| 27 Jun 2026 |
 Myria | Ah, sorry, aarch64-darwin | 12:15:51 |
 hexa | right, I don't test that in python-updates | 12:19:08 |
| hexa | https://github.com/numpy/numpy/commits/main/numpy/distutils | 12:20:05 |
| hexa | gone in 2.5.0 | 12:20:23 |
| hexa | ahh great, not my bump | 12:20:37 |
| hexa | Myria if you have everything cached until numpy, can you remove the patch, test and send a pr againnst staging-next? | 12:21:29 |
| hexa | Redacted or Malformed Event | 12:21:32 |
| Myria | Sure | 12:21:36 |
| hexa | awesome | 12:21:42 |
 Sergei Zimmerman (xokdvium) | In reply to @emilazy:matrix.org
that's references being added? Might be frankenbuilds. MacOS code signing issues persist even now | 12:24:04 |
 Grimmauld (any/all) | stdenvNoCC got a gcc leaked on aarch64-linux too though | 12:26:40 |
| Myria | Might take a bit, I was testing gcc-15.3.0 at the time so I have less in cache, and only a little poor macbook air. Will be a few hours but I'll have it done | 12:29:00 |
| hexa | ohhhh | 12:29:10 |
| hexa | only if it doesn't inconvenience you too much | 12:29:39 |
| hexa | we can probably wait until hydra has built it and fix it in an upcoming eval | 12:29:52 |
| Myria | Nah, happy to do it. Don't have anything else | 12:31:13 |
| hexa | alright | 12:31:28 |
| hexa | we're so back | 12:55:49 |
| hexa | https://grafana.nixos.org/d/MJw9PcAiz/hydra-jobs?orgId=1&from=now-30m&to=now&timezone=utc&var-machine=$__all&var-job=hydra_queue_runner | 12:56:07 |
 emily | hard to see how that would break disallowedRequisites | 13:04:40 |
| emily | oh | 13:04:46 |
| emily | unless it's sedding away a path? lol | 13:04:54 |
| emily | can we just disable fallback paths pls at least | 13:05:14 |
| emily | do I interpret correctly that Darwin is now performing on par with the Linux platforms? | 13:07:23 |
| emily | seems about true looking at the job counts on the Hydra jobset | 13:07:56 |
| hexa | it has fewwer build slots | 13:07:59 |
| hexa | Redacted or Malformed Event | 13:08:03 |
| emily | yeah, but # of completed builds / queued builds seems comparable | 13:08:17 |
| emily | the builds themselves are likely faster because of single core perf | 13:08:37 |
| hexa | yeah, we could render steps completed per platform I would hope | 13:08:51 |
