| 25 May 2025 |
 Alyssa Ross | I'll just open a PR :) | 10:38:50 |
 nh2 | Is
https://github.com/NixOS/nixpkgs/blob/4a2c6cc55d3f21dfe98ed50b171dcc3fadbec294/pkgs/development/haskell-modules/HACKING.md
outdated somehow because
regenerate-hackage-packages.sh
generates formatting disagreeing with nix-shell --run treefmt? | 12:17:27 |
 maralorn | How so? That shouldn’t be the case. | 12:21:42 |
| maralorn | We just added the formatting to regenerate-hackage-packages.sh a few weeks ago? (around the time when formatting became mandatory.) | 12:22:32 |
| nh2 | Looks like when I ran ./maintainers/scripts/haskell/regenerate-hackage-packages.sh it failed half-way with:
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 '/nix' is forbidden in restricted mode
Maybe that way it failed before the formatting run?
Not sure what the cause of that error is exactly
| 12:43:03 |
| nh2 | Another small question:
Am I doing this PR right / should it be against master?
https://github.com/NixOS/nixpkgs/pull/410785
| 12:46:12 |
| maralorn | Yeah, that’s fine against master. | 12:58:06 |
| maralorn | Huh, yeah, that is kinda troubling. | 12:59:16 |
| maralorn | You are right, I think the script failed before formatting. Question is ofc, why. | 12:59:52 |
 winston | hmm I just ran that script on c8d36fe and didn't get that error, but all I got were two maintainers changes, idk if that still means it succeeded | 13:16:05 |
 sterni (he/him) | nh2: https://github.com/NixOS/nixpkgs/issues/400784 | 13:55:35 |
| winston | ah interesting, I tried with Nix 2.28.3 | 13:58:12 |
| 27 May 2025 |
|  @irenes:matrix.org left the room. | 09:00:39 |
| 28 May 2025 |
| sterni (he/him) | Currently looks like GHC removal project will at least kill armv7l-linux and powerpc64le-linux for GHC (the latter was very unreliable anyways, so probably not a big loss). Unless, of course, we can cross compiled GHC as bootstrap seed to work. | 15:18:20 |
| sterni (he/him) | I think armv7l is de facto dead as well, but nixpkgs CI checks eval of it?? Probably more GHC availability checks need to be added before we can drop support. | 15:19:24 |
 emily | armv7 support in Nixpkgs is pretty bad at this point I think | 15:19:56 |
| maralorn | sterni: When I want GHC to set a flag when calling cc during a compilation. Is there an environment variable I can set in my Haskell package derivation? Or is it correct that I can only set that variable during compilation of GHC via CONF_CC_OPTS_STAGE2. It seems kinda fishy to me because I basically only want to affect runtime behavior of the compiler. | 15:43:27 |
 Alex | In reply to @sternenseemann:systemli.org
I think armv7l is de facto dead as well, but nixpkgs CI checks eval of it?? Probably more GHC availability checks need to be added before we can drop support. You could try asking in #exotic:nixos.org. | 15:48:51 |
 chreekat | In reply to @maralorn:maralorn.de
sterni: When I want GHC to set a flag when calling cc during a compilation. Is there an environment variable I can set in my Haskell package derivation? Or is it correct that I can only set that variable during compilation of GHC via CONF_CC_OPTS_STAGE2. It seems kinda fishy to me because I basically only want to affect runtime behavior of the compiler. I see a -optc and a -optlc (for llvm... Not sure why there's two) on GHC directly , but maybe you're asking about a different level of the stack | 16:44:04 |
| chreekat | https://downloads.haskell.org/ghc/latest/docs/users_guide/phases.html#ghc-flag-optc-option | 16:44:15 |
 Artem | In reply to @maralorn:maralorn.de
sterni: When I want GHC to set a flag when calling cc during a compilation. Is there an environment variable I can set in my Haskell package derivation? Or is it correct that I can only set that variable during compilation of GHC via CONF_CC_OPTS_STAGE2. It seems kinda fishy to me because I basically only want to affect runtime behavior of the compiler. I don't believe it can be done with env vars | 17:38:26 |
| Artem | I think it's only the flags that @b:chreekat.net: mentioned | 17:39:45 |
| Alex | In reply to @maralorn:maralorn.de
sterni: When I want GHC to set a flag when calling cc during a compilation. Is there an environment variable I can set in my Haskell package derivation? Or is it correct that I can only set that variable during compilation of GHC via CONF_CC_OPTS_STAGE2. It seems kinda fishy to me because I basically only want to affect runtime behavior of the compiler. Aren't you assuming that GHC calls cc?
It doesn't unless you're compiling C code or you have an unregisterised GHC.
Nixpkgs' cc-wrapper may allow you to add extra options via an environment variable.
I cannot recall what environment variables it reads. | 17:41:18 |
| maralorn | I think I figured it out. | 17:58:41 |
| maralorn | And yes ghc calls cc. For c code and for linking. | 17:58:41 |
| maralorn | Well I guess -optlc might actually do what I want also. | 18:19:06 |
| sterni (he/him) | -optc and friends is right yes | 18:54:28 |
| Alex | Doesn't it call ld for linking? | 19:22:42 |
| maralorn | Well, at least not for the js backend. There emscripten is used for linking iiuc. | 21:34:09 |
 alexfmpe | clearly we should have ghci running natively on browsers instead of compiling to javascript to interpret | 21:40:35 |
