issues? | 12:43:24 |
| Las | you are right | 12:43:39 |
| maralorn | Las: They have an issue about it. The problem is that GHC changed a lot of their internal APIs including nearly all of their Module names and hls links directly against their internal API. | 12:44:11 |
| Las | That sounds unfortunate | 12:44:30 |
| Las | https://github.com/haskell/haskell-language-server/pull/1635 It seems to have basic support since a few days ago? | 12:45:40 |
| Las | There is also https://github.com/haskell/haskell-language-server/pull/1649 | 12:46:21 |
 toonn | Las: Maybe haskell.nix is a better fit for you. | 12:51:39 |
| Las | Maybe yeah | 12:52:06 |
| toonn | The Nixpkgs Haskell infra's primary (only?) goal is to facilitate providing Haskell programs in Nixpkgs. | 12:52:18 |
| maralorn | toonn: That’s not how I see it. | 12:52:33 |
| toonn | It's really cool how usable it can be for general development but that is not it's goal. | 12:52:50 |
| toonn | *its | 12:53:11 |
| maralorn | I think providing a large consistent collection of precompiled libraries is great. | 12:53:23 |
| toonn | Sure, but that's what stackage is. | 12:53:46 |
| maralorn | toonn: Ours is larger ;-) | 12:54:05 |
| toonn | As soon as you step outside of the stackage purview in the Nixpkgs infra you open yourself up to trouble. | 12:54:12 |
| toonn | It's not really. | 12:54:17 |
| toonn | You said a consistent collection. | 12:54:25 |
| toonn | I assume building together is implied by that. | 12:54:35 |
| maralorn | Yeah | 12:54:47 |
| toonn | And that's only reliably true for the Stackage subset. | 12:54:52 |
| maralorn | toonn: And for everything that we actively maintain. | 12:55:55 |
| toonn | Maybe. I don't know whether anyone does build matrices comparing to all of stackage? | 12:57:03 |
| toonn | And my original point still stands. If you are the kind of dev that always wants the latest GHC, the infra isn't really for you. | 12:57:45 |
| maralorn | In reply to @toonn:matrix.org
And my original point still stands. If you are the kind of dev that always wants the latest GHC, the infra isn't really for you. Yeah, that is certainly true. | 12:58:16 |
