| 23 Sep 2025 |
 hexa | cc aloisw | 16:07:27 |
 aloisw | Current version has that toml11 4 patch reverted due to incompatibility with the toml11 3 in unstable (and 25.05). https://github.com/NixOS/nixpkgs/pull/444623 has a newer version of the patch that supports both toml11 versions. | 16:08:42 |
 emily | we already have the toml11 bump in -next | 16:09:52 |
| emily | is there a reason not to drop the patch and just merge it into there? | 16:10:08 |
 Grimmauld (any/all) | https://github.com/NixOS/nixpkgs/pull/445081
i am wondering, if cmake is broken, is just not building with cmake a fix we would want? I am a bit worried, because things downstream may break if they expect the cmake files to be present | 16:10:17 |
 Lun | I can hit that with a review run and see if the deps get upset if you want | 16:11:12 |
| Lun | seems likely to cause problems but it depends how dependents find it - they might already be shipping a FindX.cmake that works for this case or something | 16:11:45 |
| Grimmauld (any/all) | thats what i would have done in the next step too, but i first wanted to check in whether the approach is even worth considering | 16:11:49 |
| aloisw | 25.05 basically, if you prefer I can also prepare a drop of the patch against staging-next. | 16:11:52 |
| Lun | substituting the cmake version seems preferable to me assuming it's workable and there isn't a whole bunch of cruft that broke and needs a proper patch in it | 16:12:36 |
| Grimmauld (any/all) | there is also an upstream patch | 16:12:49 |
| Grimmauld (any/all) | imo fetching that is the way | 16:12:59 |
| Grimmauld (any/all) | Actually should be fine | 16:15:05 |
| Grimmauld (any/all) | the cmake files aren't being installed anyways lol | 16:15:12 |
| Grimmauld (any/all) | (checked against current unstable) | 16:15:26 |
| Grimmauld (any/all) | i'll diffoscope to make sure i am not missing anything, and then just merge tbh | 16:19:36 |
| Grimmauld (any/all) | Ah okay it does break | 16:20:38 |
| emily | whoops. I keep forgetting about 25.05. | 16:26:12 |
| emily | can we just backport the new toml11 version there? it's a pretty self-contained package, could copy-paste it as toml11_4 | 16:26:22 |
| emily | (but I'm ok with whatever you want to do) | 16:26:27 |
| emily | (this ensures more uniform behaviour between Nixpkgs releases though) | 16:26:34 |
| emily | I would rather not go CMake → autotools | 16:26:56 |
| emily | this seems worse than a substituteInPlace if it's not actually truly incompatible | 16:27:13 |
| aloisw | https://github.com/NixOS/nixpkgs/pull/445574 is the PR, currently building the package on x86_64-linux and aarch64-linux. | 16:27:24 |
| aloisw | Does that matter? 2.93 does not have uniform behaviour either. | 16:27:49 |
| emily | fair | 16:28:01 |
| aloisw | And there I would view a sudden behaviour change in 25.05 more critical. | 16:28:09 |
| emily | oh, I only meant backport it as a separate version, not backport the bump | 16:28:23 |
| emily | just for Git. but yeah, reasonable. | 16:28:29 |
| aloisw | Also 25.05 is only slightly more than 2 more months, I think I can manage the manual backports. | 16:28:46 |
