| 16 Jun 2026 |
 Vladimír Čunát | Still, the list of nixpkgs.* jobs could certainly be expanded. That should be fairly cheap. | 05:09:51 |
| Vladimír Čunát | * Still, the list of nixpkgs.* jobs among -small blockers could certainly be expanded. That should be fairly cheap. | 05:10:07 |
 linj | so just listing them in nixpkgs also affects hydra somehow? might be worth a comment for people like me who are not familiar with hydra | 05:11:44 |
| Vladimír Čunát | They get a job. | 05:12:34 |
| Vladimír Čunát | i.e. you can see them in https://hydra.nixos.org/eval/1826306 | 05:12:51 |
| Vladimír Čunát | So nixos-unstable-small waits for them to finish, but not necessarily to succeed. | 05:13:30 |
| Vladimír Čunát | Overall the channel blocker lists seem undermaintained. | 05:14:08 |
| Vladimír Čunát | It would be nice if we had some good data on usage and chose based on that. | 05:15:12 |
| linj | Another quick question: why there is no aarhc64-linux packages in the nixpkgs-unstable channel blocker but there are in the nixos-unstable channel blocker? | 05:41:14 |
| linj | * Another quick question: why are there no aarch64-linux packages in the nixpkgs-unstable channel blocker? There are aarch64-linux packages in the nixos-unstable channel blocker. | 05:42:00 |
| Vladimír Čunát | 🤔 that seems weird to me. | 05:45:42 |
| Vladimír Čunát | But you're right: https://hydra.nixos.org/job/nixpkgs/unstable/unstable#tabs-constituents | 05:45:57 |
| linj | I find emacs has a weird relationship with channel blockers: emacs.x86_64-linux and emacs.aarch64-linux are blockers for nixos-unstable, but not for nixpkgs-unstable. I think it may be a good idea to make emacs a blocker for nixpkgs-unstable and then it will automatically become a blocker for nixos-unstable. But I end up finding that no aarch64-linux packages are blockers for nixpkgs-unstable... | 05:46:49 |
| linj | This is the PR adding aarch64-linux packages as nixos-unstable blocker: https://github.com/NixOS/nixpkgs/pull/193600 | 05:47:18 |
| Vladimír Čunát |
then it will automatically become a blocker for nixos-unstable
This property is also very new.
| 05:47:33 |
| Vladimír Čunát | They were pretty random and independent. | 05:47:42 |
| Vladimír Čunát | As I wrote, this has been undermaintained. | 05:48:03 |
| Vladimír Čunát | It would be nice to organize it all better. | 05:48:13 |
| Vladimír Čunát | I'm certainly all for having aarch64-linux in nixpkgs-unstable. | 05:49:46 |
| Vladimír Čunát | * I'm certainly all for having aarch64-linux in nixpkgs-unstable blockers. | 05:49:50 |
| Vladimír Čunát | I think it just got forgotten. | 05:50:00 |
| Vladimír Čunát | On a higher level, I suppose it can be confusing that -unstable channels are split differently than stable channels, too. | 05:51:39 |
| linj | absolutely! there is missing nixpkgs-26.05 and nixpkgs-unstable-darwin | 05:57:32 |
| linj | 
Download image.png | 05:57:36 |
| Vladimír Čunát | I think the move of the nixpkgs blockers into nixpkgs only is pretty non-controversial. Also the enablement of aarch64-linux in there. | 06:01:04 |
| Vladimír Čunát | linj: feel free to create a pull request for those, I'd say. | 06:01:26 |
| linj |
move of the nixpkgs blockers into nixpkgs only
you mean nixos blockers into nixpkgs only, right?
| 06:03:40 |
| Vladimír Čunát | Adding channels is a more careful thing, as it's user-facing and going back would be quite a breaking change. The current Hydra doesn't handle duplicate jobs too well either. Maybe that will improve soon with the new queue runner, though. | 06:03:40 |
| Vladimír Čunát | Yes, I meant that the nixos blockers don't need the individual nixpkgs.* entries but only the whole nixpkgs.unstable. | 06:04:27 |
| Vladimír Čunát | (which requires filling nixpkgs.unstable with the extra jobs missing in there, I think) | 06:04:52 |
