| 15 Sep 2025 |
 Teo (he/him) | https://github.com/stable-haskell/ghc/pull/94/files | 13:47:50 |
 sterni (he/him) | Yeah, it seems the wiring it up is the problem at the moment with cabal(-install) alone, but Nixpkgs has an easier time there. | 13:47:58 |
 MangoIV | I see- so the idea is to just have some script that wrap around cabal | 13:48:03 |
 Alex | In reply to @sternenseemann:systemli.org
I would just start with adding a MicroHs package set built with GHC as we don't have hugs and then just iterate on that; maybe add hugs separately, have a hugs package set (if that even makes sense).
we don't have hugs
All it needs is an older GCC than the Nixpkgs default (or for someone to patch it to work with newer GCC). | 13:48:08 |
| sterni (he/him) | as long as it's not older than what we have in nixpkgs we can just package that shouldn't be a problem | 13:48:38 |
| sterni (he/him) | I guess still it may be easier to get MicroHs with GHC to work first before figuring out hugs, but you know that better. | 13:49:01 |
 emily | how old | 13:49:12 |
| emily | we will be dropping GCC 14 in 26.05 | 13:49:21 |
| emily | er | 13:49:34 |
| emily | 13 | 13:49:36 |
| emily | so I strongly recommend patching or -Wno-error over pinning an old GCC | 13:49:45 |
| emily | (it needs Lennart's branch of Hugs right? presumably we could just send PRs and treat that as the upstream) | 13:50:11 |
| Alex | Right now I'm using gcc13Stdenv, so I may need to take a closer look at the error from newer GCCs... | 13:50:16 |
| Alex | In reply to @emilazy:matrix.org
(it needs Lennart's branch of Hugs right? presumably we could just send PRs and treat that as the upstream) Sounds good to me. | 13:50:46 |
| emily | yeah I would prefer a moratorium on adding more GCC 13 at this point :P | 13:50:47 |
| emily | it's not hard to adapt stuff to newer versions | 13:50:54 |
| emily | just slightly tedious sometimes | 13:50:59 |
| Alex | In reply to @emilazy:matrix.org
it's not hard to adapt stuff to newer versions This one's non-trivial... Actually something wrong with Hugs' version of base IIRC (maybe a slightly less old base could be patched in?)
ERROR "libraries/bootlib/Foreign/C/Types.hs":173 - Undefined type constructor "NotReallyAType"
| 13:54:22 |
 John Ericson | Oo Haven caught up in all of this, but I like where things are going | 13:54:21 |
| John Ericson | *haven't | 13:54:25 |
| emily | this is at runtime? | 13:55:12 |
| emily | smells like latent UB in Hugs that's getting exposed | 13:55:23 |
| Alex | Nope, build time. | 13:55:29 |
| emily | but that's not C | 13:55:39 |
| emily | I mean it's Hugs outputting that right | 13:55:50 |
| emily | so it's Hugs runtime | 13:55:55 |
| Alex | The NotReallyAType comes from base's autoconf script somehow not finding the corresponding Haskell type for a few C types. | 13:56:31 |
| emily | I see | 13:58:14 |
| emily | probably there are new warnings/errors in GCC confusing it | 13:58:32 |
| emily | that happens a lot in autoconf | 13:58:41 |
