| 14 May 2025 |
 thirdofmay18081814goya | (ty for answer) | 17:20:27 |
| thirdofmay18081814goya | * (ty for answer, forgot lol) | 17:20:46 |
| thirdofmay18081814goya | * (ty for answer) | 17:21:26 |
|  keypusher joined the room. | 17:44:11 |
| keypusher | What is a good approach building a multi library project with nix? It seems cabal 3's "multiple libraries per cabal file" doesn't play well with nix. Which leads me to believe it's a better bet to keep separate cabal file per library? | 18:25:02 |
| keypusher | "seems" here is from memory and also google search on the subject that led me to a thread that seems to confirm it didn't work well by end of 2024. | 18:28:44 |
 alexfmpe | "doesn't play well with nix" ? | 19:18:45 |
| 15 May 2025 |
 Profpatsch | keypusher: yeah I’d assume with the nixpkgs haskell system you need a separate cabal | 17:17:13 |
| Profpatsch | dunno about haskell.nix | 17:17:22 |
| alexfmpe | I'm confused on what the issue is | 18:00:54 |
| alexfmpe | There's plenty of multi component packages on nixpkgs | 18:01:13 |
| alexfmpe | well, I guess one ergonomic issue I'm aware of is a single callCabal2nix derivation will build all components with different flags or for different platforms | 18:05:10 |
| alexfmpe | * well, I guess one ergonomic issue I'm aware of is a single callCabal2nix derivation will build all components with the same flags for the same platform, etc | 18:05:25 |
| alexfmpe | so if you wanted different flags for each you'd need to build all the components with each configuration | 18:05:55 |
 maralorn | I assume the support is not great. Because our generated derivation does not expose separate components. So e.g. the library dependencies are a union over all dependencies. | 18:06:07 |
| alexfmpe | yeah granularity could be improved | 18:06:35 |
| alexfmpe | but "didn't work well" sounds like there's something broken somewhere? | 18:07:23 |
 ozkutuk | Running the regenerate-hackage-packages.sh script fails with the following, any ideas?
Obtaining Hackage data …
Generating compiler configuration …
Running hackage2nix to regenerate pkgs/development/haskell-modules/hackage-packages.nix …
Regenerating transitive-broken.yaml … (pass --fast to maintainers/scripts/haskell/regenerate-hackage-packages.sh to skip this step)
error: access to absolute path '/etc/static' is forbidden in restricted mode
| 18:22:42 |
| ozkutuk | running it with --fast is fine, so it really is a problem with regenerate-transitive-broken-packages.sh | 18:27:39 |
|  Aleks Rūtiņš changed their profile picture. | 19:02:05 |
| Aleks Rūtiņš changed their profile picture. | 19:02:50 |
 sterni (he/him) | ozkutuk: looks similar to https://github.com/NixOS/nixpkgs/issues/400784 | 21:00:36 |
| sterni (he/him) | no clue what's the problem there | 21:00:41 |
| alexfmpe | ok I give, in hackage2nix/main.yaml, under supported-platforms how do I specify android? | 21:46:00 |
| 16 May 2025 |
| sterni (he/him) | the platform predicates stuff isn't supported if you need that | 10:25:55 |
| sterni (he/him) | not sure if you can express it with normal tuples somehow. | 10:26:46 |
| alexfmpe | eh? tuples? | 16:31:58 |
| alexfmpe | I mean it's not a big deal either way
the specific package I wanted to move from broken to android-only is android-activity
I don't expect anyone to try and build for non-android so in a way it can probably just be removed from the list
but then hydra will just add it back | 16:35:26 |
| alexfmpe | if doing it rightTM is a lot of work right now, maybe I just manually unmarkBroken conditional on aarch64-android ? | 16:35:47 |
| alexfmpe | in configuration-arm.nix or something | 16:36:06 |
