| 4 Sep 2025 |
 Uraraka ~ Ochaco | Absolutely a thing we would need so far to continue | 12:25:17 |
 emily | ok, how are you meant to go from an LLVM commit hash to the merge request that contains it? | 13:52:57 |
| emily | the GitLab UI completely fails to show the link since they use a merge bot | 13:53:07 |
 Alyssa Ross | LLVM? They use GitHub | 13:54:33 |
| emily | whoops. sorry. GHC. | 14:05:13 |
| emily | you see, I am working with LLVM-related GHC patches :P | 14:05:28 |
 Teo (he/him) | Do you have the commit? | 14:08:55 |
| Teo (he/him) | * Could you share the commit SHA? | 14:09:10 |
| Teo (he/him) | If you open the commit, you'll normally see back links at the bottom of the page from marge bot mentions | 14:10:19 |
|  angerman set a profile picture. | 21:15:40 |
|  Puna joined the room. | 21:21:56 |
| 5 Sep 2025 |
 John Ericson | If anyone wants to help me debug https://github.com/haskell-nix/hnix-store/pull/289 that would be great. Just wishing aloud in case I'm lucky | 01:30:55 |
| * John Ericson sleeps now | 01:31:03 |
|  @dmjio:matrix.org left the room. | 02:09:38 |
 maralorn | Aaah, what is the point of having ghc-lib-parser spanning three ghc versions when fourmolu and ormolu track cabal-syntax anyway which is tightly coupled to the ghc release … | 12:16:17 |
| maralorn | I have now the choice of downgrading ghc-lib-parser, fourmolu and ormolu to build hls on ghc 9.10 or to upgrade Cabal/Cabal-syntax. Both either globally or in a quite large overrideScope. | 12:17:35 |
| maralorn | I feel like I am overlooking something because that didn’t feel as painful before … | 12:18:04 |
 Artem | maralorn: the very strict bounds that Ormolu puts on Cabal-syntax look strange. I'd ask on their bug tracker exactly the question you're posing above: if you use ghc-lib-parser, you probably want to span over at least couple of GHC, but you also always limit Cabal-syntax to one version, so this looks like mixed messages... They always bump Cabal-syntax when they bump ghc-lib-parser, so they at least understand some connection they just overly pessimistic about it it looks like. | 15:10:21 |
| maralorn | I guess using ghc-lib-parser might have other advantages than just the compatibility range … | 15:14:22 |
| Artem | maybe, but there seems to be no reason to be as strict about Cabal-syntax bounds: they use just a few basic things from it, and those don't really change | 15:20:45 |
| maralorn | Okay | 15:22:12 |
| maralorn | It seems I am loosing my grip on being the nixpkgs maintainer when I didn’t first check whether a jailbreak works. 😄 | 15:22:36 |
| maralorn | * It seems I am loosing my grip on being a nixpkgs maintainer when I didn’t first check whether a jailbreak works. 😄 | 15:22:44 |
| maralorn | FTR, it didn’t. | 23:05:12 |
| 6 Sep 2025 |
 sterni (he/him) | the answer seems to be: detective work | 10:42:24 |
| sterni (he/him) | it is also extremely annoying to tell what versions of GHC a commit has been backported to | 10:42:49 |
| Teo (he/him) | Once you find the MR then you should be able to know the backporta since release managers leave comments when stuff gets backported, but all of it is so much friction. I occasionally consider writing a little web service to get this type of info | 11:01:37 |
| sterni (he/him) | I guess best bet is to go via the linked issue if there's one in the commit message since that's a requirement now | 11:04:31 |
| sterni (he/him) | maralorn: Cabal-syntax is reinstallable, ghc not; so that's probably the reason | 11:05:00 |
| maralorn | In reply to @sternenseemann:systemli.org
maralorn: Cabal-syntax is reinstallable, ghc not; so that's probably the reason Good point | 11:11:59 |
