| 6 Dec 2024 |
 olivia | In reply to@piegames:flausch.social
olivia: I'd hope and expect that you can just do it in git and Gerrit follows? Because that's kinda the point of referential identities for commits in the first place IMO I tried this expecting it to work and it created new CLs :( | 21:23:12 |
 jade_ | In reply to @benjamin:computer.surgery
I tried this expecting it to work and it created new CLs :( that is expected | 21:23:21 |
| jade_ | because it is how cross-branch cherry picks work | 21:23:35 |
| jade_ | one change-id can have multiple CLs | 21:23:42 |
| olivia | oh, that makes sense | 21:23:44 |
| olivia | I think I'll just create new ones and abandon the old | 21:23:55 |
| olivia | thanks for the advice! | 21:23:58 |
| jade_ | for backports you cherry pick without changing change-id, which links the CLs in gerrit | 21:24:12 |
| jade_ | it's quite nice | 21:24:13 |
| jade_ | In reply to @k900:0upti.me
Could it be that substitute-on-destination is broken maaaaybe. there was some recent settings bustage related to the evaluator refactor | 21:24:58 |
| jade_ | it might be worth trying to get a reproducible test case | 21:25:21 |
| jade_ | the thing we could also do if we have too much trouble getting things feeling stable is do an RC, but mostly i want to get enough coordinated beta bake time to believe it is okay | 21:29:08 |
| jade_ | and that is hard when n-e-j breaks (since that stops people updating) and if there are regressions (same) | 21:29:31 |
 K900 | In reply to @jade_:matrix.org
it might be worth trying to get a reproducible test case I can try tomorrow | 21:29:33 |
| jade_ | shit is allowed to break, but we just have to make sure it stays fixed long enough to believe we can release on a given day and i have not been feeling that in the past couple weeks | 21:30:18 |
| jade_ | (i reallllly wanna do n-e-j in-tree) | 21:30:36 |
| jade_ | oh another 2.92 blocker is fixing the x86_64-darwin build | 21:44:11 |
 raitobezarius | Asyncification is imminent as well and I imagine we will need a lot of testing efforts to stabilize this | 22:18:51 |
| raitobezarius | I wished we would have proper cgroups for 2.92, but it's stuck on an architectural problem, namely that remote stores (that doesn't do anything on your local machine) can... End up build things on your machine locally | 22:19:30 |
| raitobezarius | So that's deferred until we have proper separation between what can build, substitute, etc. | 22:19:50 |
| raitobezarius | As for the Nix custom subcommands experiment, I think it's mostly ready and requires only polish, but should be possible to use ; I wonder if we should include it in 2.92 under XP feature so that we can start having adoption | 22:20:28 |
| raitobezarius | Another big priority for me is identity aware sandbox, but probably 2.93 | 22:21:09 |
|  caass joined the room. | 22:54:51 |
| 7 Dec 2024 |
 piegames | I feel like Lix development is currently heavily bottlenecked on brains | 21:22:40 |
 KFears (annoying) | How so? Is this something that can be helped? | 21:53:11 |
