| 29 May 2021 |
 maralorn | Las: On nixos-unstable? | 16:13:25 |
 Las | Yeah | 16:13:31 |
| Las | It's from the 7th of May | 16:14:02 |
| Las | maralorn | 16:19:17 |
| maralorn | I see. We have quite recently merged a fix for aarch64/llvm and the ghc withPackages wrapper. But I have no clue it applies to your situation. sterni (he/him) might know more. | 16:29:31 |
| Las | Thanks, I'll try updating my nixpkgs! | 16:31:59 |
 sterni (he/him) | Las: macOS? | 16:32:27 |
| Las | No, Linux of course | 16:32:35 |
| Las | I think I'm one of the very few people using an aarch64 machine as my main machine | 16:33:00 |
| Las | * I think I'm one of the very few people using an aarch64-linux machine as my main machine | 16:33:04 |
| sterni (he/him) | depending on how your environment is set up, ghc should know how to find the LLVM stuff by default now | 16:33:59 |
| sterni (he/him) | not sure if we have fixed it everywhere | 16:34:16 |
| sterni (he/him) | Las: try updating nixos-unstable and if it still doesn't work, please open an issue on github, describe how you set up your environment and ping @NixOS/haskell. I'll try to fix this, I can't really test stuff on aarch64 though so there's always a bit of guesswork involved | 16:36:00 |
| Las | Thanks. | 16:36:11 |
 pennae | maybe we should try to reinstall our pinebook and give aarch64 haskell another try
last we tried it was a bit of a massive pain in the tail | 16:37:51 |
| sterni (he/him) | there should be quite substantial binary cache by now, I've fixed a bunch of failures related to a GHC bug | 16:38:50 |
| sterni (he/him) | the only trouble is development environments because LLVM is not always added correctly, I think | 16:39:10 |
| sterni (he/him) | it's a bit annoying that GHC expects LLVM to be in PATH | 16:39:27 |
| pennae | last we tried was almost a year ago now, and on gentoo | 16:39:40 |
| pennae | ghc just routinely failed to build at all there | 16:39:55 |
| sterni (he/him) | everything is working currently in terms of ghc except ghcHEAD | 16:41:54 |
| sterni (he/him) | which may work but the build keeps timing out for an unknown reason | 16:42:04 |
| sterni (he/him) | if someone wants to debug it I'd be very happy | 16:42:15 |
| sterni (he/him) | I think it may be because the machine hydra uses has too many cores, but who knows | 16:42:38 |
| pennae | might give it a try when(/if) we manage to put nixos on the thing | 16:48:02 |
| Las | In reply to @pennae:matrix.eno.space
maybe we should try to reinstall our pinebook and give aarch64 haskell another try
last we tried it was a bit of a massive pain in the tail FWIW I think you'll have a very hard time compiling anything until the new native code gen backend is released (GHC 9.2.1 IIRC) | 16:51:51 |
| Las | LLVM is quite slow | 16:52:09 |
| sterni (he/him) | it works at least :p | 16:52:53 |
| sterni (he/him) | In reply to @pennae:matrix.eno.space
might give it a try when(/if) we manage to put nixos on the thing yeah it's not really urgent there don't seem to many ghcHEAD users, but it'd be nice to sort these issues out until the next GHC release comes out… | 16:53:37 |
| pennae | ghc 8.6.5 took 8 hours to build last we tried, ghc 8.8 always failed with linker errors | 16:53:47 |
