| 21 Dec 2025 |
 whispers (it/fae) | * out of curiosity, if you hand it a big build graph, how does lix/nix choose what to build first? it doesn't seem to start from the most depended-on derivations or any other metric which i would expect. | 22:59:31 |
| 22 Dec 2025 |
 emily | I don't think it tries to do anything smart at all | 00:06:03 |
| emily | just picks stuff it can build and builds it until it's out of slots | 00:06:22 |
| emily | certainly Hydra isn't smart about it at all | 00:06:44 |
| emily | in practice things are usually either sufficiently wide fan-out that it doesn't really matter what you pick, or bottlenecked on few enough things that it doesn't really matter what you pick, so it's not a huge deal unless you have few build slots | 00:07:41 |
|  KFears (they/them) changed their display name from KFears (burnt out) to KFears (they/them). | 00:08:50 |
| whispers (it/fae) | huh fair, ty | 00:10:40 |
 raitobezarius | (untrusted) spare hardware still needs tending so it's up to the amount of infra resources available, we already have remote beefy builders for doing things, x86 CI is not bottlenecked in any way, Darwin is always the limiting factor | 04:23:52 |
| raitobezarius | i'd recommend joining the afnix zulip and the infra channel and looking if you can pick up some tasks so that the team and you can take the time to know each other | 04:24:47 |
| raitobezarius | many tasks doesn't require root@ perms as deployment is rarely a bottleneck | 04:25:04 |
 Sofie 🏳️⚧️ (she/her) | Is GitHub down? | 09:53:09 |
| Sofie 🏳️⚧️ (she/her) | oh, not anymore | 09:53:37 |
| Sofie 🏳️⚧️ (she/her) | for a second I got 504s | 09:53:42 |
