| 7 Feb 2026 |
 raitobezarius | https://snix.dev/docs/components/build/protocol/ is relevant too | 20:09:39 |
 Lisanna | ^-^ | 20:09:39 |
| Lisanna | so you would launch your custom builder as a standalone program, and it would then register itself as a builder through a library call? rather than being launched by the Lix cli or daemon directly? | 20:18:19 |
| raitobezarius | the deployment story is not totally hammered down, but a daemon frontend could receive build daemons URIs (remote or local), once you run a lix command that makes use of build, they would be sent to that daemon via RPC | 20:19:19 |
| raitobezarius | but yeah, in that context, Lix CLI or daemon wouldn't execve() a build component | 20:19:52 |
| raitobezarius | it would be delegated | 20:19:55 |
| Lisanna | alright | 20:20:02 |
| Lisanna | I'll keep watch for progress on this front, and see about adding a mode to my program for this kind of usage | 20:21:15 |
| Lisanna | is there an issue tracking it I can follow? | 20:22:28 |
| raitobezarius | https://git.lix.systems/lix-project/lix/issues/911 | 20:23:49 |
| raitobezarius | for build-hook | 20:23:53 |
| raitobezarius | i think we may remove it only once we have hit a blocker due to it OR we have replaced it | 20:24:15 |
| raitobezarius | the full core team is usually at https://zulip.lix.systems/#narrow/channel/9-Store to discuss these topics on higher bandwidth | 20:24:57 |
| Lisanna | Like the linked issue mentions, I've considered reimplementing this as a custom store plugin, but that presents complications for this particular use-case. I think you'd have to manually retrieve the build result from the custom store after the build completes, because it won't be copied back to the local store automatically. Unless the remote's local store is on a shared filesystem, it would need to keep track of which node has which build result, which would be hard to do in-between invocations, potentially causing rebuilds or not being able to find the already built derivation. | 20:42:46 |
| raitobezarius | yeah, a store plugin is highly non trivial for what you are doing | 21:44:45 |
| raitobezarius | Changes performed | 23:14:06 |
 K900 | Someone do the dark souls image macro thing | 23:14:55 |
| K900 | I'm on my phone | 23:14:58 |
| raitobezarius | 
Download clipboard.png | 23:15:21 |
| K900 | Thank | 23:19:58 |
| 8 Feb 2026 |
|  Grimmauld (migrated to @grimmauld:m.grimmauld.de) changed their display name from Grimmauld (any/all) to Grimmauld (migrated to @grimmauld:m.grimmauld.de). | 11:04:24 |
| Grimmauld (migrated to @grimmauld:m.grimmauld.de) left the room. | 11:12:01 |
| raitobezarius | Final actions commence now | 11:35:34 |
| raitobezarius | All actions have been done; there may be an impact in CI latency at the moment due to our naive scheduling right now: https://zulip.afnix.fr/#narrow/channel/4-infra/topic/Rethinking.20Buildkite.20scheduling.20in.20multi-tenant/with/23171. Ping me if this is affecting you. | 13:05:32 |
| raitobezarius | CI is currently down due to issues | 17:54:42 |
| 9 Feb 2026 |
 Rutile (rootile) | 
Download image.png | 18:23:39 |
| Rutile (rootile) | having fun migrating a test and found this pearl while looking for why a test is commented out | 18:23:54 |
| raitobezarius | :v | 18:25:43 |
| raitobezarius | FWIW, CI is back to mostly normal since yesterday night (1am / 2am CEST), there's still some major bugs:
- Final status report doesn't happen, manual
+1 Verified has to be done
- Some namings are wrong on Buildkite
/xxx instead of cl/xxx
| 18:31:19 |
 Winter | is anyone aware of semi-recently-ish changes to how errors in binary caches, specifically S3, are reported? I noticed a severe lack of "failed to upload s3://.../nix-cache-info" in the last few months; I've always wondered why exactly those are even occurring due to the fact that I have a nix-cache-info in my cache | 19:43:58 |
