| 17 Dec 2024 |
 Tristan Ross | I could do that in the PR which is the next step: making linker work as expected | 17:52:07 |
 Randy Eckenrode | I was looking at doing that for the LLVM bootstrap, but adding Darwin checks everywhere was ugly. It also seemingly can’t use lld. | 17:52:44 |
| Randy Eckenrode | In reply to @rosscomputerguy:matrix.org
I could do that in the PR which is the next step: making linker work as expected I’m not sure the LLVM changes would be right in that case. Is linker meant to be independent of bintools? | 17:54:15 |
| Tristan Ross | Gotcha, yeah I'm not entirely sure how LLVM bootstrapping will be done yet | 17:54:26 |
| Tristan Ross | In reply to @reckenrode:matrix.org
I’m not sure the LLVM changes would be right in that case. Is linker meant to be independent of bintools? Yes | 17:54:32 |
| Randy Eckenrode | I think right now the useLLVM case assumes LLD. Darwin would use LLVM bintools and ld64 (for now). | 17:55:14 |
| Randy Eckenrode | Darwin also needs opted out of libunwind. | 17:56:23 |
| Tristan Ross | Alright, I'll get that done when I work on the linker stuff. We might get a world rebuild from it. | 17:56:35 |
| Randy Eckenrode | That’s why I wanted to do Darwin separately. It’s going to involve cleanup. | 17:56:53 |
| Tristan Ross | Gotcha, do you want it completely separate from the linker PR or just separate from this first one? | 17:57:24 |
| Randy Eckenrode | It could possibly be part of the linker PR if that is updating the LLVM build. | 17:58:03 |
| Tristan Ross | The linker PR might be a bit big since it adjusts things like build systems or where the linker is explicitly defined. | 17:58:09 |
| Tristan Ross | In reply to @reckenrode:matrix.org
It could possibly be part of the linker PR if that is updating the LLVM build. Wdym by updating the LLVM build? | 17:58:31 |
| Randy Eckenrode | Hold on, misread the PR. It’s checking toolchain not bintools. | 17:59:14 |
| Randy Eckenrode | Darwin would still have an apple toolchain. | 17:59:30 |
| Randy Eckenrode | In reply to @rosscomputerguy:matrix.org
Wdym by updating the LLVM build? I was thinking it needed more granularity like if a platform used LLVM bintools, it would default to building them in the LLVM build like it does currently when useLLVM. | 18:00:30 |
| Tristan Ross | Oh gotcha | 18:01:21 |
| Randy Eckenrode | That seems like something to do in a follow up. | 18:01:45 |
| Tristan Ross | Yeah | 18:01:56 |
| Randy Eckenrode | If you can get away with setting bintools to llvm on Darwin now, I’d say go for it. Otherwise, if it requires cleanup, I say do it separately. | 18:02:28 |
| Tristan Ross | I'm not sure if it'll require a cleanup | 18:02:50 |
| Tristan Ross | I don't have time to test atm and I don't do much on Darwin. | 18:03:22 |
| Randy Eckenrode | In reply to @rosscomputerguy:matrix.org
I'm not sure if it'll require a cleanup Not sure either. If nothing is being changes to check bintools, then probably not. | 18:04:51 |
| Tristan Ross | Stuff does check it | 18:05:09 |
| Tristan Ross | The bintools derivation in nixpkgs itself | 18:05:21 |
| Tristan Ross | It's an alias which changes based on linker and bintools in the platform | 18:05:40 |
| Tristan Ross | Plan is to make that only change on bintools in the platform | 18:06:04 |
| Randy Eckenrode | I don’t see that in the PR? | 18:07:19 |
| Randy Eckenrode | Oh, there it is in all-packages.nix. | 18:07:49 |
| Tristan Ross | https://github.com/NixOS/nixpkgs/pull/365057/files#diff-ab5748dc9567516fefba8344056b51ec1866adeace380f46e58a7af3d619ea22R7903 | 18:08:04 |
