8 Oct 2024 |
Randy Eckenrode | * Given the Darwin stuff can’t go to master, what the policy is for master doesn’t seem pertinent. My only concern is getting the Darwin stuff into 24.11. | 14:14:46 |
Lily Foster | looke like @hexa first started to change it in 22.11 https://github.com/NixOS/nixpkgs/issues/193585 but it was not included in the post-release fixup https://github.com/NixOS/release-wiki/pull/60 | 14:15:39 |
emily | In reply to @reckenrode:matrix.org Given the Darwin stuff can’t go to master, what the policy is for master doesn’t seem pertinent. My only concern is getting the Darwin stuff into 24.11. I cared about master for my own stuff | 14:43:56 |
emily | which is why the schedule confused me | 14:44:18 |
Tristan Ross | In reply to @reckenrode:matrix.org Given the Darwin stuff can’t go to master, what the policy is for master doesn’t seem pertinent. My only concern is getting the Darwin stuff into 24.11. How far do you think we are from being able to merge it? | 17:59:47 |
emily | a few days | 18:05:40 |
emily | it's just pending review from two other people who have started going through it already | 18:05:58 |
emily | the follow-up PRs are mostly ready | 18:06:09 |
emily | it'll definitely make the next staging-next cycle, which at this rate might not even start before the 11th | 18:06:57 |
Tristan Ross | Ok, I think we're good then to get it into 24.11 | 18:25:06 |
vcunat | I wonder about staging-next-24.05 as it's been quite some time, too. | 18:25:09 |
vcunat | * I wonder about staging-next-24.05 as it's been quite some time, too. (4 weeks) | 18:25:33 |
Tristan Ross | We can probably squeeze in 1 last staging-next for 24.05? | 18:25:38 |
vcunat | 1 last seems way too little to me. Given that we promise support until the end of the year? | 18:26:16 |
Tristan Ross | Oh | 18:26:54 |
Tristan Ross | Then we probably can get like 4 or 5 staging-nexts for 24.05? | 18:27:35 |
emily | In reply to @vcunat:matrix.org I wonder about staging-next-24.05 as it's been quite some time, too. (4 weeks) I worry that if we do 24.05 next, the Darwin rework cycle becomes the last cycle before release? | 18:33:29 |
emily | which would be pretty not good | 18:33:38 |
emily | also, Randy is going to be unavailable starting ~10 days from now, so we should get to any issues ASAP. | 18:34:06 |
vcunat | Rushing big features late. That's a tension before all NixOS stable releases. | 18:37:41 |
emily | indeed | 18:42:05 |
emily | in this case, it's not so much a feature as a vital refactor | 18:42:18 |
emily | Darwin support desperately needs it to be viable; we rushed the 12.3 SDK as an interim measure and we've already had major packages (QEMU, Qt) be unable to be updated without it | 18:42:41 |
emily | those two projects don't support 12 officially, so we could easily end up in trouble within 6 months even doing security updates | 18:43:19 |
emily | it's also going to be painful to do backports across different macOS SDK patterns, so landing it early next cycle would be really bad for 24.11 in general | 18:43:47 |
emily | I think everyone would have preferred this to be happening a month ago but you know how things go | 18:44:02 |
Randy Eckenrode | Two months to develop and land is about what I expected. I wish I had been able to start this sooner, but it took some time to get the cctools and ld64 updates landed. | 19:43:20 |
Randy Eckenrode | While I lost some time due to poor Git hygiene and cleaning up my branch, I don’t think it would have made the current cycle regardless. | 19:44:03 |
Randy Eckenrode | https://gist.github.com/reckenrode/58933e82c687d8026fe19630c0ed699f | 19:44:33 |
Randy Eckenrode | That’s the flake I used for testing, which took half a week to build. I feel pretty about what kind of breakage to expect. | 19:45:12 |