| 29 May 2026 |
 Alex | Ok, found the commit (it was). | 11:04:03 |
 alexfmpe | In reply to @alexfmpe:matrix.org
Hmm cabal itself seems to default to -O1 Actually... would it make sense for us to use -O2 by default? | 17:01:12 |
| alexfmpe | I'm not sure exactly how much that slowdowns compilation and don't know the benefit for cpu/memory what with ghc not doing whole-program optimization, but it might also decrease binary size which is a win on nixpkgs, hydra etc? | 17:03:28 |
| alexfmpe | wonder if just building cabal itself with -O2 would make compiling everything else faster enough that it pays for itself | 17:04:52 |
 Teo (he/him) | Often -O2 is not worth it | 17:07:52 |
| Teo (he/him) | It's the sort of thing that should be benchmarked on a case by case basis | 17:08:41 |
| Teo (he/him) | Redacted or Malformed Event | 17:08:42 |
| 31 May 2026 |
 whispers [& it/fae] | hii! i have a procedural haskellPackages question. i have a few packages i'd eventually like to mark broken (haskellPackages.gi-soup2, haskellPackages.spike) because they depend on the libsoup_2_4, and i'd like to get rid of that (not yet, but planning to look into it soon). the dependency is defined in hackage-packages.nix, but aiui that's not something you're supposed to touch directly. where in nixpkgs would i (a) drop the dependency on the system library so that we don't get eval failures, and (b) mark it as broken? i'm looking through the manual and haskell-modules/HACKING.md but can't find anything that seems immediately relevant | 01:35:33 |
| whispers [& it/fae] | hii! i have a procedural haskellPackages question. i have a few packages i'd eventually like to mark broken (haskellPackages.gi-soup2, haskellPackages.spike) because they depend on the libsoup_2_4 derivation, and i'd like to get rid of that (not yet, but planning to look into it soon). the dependency is defined in hackage-packages.nix, but aiui that's not something you're supposed to touch directly. where in nixpkgs would i (a) drop the dependency on the system library so that we don't get eval failures, and (b) mark it as broken? i'm looking through the manual and haskell-modules/HACKING.md but can't find anything that seems immediately relevant | 02:02:58 |
| alexfmpe | probably https://github.com/NixOS/nixpkgs/blob/haskell-updates/pkgs/development/haskell-modules/configuration-common.nix | 13:33:21 |
| alexfmpe | or configuration-nix.nix, if it's due to nix(pkgs) specific reasons, rather than general distro things | 13:34:19 |
 sterni (he/him) | if the library is removed, the dependency will also be removed on the next regeneration. | 14:03:46 |
| sterni (he/him) | libs can be marked broken in the generator's configuration | 14:04:37 |
 aiya | Redacted or Malformed Event | 17:44:16 |
| whispers [& it/fae] | seems to work, thank you both! | 19:37:51 |
| 1 Jun 2026 |
 bglgwyng | Is is possible to use wasm compiler with pkgs.haskellPackages now? It doesn't seem so. | 13:33:25 |
| alexfmpe | TL;DR: not yet
https://github.com/NixOS/nixpkgs/pull/225000 | 13:59:22 |
| alexfmpe | Use ghc-wasm-meta for now | 13:59:34 |
| 3 Jun 2026 |
|  Jade w changed their display name from Philip W to Jade w. | 15:01:19 |
| 4 Jun 2026 |
|  Echo changed their profile picture. | 04:23:34 |
| sterni (he/him) | I think downgrading is no longer an option. https://github.com/NixOS/nixpkgs/pull/527962 | 12:53:48 |
 Artem | sterni: 4.17 is good enough: it doesn't break the PVP... | 12:54:51 |
| Artem | also, sterni , regarding https://github.com/NixOS/nixpkgs/pull/521260#issuecomment-4622508007 let me know what you think. I think https://github.com/NixOS/nixpkgs/pull/520243 looks much better than the last time you looked at it. Even it's LLM-assisted, I think it's a good step forward. | 14:02:50 |
| alexfmpe | on our stackage nightly bump, HLS is broken on 9.10 due to apply-refact
what I don't understand is how we made it work on master as that dep was always there on cabal but was missing from hackage-packages.nix | 14:52:34 |
| alexfmpe | * on our stackage nightly bump, HLS is broken on 9.10 due to its dep, apply-refact, being broken on 9.10 period
what I don't understand is how we made it work on master as that dep was always there on cabal but was missing from hackage-packages.nix | 14:52:47 |
| alexfmpe | huuuuh technically the dependency is on HLS's cabal file, but on the hls-hlint-plugin sublibrary | 14:53:54 |
| alexfmpe | we've been adding apply-refact manually
https://github.com/NixOS/nixpkgs/blob/0a130039b0ecb0f9690581660080baca073fdd5a/pkgs/development/haskell-modules/configuration-common.nix#L237
which makes sense given the dep wasn't on hackage-packages.nix (presumably due to it being behind a cabal conditional)
but then why does the dep show up now | 14:59:46 |
| sterni (he/him) | because the conditionals are resolved with impl being ghc 9.12 | 17:44:09 |
| alexfmpe | Aaah | 17:44:22 |
| 6 Jun 2026 |
 hellwolf | trying to fixing my part in haskell-updates. however, facing this optics package issue Is it being dealt:
error: Cannot build '/nix/store/6fj48f1agb8y7wyvjxqv8lmc6jabaaxq-optics-0.4.2.1.drv'.
Reason: builder failed with exit code 1.
Output paths:
/nix/store/2gvmjpzcgqa8xhq47mz9xrd5w2hm86pw-optics-0.4.2.1
/nix/store/xkdgyjjf6wvb5lp51m940ffrg81lwywf-optics-0.4.2.1-doc
| 11:38:19 |
