!RbXGJhHMsnQcNIDFWN:nixos.org

Haskell in Nixpkgs/NixOS

740 Members
For discussions and questions about Haskell with Nix, cabal2nix and haskellPackages in nixpkgs | Current Docs: https://haskell4nix.readthedocs.io/147 Servers

Load older messages


SenderMessageTime
6 May 2026
@alexfmpe:matrix.orgalexfmpeor switch to pkgs.haskell.lib.compose18:36:05
7 May 2026
@agahnim:matrix.orgAgahnim joined the room.12:02:42
@mangoiv.:matrix.orgMangoIV what is the reason for useLLVM being off-by-default? 14:21:39
@noiobeforebedtime:winesj.comJack (he/him)Wasn’t the person who made that switch. But it takes longer to compile, and that’s GHC’s default behavior unless you use -fllvm21:28:16
@alexfmpe:matrix.orgalexfmpeI think it's also currently broken for windows, something about split sections21:45:09
@alexfmpe:matrix.orgalexfmpethere was a recent PR fixing it, dunno if anything released had it backported21:45:27
8 May 2026
@mangoiv.:matrix.orgMangoIV I mean. Enabling only on platforms that it works on would go a long way. IMO it’s just a bit weird that you need to rebuild the compiler even though LLVM is basically a runtime dependency. 09:17:04
@emilazy:matrix.orgemilyI believe it affects the config11:34:02
@emilazy:matrix.orgemilylike the settings file stuff11:36:21
@emilazy:matrix.orgemilyidk if that could be split out to a separate derivation11:36:32
@emilazy:matrix.orgemilyLLVM is pretty chunky to pull in non-optionally11:36:46
@emilazy:matrix.orgemilyconsidering stuff depending on Pandoc, ShellCheck etc. and GHC's historical problems with LLVM version support11:37:16
@emilazy:matrix.orgemilyif it could just be a wrapper rebuild to use LLVM that would be one thing11:37:57
@emilazy:matrix.orgemilyoh11:38:53
@emilazy:matrix.orgemilyhttps://github.com/NixOS/nixpkgs/blob/8554fb1b7455107dda8184fff913e601254ab527/pkgs/development/haskell-modules/with-packages-wrapper.nix11:38:56
@emilazy:matrix.orgemilypretty sure you can just override withPackages and skip the compiler build11:39:12
@emilazy:matrix.orgemilyoverriding the compiler build itself is requesting it do the actual bootstrap with LLVM I think11:39:50
@teoc:matrix.orgTeo (he/him)This would be great imo11:41:05
@emilazy:matrix.orgemilywell I think ^ is that11:49:40
@emilazy:matrix.orgemilyhttps://github.com/NixOS/nixpkgs/blob/8554fb1b7455107dda8184fff913e601254ab527/pkgs/development/haskell-modules/with-packages-wrapper.nix#L20211:50:21
@emilazy:matrix.orgemily I guess it basically just adds it to the path and doesn't even add -fllvm itself? 11:51:44
@dandart:matrix.orgEmber changed their display name from Ember Void to Ember.15:21:36
@alex:tunstall.xyzAlex

Yes, to build the boot libraries and the stage2 compiler.

(This has been said before, but it would be really nice to find a way to split stage1 and stage2 into separate derivations.)

17:00:27
9 May 2026
@whispers:catgirl.cloudwhispers [& it/fae] joined the room.05:09:50
10 May 2026
@eveeifyeve:matrix.orgeveeifyeveA good question, has anyone seen if https://gitlab.haskell.org/ghc/ghc/-/commit/78ad81ecef846f73fee0f6c1a86cd6f19aa29b21.patch was backported to GHC 9.8? Because I only see 9.6.x and 9.10.x have it.14:00:26
@eveeifyeve:matrix.orgeveeifyeve* A good question, has anyone seen if https://gitlab.haskell.org/ghc/ghc/-/commit/78ad81ecef846f73fee0f6c1a86cd6f19aa29b21 was backported to GHC 9.8? Because I only see 9.6.x and 9.10.x have it.14:00:46
@alexfmpe:matrix.orgalexfmpedoesn't look it14:06:49
@alexfmpe:matrix.orgalexfmpehttps://gitlab.haskell.org/ghc/ghc/-/blob/ghc-9.8.4-release/m4/fp_find_cxx_std_lib.m414:06:51
@alexfmpe:matrix.orgalexfmpechanges are not present14:07:28
@ashinnv:matrix.orgMr Mayhem's Dumbass Labradoodle changed their display name from 👿👿👿M̸̙̜̔̇Ǎ̴͎̙͔G̸̞̈N̸͔͍̝͗͋̾Ő̷͖̼͈̽̚L̷̻͚̓̔I̷̛͔̰̟̔Å̴̩̍ ̷̦̒̇͝M̷̱̠̺̉̎A̵̼̎͗͘Ỹ̸̬̲͂̕H̷̙̖͂Ē̷͉̦̌͒M̶͈̥̽̐ 👹👹👹👹😈😈😈 to M̸̙̜̔̇Ǎ̴͎̙͔G̸̞̈N̸͔͍̝͗͋̾Ő̷͖̼͈̽̚L̷̻͚̓̔I̷̛͔̰̟̔Å̴̩̍ ̷̦̒̇͝M̷̱̠̺̉̎A̵̼̎͗͘Ỹ̸̬̲͂̕H̷̙̖͂Ē̷͉̦̌͒M̶͈̥̽̐.19:20:44

Show newer messages


Back to Room ListRoom Version: 6