| 10 Jan 2026 |
tks_123 | are there examples of this out there? | 13:03:45 |
Acid Bong | In reply to @tks_123:matrix.org Is there a way to add a haskell package as a flake (i.e. some other haskell project with its own flake.nix) into the haskellPackages of my own haskell project? yup, but i hope they expose a package.nix with the package recipe alone, so you could haskellPackages.callPackage it | 13:25:26 |
Acid Bong | (for the sake of using the same version and edition of GHC) | 13:26:01 |
maralorn | Its complicated, the default flake interface is not really meant to be composable. | 16:03:30 |
maralorn | Some flake authors design their flake in a way that they expose a) the package and b) an overlay which injects the package. | 16:04:05 |
| aiya joined the room. | 19:38:17 |
aiya | so the 25.11 "default" GHC is 9.10, but LLVM only has 18-20 which is incompatible. is this intentional, or is LLVM 15 supposed to be in nixpkgs just to support 9.10 -fllvm?
i back-searched this chat for LLVM but only found conversation on bumping LLVM 12 -> 15 | 19:50:59 |
emily | we backport patches for newer LLVM support to older GHCs | 19:52:34 |
emily | (there aren't that many, the backend doesn't change much since it uses the LLVM textual IR format) | 19:52:52 |
emily | (mostly it's just backporting bumps to the version number it checks) | 19:53:02 |
emily | see pkgs/development/compilers/ghc/common-llvm-patches.nix | 19:53:14 |
| 11 Jan 2026 |
| ivan joined the room. | 01:54:59 |
iqubic (she/her) | What should I do if Nix is refusing to build a Haskell Library for me? Right now, hip is failing to build! https://hackage.haskell.org/package/hip | 03:30:09 |
iqubic (she/her) | error: Package ‘hip-1.5.6.0’ in /nix/store/aqhcr0kv7d91rfm0ca77rbakjk5gkbq6-source/pkgs/development/haskell-modules/hackage-packages.nix:330406 is marked as broken, refusing to evaluate.
a) To temporarily allow broken packages, you can use an environment variable
for a single invocation of the nix tools.
$ export NIXPKGS_ALLOW_BROKEN=1
Note: When using `nix shell`, `nix build`, `nix develop`, etc with a flake,
then pass `--impure` in order to allow use of environment variables.
b) For `nixos-rebuild` you can set
{ nixpkgs.config.allowBroken = true; }
in configuration.nix to override this.
c) For `nix-env`, `nix-build`, `nix-shell` or any other Nix command you can add
{ allowBroken = true; }
to ~/.config/nixpkgs/config.nix.
note: trace involved the following derivations:
derivation 'mh26'
| 03:36:54 |
| Dr Mayhem, esq changed their display name from Mr Mayhem to Dr Mayhem, esq. | 13:34:01 |
| ghpzin changed their display name from ghpzin (moved to @ghpzin:envs.net) to ghpzin. | 15:04:53 |
| @ghpzin:envs.net left the room. | 16:16:06 |
| 12 Jan 2026 |
aiya | when you get that message, most likely it's a compilation error. try running NIXPKGS_ALLOW_BROKEN=1 nix-shell -p haskellPackages.hip to isolate the error | 02:03:47 |
aiya | here's the error i got when running that command: | 02:04:06 |
aiya | [34 of 40] Compiling Graphics.Image.Types ( src/Graphics/Image/Types.hs, dist/build/Graphics/Image/Types.o, dist/build/Graphics/Image/Types.dyn_o )
<no location info>: error:
Simplifier ticks exhausted
When trying RuleFired Class op $p8ColorSpace
To increase the limit, use -fsimpl-tick-factor=N (default 100).
In addition try adjusting -funfolding-case-threshold=N and
-funfolding-case-scaling=N for the module in question.
Using threshold=1 and scaling=5 should break most inlining loops.
If you need to increase the tick factor substantially, while also
adjusting unfolding parameters please file a bug report and
indicate the factor you needed.
If GHC was unable to complete compilation even with a very large factor
(a thousand or more), please consult the "Known bugs or infelicities"
section in the Users Guide before filing a report. There are a
few situations unlikely to occur in practical programs for which
simplifier non-termination has been judged acceptable.
To see detailed counts use -ddump-simpl-stats
Total ticks: 4241
| 02:04:12 |
aiya | someone's already opened an issue about this upstream | 02:11:34 |
aiya | * someone's already opened an issue about this upstream
as for a fix, you can always increase the tick tolerance as specified in the error message above or reduce inlining in colorSpace.hs | 02:12:23 |
aiya | so i'm trying to fix a package that requires random>=1.3 as a dependency but the version of random in nixpkgs is 1.2. what's the best practice here? do i: | 02:19:34 |
aiya | * so i'm trying to fix a package that requires random>=1.3 as a dependency but the version of random in nixpkgs is 1.2. what's the best practice here? do i:
- create a new package
random_1_3_1
- try and back-patch the package to work with
random 1.2
- override the build deps for that specific package's derivation? (i don't like this option, other packages require
random>=1.3 too)
| 02:21:05 |
aiya | * so i'm trying to fix a package that requires random>=1.3 as a dependency but the version of random in nixpkgs is 1.2. what's the best practice here? do i:
- create a new package
random_1_3_1
- try and back-patch the package to work with
random 1.2
- update
random to 1.3 (probably not recommended, stackage lts has it as 1.2.1)
- override the build deps for that specific package's derivation? (i don't like this option, other packages require
random>=1.3 too)
i couldn't find specific guidance on CONTRIBUTING.md and while i know some packages get multiple version i've yet to see a Haskell non-core package have multiple versions on nixpkgs | 02:23:40 |
| aiya set a profile picture. | 02:33:02 |
| aiya changed their profile picture. | 02:34:54 |
| aiya changed their profile picture. | 02:36:52 |
Artem | @aiya:catgirl.cloud: random_1_3_1 is already there (I think because haskellPackages by default include LTS versions and latest versions too) | 04:56:53 |
Artem | trying to fix liquidhaskell in nixpkgs in the light of https://github.com/ucsd-progsys/liquidhaskell/issues/2188:
https://github.com/NixOS/nixpkgs/pull/479484 | 20:52:29 |