| 15 Nov 2024 |
 p14 | I could believe that. | 09:10:36 |
 emily | I do think we should find a way of solving this libc++ issue that isn't -nostdlibinc, but if we move it to the wrapper the pressure is reduced | 09:11:05 |
| emily | and we get a fully target-independent Clang :) | 09:11:21 |
| p14 | Yeah, and it's easier to develop since it won't require rebuilding clang. | 09:11:24 |
| p14 | (When someone does get around to it) | 09:11:34 |
| emily | well, wrapper changes aren't much kinder 😅 | 09:11:37 |
| emily | though you can at least override the wrappers for specific derivations | 09:11:43 |
| p14 | Yup | 09:11:48 |
| emily | you Linux people don't know how good you have it – changing LLVM doesn't even restart the entire bootstrap | 09:11:58 |
| p14 | Muahah | 09:12:04 |
| p14 | Yeah, definitely glad I don't have to deal with that. | 09:12:29 |
| emily | instead you have to deal with GCC. | 09:12:50 |
| p14 | Well there is that. | 09:12:56 |
| p14 | I've got a moment to mess with this and send a patch to drop the patch unless you want the fun. Will let you know if I have to give up for some reason. | 09:14:08 |
| p14 | I have a handful of trivial patches outstanding which would be nice to not let languish until I forget about them: https://github.com/NixOS/nixpkgs/pulls/pwaller | 09:15:07 |
| emily | would be happy to review :) | 09:15:44 |
| p14 | The failures mentioned in https://github.com/NixOS/nixpkgs/pull/354471 raise my eyebrows a bit, not sure what's going on there. | 09:15:45 |
| emily | I've been having too much LLVM fun myself already lately: https://github.com/NixOS/nixpkgs/pull/354107 | 09:16:13 |
| emily | should be landing in staging soon | 09:16:19 |
| emily | plan is to update LLVM for Darwin and other platforms in lockstep in future | 09:16:59 |
| emily | which should be very good for LLVM in general, because we're the only ones building most of the package set with it | 09:17:12 |
| p14 | Very cool to have gotten things into this state. | 09:17:13 |
| p14 | It (the state of nixpkgs generally and the llvm side specifically) is a seriously impressive thing. | 09:17:55 |
| p14 | Would be nice if more stuff built happily out of the box with clang. | 09:18:03 |
| emily | the more we take an upstream-first approach the more the ecosystem will move in that direction :) | 09:18:24 |
| emily | I'd rather Linux was using LLVM personally! | 09:18:39 |
 @paparodeo:matrix.org | In reply to @emilazy:matrix.org
I've been having too much LLVM fun myself already lately: https://github.com/NixOS/nixpkgs/pull/354107 eh I broke eval on that one -- will fix. | 09:19:39 |
| emily | In reply to @p14:matrix.org
The failures mentioned in https://github.com/NixOS/nixpkgs/pull/354471 raise my eyebrows a bit, not sure what's going on there. should maybe just send this to staging so you can do the --replace-fail thing? | 09:19:46 |
| p14 | I mean, I could also do replace-fail in a separate patch. | 09:20:04 |
| emily | sure :) | 09:20:09 |
