| 3 Sep 2023 |
 raitobezarius | I still think getting rid of it is faster | 11:28:17 |
 artemis | pkgsCross is like the only reason i use nix haha | 11:40:58 |
| raitobezarius | Well, the good news is you will still be able to use it :) | 11:41:43 |
| artemis | what does getting rid of it entail then, moving it out of the nixpkgs tree? | 11:42:03 |
| raitobezarius | No | 12:08:59 |
| raitobezarius | pkgsCross is just convenience for | 12:09:08 |
| raitobezarius | nix-build -A something --argstr crossSystem "a system config" | 12:09:35 |
| raitobezarius | It is an arbitrary limitation of the full power of the cross compilation infra | 12:09:55 |
| raitobezarius | And is useful to refer to cross compiled package internally to nixpkgs via the available examples | 12:10:16 |
 ash (it/its) 🏳️⚧️ | alright i'm back on the case | 17:02:06 |
| ash (it/its) 🏳️⚧️ | getting firefox to build for ppc64le-musl | 17:02:19 |
| ash (it/its) 🏳️⚧️ | with obnoxiously cryptic errors | 17:02:36 |
| ash (it/its) 🏳️⚧️ | does anyone want to try their hand? i pushed my nixpkgs changes to github: https://github.com/ashkitten/nixpkgs/tree/fix-firefox-ppc64le-musl | 17:06:56 |
| ash (it/its) 🏳️⚧️ | nix build github:ashkitten/nixpkgs/fix-firefox-ppc64le-musl#pkgsCross.musl-power.firefox | 17:07:38 |
| ash (it/its) 🏳️⚧️ | here's my build log: https://gist.github.com/ashkitten/5bfa92664c80e5e4082eca94f5dbfdfd | 17:08:19 |
| ash (it/its) 🏳️⚧️ | it truly beats me why exit status 0 is an error here... | 17:10:52 |
| ash (it/its) 🏳️⚧️ | okay.. got past that. now i'm having different errors! | 18:22:59 |
| ash (it/its) 🏳️⚧️ | https://gist.github.com/ashkitten/a8f46190ae9c734bc7b0fe1a4581700f | 18:23:21 |
| ash (it/its) 🏳️⚧️ | guess it's trying to run the wrong as but i'm not sure where it's getting that from | 18:23:41 |
| ash (it/its) 🏳️⚧️ | interesting... pkgsCross.musl-power.buildPackages.rustc.llvmPackages.bintools doesn't contain as | 18:35:38 |
| ash (it/its) 🏳️⚧️ | llvmPackages.bintools doesn't at all, actually | 18:36:03 |
| ash (it/its) 🏳️⚧️ | i'll try building without lto.. see if that at least gets it to build, then work on lto | 18:40:11 |
| ash (it/its) 🏳️⚧️ | the issue is it's passing -fno-integrated-as to clang, but there is no as in the closure if cc is overridden with a bintools that doesn't contain as i think | 18:41:36 |
| ash (it/its) 🏳️⚧️ | ayyyyy it built!! | 19:08:03 |
| ash (it/its) 🏳️⚧️ | so now i need to make sure the wrapper builds, and see if i can add as back to stdenv without messing everything up | 19:10:57 |
| ash (it/its) 🏳️⚧️ | got wrapper building!! | 21:23:12 |
| ash (it/its) 🏳️⚧️ | and it works! ..now building again to fix sqlite endianness issue | 21:23:39 |
 rhelmot | is clang known to be problematic as a basic stdenv compiler | 21:35:45 |
 p14 | In reply to @rhelmot:matrix.org
is clang known to be problematic as a basic stdenv compiler There are issues for example https://discourse.llvm.org/t/configure-script-breakage-with-the-new-werror-implicit-function-declaration
They can be worked around with effort but lots of stuff doesn't currently work out of the box with a recent LLVM in nix. | 21:37:26 |
| p14 | It's slightly unfortunate breakage particularly because it confuses a lot of configure scripts into awkwardly silently (or nonobviously at least) broken states. | 21:38:21 |
