| 5 Jun 2021 |
LnL | but you have very little available there so it could be tricky and based on the commits I assume we need a never compiler in bootstrap-tools anyway | 15:16:14 |
LnL | well maybe that's not strictly necessary, it might be possible to bootstrap the new defaults with an older compiler but it should probably be updated in both places so needs to happen either way | 15:19:33 |
toonn | LnL: Ok, so I've just force pushed the changes you asked for, I believe. | 18:37:18 |
| 6 Jun 2021 |
| smittie joined the room. | 10:46:18 |
sterni (he/him) | any one around here I could ping for testing haskell / GHC on aarch64-darwin | 11:12:11 |
sterni (he/him) | * any one around here I could ping for testing haskell / GHC on aarch64-darwin? | 11:12:16 |
sterni (he/him) | We are looking into updating to GHC 8.10.5 currently which supposedly supports aarch64-darwin via the LLVM backend. | 11:12:40 |
smittie | I'm fairly new to nix, but depending on the workload I'm up for it. Is it just installing the nix package? | 11:15:14 |
sterni (he/him) | It would already be helpful if someone could checkout the PR, build GHC and try compiling a package or two, so all variations of running nix-build in a nixpkgs checkout | 11:16:30 |
sterni (he/him) | however it would involve compiling GHC | 11:16:48 |
sterni (he/him) | I have no idea how fast the M1 macbooks are | 11:17:04 |
sterni (he/him) | but let me first see if it works on x86_64-linux | 11:17:48 |
smittie | Oh, well. I'm running on a 2014 15" intel model. Without further explanations and links (to the PR) I am probably not the best candidate in helping out here. | 11:27:48 |
smittie | * Oh, well. I'm running on a 2014 15" intel model. | 11:31:10 |
andi- | sterni (he/him): link me the PR and I'll test on an M1 | 12:16:06 |
sterni (he/him) | andi-: https://github.com/NixOS/nixpkgs/pull/125936 | 14:12:51 |
andi- | Is it normal for GCC to throw these during compilation and just carry on?
<no location info>: error:
Warning: Couldn't figure out C compiler information!
Make sure you're using GNU gcc, or clang
| 15:58:01 |
andi- | Error sounds like it should fail the build but then it says "just" warning... | 15:58:27 |
sterni (he/him) | you mean GHC right? | 16:19:33 |
andi- | Yes, only now realized that this isn't the haskell channel | 16:24:53 |
sterni (he/him) | andi-: I'll check, I am also no expert on GHC on darwin unfortunately | 16:50:18 |
LnL | toonn: I think you forgot to remove the darwin.Libsystem part, if that's gone the rebuild tags should disappear | 17:01:19 |
andi- | In reply to @sternenseemann:systemli.org andi-: I'll check, I am also no expert on GHC on darwin unfortunately So far it looks good. It is building the stage1. (And I also started working on aarch64-darwin support in GHC now that the release supports that which is also at the sage1 build) | 17:03:19 |
sterni (he/him) | right yeah, the PR didn't expand the platform list etc. | 17:04:18 |
sterni (he/him) | the whole platform stuff is gonna be a bit of extra work which I'm dreading | 17:04:33 |
sterni (he/him) | plus it kinda hinges on updating stuff in a repo only peti has access to and releasing a new package on hackage which also only peti can do which is extremely annoying | 17:05:06 |
sterni (he/him) | I'm gonna open a tracking issue for that we really need that stuff in the NixOS org | 17:05:23 |
toonn | LnL: Just pushed. | 17:26:36 |
toonn | LnL: Yep, the stdenv rebuild label's removed. | 17:39:40 |
LnL | cool, are you confident that this still is everything needed to unblock the llvm changes or you want to validate that first before we merge? | 17:47:47 |