| 10 Feb 2025 |
 sterni (he/him) | maralorn: why did they even upstream 8.10.7 ghcjs lol | 17:05:33 |
 maralorn | Well, that was years ago and at that point they were using it?^^ | 17:06:02 |
| sterni (he/him) | it's getting to be just one big blur | 17:06:34 |
 emily | I'm tired of the fact that we have to keep backporting LLVM patches that require manual work to apply to LLVM 12 that we can't just skip and leave broken on Darwin because the GHC bootstrap pulls it in and GHC is now used by everything. and I'm tired of the constantly-shifting goalposts around GHC versions that there has never been any evidence that the people who would use them are actually using modern Nixpkgs and substantial amounts of evidence that they're not. | 17:06:49 |
| sterni (he/him) | emily: i.e. we kind of need 8.10.7 or 9.0.2 to bootstrap GHC 9.4.8 due to build system bugs. Simultaneously, these compilers are the only remaining compilers that are not affected by various problems relating to the build system. Without 9.4.8 we would be unable to cross compile GHC to riscv64 and do TemplateHaskell in pkgsStatic. | 17:08:32 |
| emily | if there's an agreement that they can be dropped if the internal uses are sorted, I'll help | 17:07:03 |
| emily | if you just personally don't want those versions to be removed, then ok | 17:07:17 |
| emily | but while it's still unclear if there will be any point to me as a Darwin/LLVM person to put in the effort I'm not going to allocate my limited time to it | 17:07:46 |
| emily | if a survey is what's required I'll wait for the survey. I already sorted things out with other users of old compilers like the CUDA team and I am convinced that GHC is the only substantive blocker | 17:08:37 |
| emily | riscv64 is tier nothing, it can wait for build system bugs to be fixed (assuming we can't just patch or work around them) | 17:09:13 |
| emily | avoiding an unsupported platform placing maintenance burdens on one of our major ones is part of the reason we have a vague notion of support tiers in the first place | 17:09:51 |
| emily | if static is a blocker then that can be discussed but then why was this not raised the previous 10 times this was talked about? | 17:10:44 |
| maralorn | sterni: If you want the details: reflex-plattform currently still pins older nixpkgs, no one is planing on bumping that pin while using ghcjs. Obsidian systems has already switched to using haskell.nix and the new ghc js backend is expected to be usable with that whole setup very soon. | 17:10:51 |
| maralorn | And I actively asked the people involved whether they’d be affected by us dropping ghcjs and they all said no. | 17:12:28 |
| emily | saying that there's no need for a survey because the internal users are what matter, and then saying that actually we'd need to survey external users when I say that the internal users seem tractable and I'm happy to help, and then bringing up new concerns about RISC-V and static when I say I'll wait for the survey, just makes it seem like you want to stall indefinitely | 17:16:28 |
| emily | if old GHCs are actually required and that's just a fact of life then sure, I can deal with that, although it would be nice if the GHC maintainers took over responsibility for quickly backporting fixes to old LLVMs that they are consequently the only substantive users of | 17:17:35 |
| emily | but I haven't seen any evidence that more than one person wants to keep them, and plenty of evidence that nobody else really cares, but the reasons pile keeps growing every time it's brought up all the same | 17:18:09 |
| emily | I'm happy to put in effort to solve any technical obstacles, or wait for a survey or whatever else, but I'm not convinced that technical obstacles are what's blocking here | 17:18:52 |
|  mjm joined the room. | 17:19:48 |
| emily | GHC is in the critical path for building a working Nixpkgs. the bootstrap pulls in an unsupported LLVM that doesn't build with modern compilers without patching and keeps requiring manual work to backport fixes. it has broken during recent staging cycles and it has fallen on Darwin maintainers to fix it because we cannot just ignore it. if there are substantive technical blockers to fixing this, then I want to work to address them. if there is a meaningful external userbase that relies on using those unsupported compilers with a supported Nixpkgs then I want to hear from them and discuss what the long-term options are. but it is a major burnout factor to have to repeatedly take responsibility for maintaining a codebase that we would rather be outside our purview because it very indirectly blocks the world and one maintainer can't decide what the reason it has to stay is | 17:33:32 |
| emily | if a survey will settle the issue then I'll wait for a survey. but the reason there are still internal users is that it seems like a bad use of my time and energy when I get the feeling there'd be another reason even if static and RISC-V were solved and the survey got 1000 responses all saying they don't care about it | 17:34:41 |
| emily | so instead I spend my time shepherding LLVM 12 fixes because I have no choice other than giving up on Darwin maintenance | 17:35:59 |
| emily | for the benefit of what may well be zero users who want to use old ghcjs on a fresh Nixpkgs release that is still over 3 months away | 17:36:36 |
| maralorn | sterni: I am currently not super motivated to do that survey. But what do you think about dusting of my deprecation proposal: https://md.darmstadt.ccc.de/FZwe_M9GQwWVnYfO0U4qJA (Would need a few updates.) The idea is still to announce it loud and wide and then people can complain, which I don’t expect them to do. The goal here is to give ourselves the permission to remove 8.10-9.2 if we can manage it. Of course we wouldn’t do it if we can’t manage building stuff like elm or bootstraping relevant configurations, but that’s orthogonal. No deprecation policy can force us to break our builds. This is just to unblock contributors like emily so that we are clear about internal and external requirements and expectations. | 18:05:09 |
 Tristan Ross | In reply to @emilazy:matrix.org
GHC is in the critical path for building a working Nixpkgs. the bootstrap pulls in an unsupported LLVM that doesn't build with modern compilers without patching and keeps requiring manual work to backport fixes. it has broken during recent staging cycles and it has fallen on Darwin maintainers to fix it because we cannot just ignore it. if there are substantive technical blockers to fixing this, then I want to work to address them. if there is a meaningful external userbase that relies on using those unsupported compilers with a supported Nixpkgs then I want to hear from them and discuss what the long-term options are. but it is a major burnout factor to have to repeatedly take responsibility for maintaining a codebase that we would rather be outside our purview because it very indirectly blocks the world and one maintainer can't decide what the reason it has to stay is Yeah, this is why I'm being kinda strict with removing LLVM 12. Less of a burden. And with the Darwin, really seems much more of a reason to remove it. We've basically got everything else except GHC off LLVM 12 so GHC is the last thing standing. It's the last release pre monorepo, doesn't get any patches. | 19:19:13 |
 alexfmpe | In reply to @maralorn:maralorn.de
sterni: I am here. I actually use ghcjs every day. I can tell you that we can drop it. Also use it, also on said call and I'm with maralorn on this.
Kill it with fire, brimstone and extreme prejudice. Miso and reflex-dom now build with pkgsCross.ghcjs.haskell.packages.ghc912 on haskell-updates or open PRs to it | 21:08:16 |
| alexfmpe | Obsidian isn't even fussing over nixpkgs haskell, they adopted haskell.nix | 21:09:55 |
| alexfmpe | Miso does use nixpkgs haskell but they're also on a 2 years old and in the process of jumping directly to modern nixpkgs | 21:12:28 |
| alexfmpe | Er, modern nixpkgs for 9.12 ghc | 21:13:08 |
|  Mr Mayhem changed their display name from magnolia_mayhem to Mississippi_metalhead_mailman_magnolia_mayhem. | 21:47:39 |
