| 12 Jul 2025 |
 Sarah Clark | At least, that's where git-bisect took me when testing gpytorch. | 21:40:54 |
 Randy Eckenrode | Probably the line where it adds llvmPackages.libomp. | 21:41:14 |
| Sarah Clark | I need to bisect a few more packages. | 21:41:15 |
| Randy Eckenrode | That’s new. | 21:41:22 |
| Sarah Clark | Frack | 21:41:33 |
| 13 Jul 2025 |
| Randy Eckenrode | mypy fails to build after the Lix change to build on the same volume as the store when the store is case sensitive | 03:30:28 |
| Randy Eckenrode | https://github.com/python/mypy/blob/db6788868b07fd13bcc8606a165f2ca90a68c76c/mypy/test/testfscache.py#L78-L89 | 03:30:30 |
| Randy Eckenrode | They have a test that fails when Windows or Darwin have a case-sensitive system. | 03:31:09 |
 Katalin 🔪 | feels like that should be reported to them since windows and darwin can absolutely have case-sensitive FSes :V | 03:44:49 |
| Katalin 🔪 | especially darwin | 03:44:52 |
| Katalin 🔪 | (it’s a bit harder to do on windows, right?) | 03:45:37 |
 emily | yeah correct way is to sniff with a test file before checking | 10:19:43 |
 symphorien | what is the current vresion of clang on darwin ? | 12:18:28 |
| Randy Eckenrode | Clang 19. | 12:19:20 |
| symphorien | thank you | 12:19:25 |
| Randy Eckenrode | It’s expected to be updated to Clang 20 for 25.11, but Darwin will be switching to the system libc++ for that release as well. | 12:20:46 |
 ethancedwards8 | In reply to @reckenrode:matrix.org
It’s expected to be updated to Clang 20 for 25.11, but Darwin will be switching to the system libc++ for that release as well. Can you elaborate/provide a link for the second half of this statement? | 13:19:49 |
|  Max set a profile picture. | 16:04:56 |
| Max changed their display name from cmltawt0 to Max. | 16:05:31 |
| Max changed their profile picture. | 16:06:10 |
| Max changed their profile picture. | 16:06:25 |
 prince213 | I think it's https://github.com/NixOS/nixpkgs/pull/398727 | 18:59:17 |
| ethancedwards8 | Interesting. Am I understanding this correctly in thinking that instead of building packages with the cxx library from Nixpkgs it will use the native Darwin library?
This seems slightly impure?
| 19:45:10 |
| emily | Darwin system libraries are already late-bound in general | 19:47:38 |
| emily | you can't build your own libSystem | 19:47:44 |
| emily | for the C library, every system framework, etc. we link against headers and (non-code-containing) stubs from the SDK and the corresponding library is loaded from the OS's dynamic linker cache at runtime | 19:48:19 |
| emily | we are just moving to using that libc++ instead of one we build ourselves | 19:48:49 |
| emily | (we do build our own toolchain etc. from scratch, we don't use Xcode) | 19:48:56 |
| ethancedwards8 | Oh, I did not realize that. | 19:50:06 |
| Randy Eckenrode | It’s an ODR violation to use multiple libc++ implementations with the same ABI. | 19:50:07 |
