| 23 Jan 2025 |
 hexa | two cases come to mind: multiple jobs building into staging and changes to the same PR, but the builds land on another builder | 20:28:30 |
 emily | x86_64-darwin build failures should track aarch64-darwin much more reliably on unstable these days | 20:44:40 |
| emily | but not on 24.11 | 20:44:42 |
| emily | aren't there still actual physical Intel Macs for ofborg though? | 20:45:06 |
 @janne.hess:helsinki-systems.de | there are (still) | 20:51:49 |
| emily | seems like a waste not to use them :) | 20:57:16 |
| emily | (though I guess electricity cost is a thing) | 20:57:26 |
 adamcstephens | just ship them somewhere with cheap electricity costs :) | 20:58:34 |
| hexa | they are provided by macstadium | 20:59:53 |
| emily | oh, I thought it was the physical machines | 21:00:10 |
 pbsds | there seems to be an eval oopsie in https://github.com/NixOS/nixpkgs/pull/376117, anyone here familiar with the error? | 21:54:53 |
| pbsds | (also, if there is a new matrix space for the gha actions please let me know) | 21:55:28 |
 Winter | just use #dev, not here | 21:57:24 |
 leona | In reply to @pederbs:pvv.ntnu.no
there seems to be an eval oopsie in https://github.com/NixOS/nixpkgs/pull/376117, anyone here familiar with the error? This was a broken merge that was already reverted | 21:59:45 |
| leona | In reply to @pederbs:pvv.ntnu.no
there seems to be an eval oopsie in https://github.com/NixOS/nixpkgs/pull/376117, anyone here familiar with the error? * | 21:59:53 |
| pbsds | ah, i had to also re-run the resolve-merge-commit job, thanks! | 22:03:39 |
| 24 Jan 2025 |
|  SomeoneSerge (back on matrix) joined the room. | 11:31:26 |
| adamcstephens | this kanidm job timed out on aarch64-linux. Is this just a capacity issue, or should I expect this to get better in the future? https://logs.ofborg.org/?key=nixos/nixpkgs.376286&attempt_id=e8166195-320a-4fb7-bde6-7af6a6daa646 | 15:27:37 |
| adamcstephens | (I know this is a big ask of a job, as it builds two copies of kanidm + nixos tests for both) | 15:28:13 |
| hexa | and a day later we have 7k remaining 🤡 | 17:06:09 |
| adamcstephens | so, no, we're not keeping up them :) | 18:27:23 |
| emily | fwiw Domen's Nix <3 macOS Open Collective thing is sitting on a bunch of capital and continues to get ~$1k/month that isn't really going anywhere right now | 19:24:23 |
| emily | I was planning on reaching out to see if we could get some ofborg builders funded with that before the ofborgalypse happened, might still be worth doing so | 19:24:40 |
| emily | since after the SDK rework CI backlog certainly became #1 macOS platform support issue | 19:25:03 |
| emily | (that said, is it definitely not timing out on LLVMs again? :) ) | 19:38:55 |
| adamcstephens | * so, no, we're not keeping up then :) | 20:00:58 |
| adamcstephens | So is the question: At what point do we throw more hardware at the problem to see if it goes away? :) | 20:04:43 |
| emily | I think that's a question at least | 20:11:09 |
| emily | if there's spare money earmarked for making Darwin support better that isn't getting used on anything and slow CI is the #1 Darwin pain point then it seems like a good thing to do | 20:11:53 |
| emily | but also getting more Macs just because they're constantly failing to build LLVM and would be totally fine keeping up if they got their timeout bumped by 30 minutes or something is a bit silly | 20:12:13 |
