| 7 Jun 2021 |
 joe (he/him) | In reply to @sternenseemann:systemli.org
what's frustrating me about the situation is that none of the current haskell maintainers has a stake in darwin, but due to the channel situation we need to invest time into keeping a significant part of the ecosystem on darwin working Do we know anyone who does use darwin nix and haskell? | 12:45:25 |
 sterni (he/him) | In reply to @domenkozar:matrix.org
There are people heavily invested into darwin+GHC, maybe we should invite them into the team? well they don't seem to contribute too much tbh | 12:45:42 |
 Domen Kožar | I think that's more of a communication issue instead of no users issue | 12:46:20 |
| Domen Kožar | but both are assumptions :) | 12:46:33 |
 maralorn | In reply to @sternenseemann:systemli.org
what's frustrating me about the situation is that none of the current haskell maintainers has a stake in darwin, but due to the channel situation we need to invest time into keeping a significant part of the ecosystem on darwin working The interesting thing is, that ghc is actually working on ghc, and from my outside perspective it has been doing that relatively realiable for quite a while. Still peti did not care about this as far as I can tell? So there have to be some people who have been jumping in when there where serious issues? | 12:46:46 |
| maralorn | In reply to @sternenseemann:systemli.org
what's frustrating me about the situation is that none of the current haskell maintainers has a stake in darwin, but due to the channel situation we need to invest time into keeping a significant part of the ecosystem on darwin working * The interesting thing is, that ghc is actually working on darwin, and from my outside perspective it has been doing that relatively realiable for quite a while. Still peti did not care about this as far as I can tell? So there have to be some people who have been jumping in when there where serious issues? | 12:47:02 |
| maralorn | * The interesting thing is, that ghc is actually working on darwin, and from my outside perspective it has been doing that relatively realiable for quite a while. Still peti did not care about this as far as I can tell? So there have to be some people who have been jumping in when there were serious issues? | 12:47:19 |
| sterni (he/him) | well it works to a lesser degree than on linux actually | 12:47:21 |
| Domen Kožar | I know that obsidian did a lot of work to support darwin+GHC | 12:47:27 |
| Domen Kožar | like John and Matthew | 12:47:38 |
| sterni (he/him) | not that active anymore, but yeah that's true | 12:48:05 |
| Domen Kožar | same for Angerman | 12:48:12 |
| sterni (he/him) | ghc.withPackages doesn't work properly on macOS and I think standalone ghc (which is not in stdenv) can't currently compile non-trivial packages | 12:48:26 |
| sterni (he/him) | i. e. tests.writers fails on darwin because of that | 12:48:36 |
| Domen Kožar | Are there issues opened for these problems? | 12:48:50 |
| Domen Kožar | (getting to the bottom if this is a communication problem or lack of resources) | 12:49:09 |
| sterni (he/him) | https://github.com/NixOS/nixpkgs/issues/123228, I have even started on fixing the issues; fixed the stdenv assumption already, but haven't been able to debug the linker issue fully yet https://github.com/NixOS/nixpkgs/pull/125367 | 12:49:51 |
| sterni (he/him) | looks similar to a bug matthew fixed in the past, but may be a new or related regression, I have no idea yet | 12:50:16 |
| Domen Kožar | So it seems that the main problem is to pull in people that care about darwin+GHC into such issues, as it's hard to keep up with nixpkgs issues in general | 12:53:49 |
| sterni (he/him) | yeah, definitely | 12:54:15 |
| sterni (he/him) | John Ericson: I wonder whether this actually did or did not happen in the meantime since we still use those flags: https://github.com/NixOS/nixpkgs/issues/46814#issuecomment-424792721 | 12:55:34 |
| sterni (he/him) | okay, hm this is definitely a different issue than the issue back then | 12:56:34 |
| sterni (he/him) | back then the issue was that -liconv was missing, nowadays the -L flag for iconv is missing, but -liconv is passed correctly | 12:56:59 |
| sterni (he/him) | cdepillabout: maralorn: seems like doctests with GHC 8.10.5 are broken, I hope this can be fixed by just going ahead of stackage for the doctests package | 12:59:24 |
| maralorn | In reply to @domenkozar:matrix.org
So it seems that the main problem is to pull in people that care about darwin+GHC into such issues, as it's hard to keep up with nixpkgs issues in general Technically if people were to regularly filter for 6.topic: darwin and 6.topic: haskell that would help. But some kind of push based solution is probably more effective. | 13:00:20 |
| Domen Kožar | Yeah, we need something like that for whole nixpkgs | 13:00:45 |
| maralorn | In reply to @sternenseemann:systemli.org
cdepillabout: maralorn: seems like doctests with GHC 8.10.5 are broken, I hope this can be fixed by just going ahead of stackage for the doctests package Is it really an issues with doctest? | 13:00:49 |
| maralorn | It says doctests: unable to load package ghci-8.10.5'`. Couldn‘t that be also an error in how we setup the build environment? | 13:02:07 |
| sterni (he/him) | we'll see I have no idea yet | 13:02:59 |
 siraben | sterni (he/him): indeed, having more darwin testers would be good | 13:07:30 |
