| 2 Oct 2025 |
|  Guilherme Espada joined the room. | 18:19:30 |
| 3 Oct 2025 |
|  -> @oli:catgirl.cloud changed their display name from olivia floof company edition [she/it] 🏳️⚧️ to -> @oli:catgirl.cloud. | 09:33:58 |
|  ulucs changed their display name from wtvb to ulucs. | 15:05:21 |
 delroth | linked to the recent gerrit outage / mild data loss: we're going to migrate gerrit.lix.systems to a new host on Tuesday 2025-10-07. Maintenance window is currently from 17:00 to 20:00 CEST, expect downtime during this period (hopefully we won't need the whole 3h). The host will also be relocated from the US to Western Europe. IP addresses will change but I expect to be able to keep the SSH key, so this should be fairly transparent. We'll re-announce a bit before the maintenance window, at the beginning of the maintenance window, and when the migration ends. | 21:56:04 |
| 5 Oct 2025 |
 piegames | https://buildkite.com/lix-project/lix/builds/4982/steps/canvas?jid=0199b4f9-45b7-4be9-bf12-d510cc946935 pipeline job running for 3.5h, dis normal? | 18:52:00 |
| piegames | Grr, Darwin CI seems to be down again | 18:53:10 |
 raitobezarius | I don't think it's down | 18:55:20 |
| raitobezarius | We have been just hammering it | 18:55:23 |
| raitobezarius | 
Download image.png | 18:56:02 |
| raitobezarius | We just don't have enough Darwin capacity on our peaks contribution periods | 18:56:18 |
| raitobezarius | I'm planning to connect something like 4 more jobs or slightly more | 18:56:26 |
| raitobezarius | But it won't give us 3.5hr/3 peaks | 18:56:47 |
| raitobezarius | If we had a way to make certain Darwin tests which are extremely slow faster, we could cut | 18:57:02 |
| piegames | can we make Darwin tests non-blocking for merge maybe? :) | 19:34:52 |
| piegames | how many Darwin-specific tests do we have? | 19:35:24 |
| raitobezarius | unit tests mostly? | 19:36:16 |
| raitobezarius | we don't have any integration tests for darwin | 19:36:27 |
| raitobezarius | not a big fan of not making them blocker for merge | 19:36:38 |
| raitobezarius | but happy to drop for example nej | 19:36:44 |
| raitobezarius | * but happy to drop for example nej tests | 19:36:46 |
| raitobezarius | which are notoriously slow | 19:36:49 |
| piegames | I think we should primarily be smarter about our scheduling | 19:40:20 |
| piegames | stuff like making round-robin across chains, build first and last commit first, skip all in-between on failure of both | 19:41:19 |
| raitobezarius | besadii can be extended to support this scheduling | 19:53:30 |
| raitobezarius | it contains hooks for result reporting and starting up builds | 19:53:54 |
| raitobezarius | so it can delay to wait for a chain upload rather than firing up CI jobs immediately | 19:54:03 |
| raitobezarius | and it can cancel many builds in-between if failure occur | 19:54:13 |
| raitobezarius | though my priority queue for CI infra is:
- merge queue
- macOS capacity
- benchmarking infra
| 19:54:34 |
| raitobezarius | I cannot touch besadii for October | 19:54:47 |
| piegames | I agree on the first two, can besadii be put over benchmarking infra? | 19:58:15 |
