Haskell in Nixpkgs/NixOS | 730 Members | |
| For discussions and questions about Haskell with Nix, cabal2nix and haskellPackages in nixpkgs | Current Docs: https://haskell4nix.readthedocs.io/ | More Nix: #community:nixos.org | More Haskell: #haskell-space:matrix.org | 146 Servers |
| Sender | Message | Time |
|---|---|---|
| 31 Dec 2024 | ||
| Huh, I just enabled the multipleComponents setting and suddenly the extension works again. Let's see for how long | 17:23:57 | |
| Problem is it's hard to say if it just randomly works again because i restarted vscode or if the setting actually helped | 17:24:32 | |
| I haven't enabled it myself because my old ass vscode extension doesn't support it, so I know it's not absolutely necessary | 17:25:00 | |
| But I don't use stack so that could matter | 17:25:10 | |
In reply to @megmug:matrix.orgYeah you want to restart once per change so you later know what was actually needed | 17:25:44 | |
| It makes sense to start from the 'sane defaults' state where you seem to be now anyway | 17:26:40 | |
| alexfmpe: Anyways, maybe your tip has helped me, at least for now, so thanks for that! Have been trying stuff for hours and happy that it works again at all 😅 | 17:29:47 | |
| 18:10:43 | ||
| hi I'm trying to cross compile nix-tree (from x86_64 to armv7, but I have the same issue to aarch64) and I get this error:
is there a standard solution to provide this -fexternal-interpreter ? | 19:40:45 | |
symphorien: no, not at the moment. haskell.nix has a scheme to start the iserv via qemu user mode emulation which we should eventually copy (last time someone tried they ran into some issues, but they may be solvable) | 20:50:27 | |
| if you can be bothered to set up haskell.nix, it should work | 20:50:47 | |
| not really, I have never touched haskell | 20:51:14 | |
| other question: when I cross compile nixfmt I get a cycle error: | 20:56:15 | |
hm interesting I've never noticed that, will need to investigate that. I guess dead code elimination doesn't work there… You can probably work around it by using haskell.lib.compose.overrideCabal { enableSeparateBinOutput = false; } nixfmt. | 20:59:53 | |
| I must be holding it wrong | 21:05:56 | |
|
| 21:06:12 | |
| interesting it doesn't work if the output is preselected, probably some fix point is wrong there | 21:11:19 | |
try haskellPackages.nixfmt instead that works! | 21:11:27 | |
| the interpreter issue is problematic because nixos now depends on shellcheck and shellcheck also exhbits this issue | 21:11:32 | |
| I cannot build ghc natively on armv7 because the boostrap segfaults and presumbaly it would oom and cross is blocked on this interpreter thing | 21:12:22 | |
| hum now that I think of it nixos should depend on buildPackages.shellcheck... | 21:14:00 | |
| Yeah, that's unfortunate. The shellcheck stuff is unfortunate and should be toned down a bit imo. | 21:14:30 | |
| I'm not sure when I'll have a minute again to look into external interpreter unfortunately. | 21:14:50 | |
| for now I overlay these haskell dependencies with a derivation that contains an empty shell script, and my system builds fine :) | 21:15:37 | |
| The armv7l bootstrap problem is annoying, unfortunately upstream has stopped providing bindists. We can cross compile GHC itself now, so we can hopefully just build our own bootstrap compilers, but we've yet to get the infrastructure for that up | 21:15:39 | |
| I see | 21:15:51 | |
| anyway it could be argued that spending time on armv7 is just wasteful | 21:17:32 | |
| feel like we're at the point where 32-bit ARM support in Nixpkgs will inevitably get worse over time rather than better :/ | 21:19:31 | |
| yes | 21:20:25 | |
| also y2k38 is coming :) | 21:20:49 | |