| 19 Sep 2025 |
 emily | updated the hook PR with a missing Darwin fix and some tweaks | 17:12:21 |
| emily | going to try re-bootstrapping staging with it because apparently Perl might be broken | 17:12:31 |
| emily | https://github.com/NixOS/nixpkgs/pull/444348 ? | 17:20:15 |
| emily | cc K900 Alyssa Ross | 17:20:18 |
 Alyssa Ross | already discussed | 17:20:31 |
| Alyssa Ross | i'm building | 17:20:44 |
 K900 | On Darwin? | 17:20:43 |
| Alyssa Ross | referring to the PR link, not anything Darwin | 17:21:40 |
 connor (burnt/out) (UTC-8) | I do, thanks for the reminder | 17:22:45 |
| K900 | No I mean Perl being broken on Darwin | 17:24:14 |
| emily | yes, checking | 17:30:15 |
| K900 | What I mean is | 17:30:57 |
| K900 | Should I try Linux | 17:31:00 |
| emily | probably not since it looked like LLVM 21 | 17:31:08 |
| emily | also because people actually test Linux | 17:31:18 |
| emily | nim65s: fwiw trlib, hpipm, openscenegraph, corto, bullet, libccd, cxxopts, pystring, python3Packages.qdidl, thrift, qgv, python312Packages.pymeshlaball look good to PR to me.glewneeds to usefetchpatchinstead offetchurl`. other things I adjusted in my branch or look like they might need some more polish, I forget the exact list :) | 17:43:40 |
| emily | * nim65s: fwiw trlib, hpipm, openscenegraph, corto, bullet, libccd, cxxopts, pystring, python3Packages.qdidl, thrift, qgv, python312Packages.pymeshlab all look good to PR to me. glew needs to use fetchpatch instead of fetchurl. other things I adjusted in my branch or look like they might need some more polish, I forget the exact list :) | 17:43:57 |
| connor (burnt/out) (UTC-8) | emily: is there an easy way to test that PR against master, or does it rely on staging? | 17:50:05 |
| emily | it may be possible to cherry-pick the CMake bump and the hook changes, but I wouldn't count on it, since there are fixes of other packages that went to staging | 17:51:42 |
| emily | no idea what's in CUDA's dependency closure | 17:51:42 |
| connor (burnt/out) (UTC-8) | angry ghosts and bad CMake, mostly :l | 17:53:36 |
 Vladimír Čunát | We have binaries for linux stdenvs for the latest staging in the cache. | 17:56:46 |
| Vladimír Čunát | So if you'd change cmake, you may as well use staging directly. | 17:57:03 |
| Vladimír Čunát | (I don't think there will be much of a difference in the rebuild amount.) | 18:00:00 |
 nim65s | In reply to @emilazy:matrix.org
nim65s: fwiw trlib, hpipm, openscenegraph, corto, bullet, libccd, cxxopts, pystring, python3Packages.qdidl, thrift, qgv, python312Packages.pymeshlab all look good to PR to me. glew needs to use fetchpatch instead of fetchurl. other things I adjusted in my branch or look like they might need some more polish, I forget the exact list :) Ok thanks :)
For corto I have something else almost ready, but the others I can open a PR.
On master ? One PR with one commit per package ? Or a PR per package ? | 18:26:09 |
| emily | I'm okay with one big PR if it's more convenient | 18:31:18 |
| emily | master should be fine rebuilds permitting | 18:31:30 |
| emily | Fabián Heredia: are you sure about not doing GCC 15 for 25.11? I plan to start running builds with it while we do the current -next cycle | 18:42:07 |
| emily | and it seems like there is a lot of activity to fix packages already before it even hits staging | 18:42:30 |
| emily | I would prefer to get it in if we can, it doesn't immediately seem worse than GCC 14 was, or LLVM 16→19 | 18:43:17 |
