| 28 Sep 2025 |
|  libman joined the room. | 05:35:06 |
 sterni | bglgwyng: as far as I know, there is no way to specify the build targets at configure time | 10:28:22 |
| sterni | The package database is prepared by us and includes all dependencies Cabal reports to us when generating the expression. Not sure if you could influence this to trim down to specific targets, in any case cabal2nix does not support this at the moment. | 10:29:38 |
| sterni | interesting that doesn't seem to be documented? | 10:31:01 |
| sterni | it would be nice if we could set testTargets and buildTargets at configure time already | 10:31:28 |
 bglgwyng | Indeed I tried the same thing but it didn't work | 11:46:44 |
| bglgwyng | Maybe haskell.nix uses its own patched version? | 11:47:06 |
| bglgwyng | No, it doesn't. | 13:20:35 |
| sterni | maybe it is possible with cabal-install, but not Setup.hs? though the cabal(1) man page doesn't seem to indicate this… | 16:33:29 |
| 29 Sep 2025 |
|  @tomodachi94:matrix.org left the room. | 03:15:05 |
| bglgwyng | I still can't reproduce haskell.nix's behavior, which I didn't check if it work though. But I'm failing with both cabal-instsall and Setup.hs. | 03:28:08 |
|  ivy joined the room. | 04:22:03 |
| ivy | i wanna see if i can fix kmonad on macos | 04:23:39 |
| ivy | im unsure how i can "unbreak" like i deleted the line from main.yaml | 04:23:58 |
| ivy | but how do i regenerate package-list | 04:24:11 |
| ivy | cause it builds for me on my mac | 04:49:30 |
| ivy | so idk why considered "broken" | 04:49:35 |
| ivy | if its the drivers i have a pr for that | 04:49:45 |
| bglgwyng | I hardwired CABAL_CONFIG and CABAL_DIR as Nix store paths in nix-x-cabal. And it broke hls since it tries to build on start up emitting intermidiate outputs in CABAL_DIR.
Is it basically a bad idea to do so, even though I can figure out how to work around this specific issue? | 06:11:52 |
| bglgwyng | * I hardwired CABAL_CONFIG and CABAL_DIR as Nix store paths in nix-x-cabal. And it broke HLS since it tries to build on start up emitting intermidiate outputs in CABAL_DIR.
Is it basically a bad idea to do so, even though I can figure out how to work around this specific issue? | 06:24:51 |
| sterni | ivy: I've looked into it and it seems like it was broken at one point and the Nixpkgs maintainer had no access to a Darwin machine to investigate the failure. Probably has been resolved by Apple SDK changes or whatever. I've proposed the correct meta data update at https://github.com/NixOS/nixpkgs/pull/447134. | 10:35:13 |
| sterni | It is currently not (easily) possible to regenerate hackage-packages.nix on Darwin since the tool that does it uses a fs based database that can't be on a case insensitive file system. | 10:36:09 |
| ivy | done it all | 11:51:37 |
| ivy | ahh easy | 11:51:42 |
| ivy | works on mine perfectly | 11:51:46 |
| ivy | let's get that metadata fixed then, and i can rebase on it | 11:52:20 |
| ivy | also sterni i dont think ur ofborg will work | 11:54:26 |
| 30 Sep 2025 |
| bglgwyng | Does nixpkgs.haskell do sometihng special on HLS, wrapping with some env vars or something?
I found that HLS can't find the right package db I configured. | 04:23:10 |
| bglgwyng | https://discord.com/channels/280033776820813825/505370075402862594/1422409401859575969
this is relevant | 04:23:18 |
 maralorn | No, it doesn’t. HLS asks e.g. cabal-install, cabal-install looks at ghc. ghc has the right package db. It’s relevant that you start hls from the right nix-shell for your projetc so that the correct ghc is available. | 07:11:10 |
