| 28 Jan 2025 |
Vladimír Čunát | The queue runner, probably. | 15:45:41 |
K900 | Detached signatures would be the biggest effect to complexity ratio probably | 15:45:54 |
K900 | But I can't even explain what that is without explaining the other half of Hydra | 15:46:17 |
Sami Liedes | Right. | 15:46:24 |
K900 | The short version is that we are currently at a stage where we have enough hardware to make the coordinator the bottleneck | 15:47:21 |
K900 | So adding more hardware will not help (much) | 15:47:29 |
K900 | And fixing the coordinator requires very invasive surgery on a 20 year old Perl/C++ codebase | 15:47:47 |
K900 | Or a from-scratch rewrite | 15:48:04 |
Sami Liedes | Is what we have traditional, inherently serialized and unscalable CI in the sense that it will separately analyze each commit, make a verdict and push, guaranteeing both that the target branch always passes tests and that it will be able to pinpoint exactly what change caused a failure? | 15:49:53 |
K900 | No | 15:50:04 |
Sami Liedes | Ok. | 15:50:15 |
K900 | It's evaluating snapshots of master on a fixed timer | 15:50:23 |