30 Oct 2022 |
Madoura | it's in nixpkgs, but yes | 07:58:47 |
pie_ | I have no idea how that part of the nix infra works but apparently the stdenv compiler stuff is kind of complicated | 07:58:56 |
pie_ | well its good that its already in nixpkgs | 07:59:04 |
pie_ | I dont know if for something relatively obsruce like rocm whether that means it should work or not | 07:59:29 |
pie_ | youve tried google? | 07:59:38 |
pie_ | wait | 07:59:55 |
pie_ | wait | 07:59:56 |
pie_ | wait | 07:59:57 |
Madoura | Yeah, can't find anything related. | 08:00:00 |
pie_ | is it just the path | 08:00:07 |
pie_ | or is it trying to pull this from gcc | 08:00:12 |
pie_ | becaues thats probably yourproblem right there if it is | 08:00:20 |
Madoura | That's what I thought too, it could possibly be toolchain mixing. | 08:00:26 |
pie_ | its almost definiteƩy | 08:00:33 |
pie_ | * its almost definitely | 08:00:36 |
Madoura | That's something I have noooo idea how to fix. | 08:00:40 |
pie_ | I think I managed that at some point before but I dont remember how | 08:00:51 |
pie_ | well, for starters we need to figure out where its coming from | 08:01:00 |
pie_ | maybe try nix why-depends on the derivations? | 08:01:06 |
Madoura | Good idea. Haven't used that before, nix why --depends ? | 08:01:35 |
pie_ | have you swapped out the stdenv? | 08:01:36 |
| * pie_ tries to check script | 08:01:41 |
Madoura | Yeah, tried with stdenvNoCC | 08:01:50 |
pie_ | thats still gcc isnt it? I honestly have nmo idea | 08:02:03 |
pie_ | I gues it would make sense that it isnt because its called NoCC | 08:02:14 |
Madoura | yeah, it shouldn't have a compiler | 08:02:24 |
pie_ | but yes youre using stdenv in what you linked and im pretty sure that will at least be one of the starting points of the problem | 08:02:43 |
Madoura | Oh it's actually why-depends , okay. | 08:02:53 |
Madoura | I'll try it again as well as check with that nix command. | 08:03:07 |
| * pie_ tries to remember how he helped someone else before with mixed toolchain issues | 08:03:20 |