| 26 Jun 2026 |
 hexa | that looks much better | 10:08:17 |
| hexa | https://github.com/NixOS/nixpkgs/pull/535575 | 10:24:29 |
| hexa | vcunat did you restart the graphical iso just now? | 10:51:25 |
 Vladimír Čunát | I did, but it didn't help. | 10:51:43 |
| hexa | https://hydra.nixos.org/build/332754507 | 10:51:43 |
| hexa | Use of uninitialized value $key in hash element at /nix/store/jvlbqylkvks50vfiqncbyxrgas38bb5z-nixos-channel-scripts/bin/.mirror-nixos-branch-wrapped line 193.
Use of uninitialized value $productType in concatenation (.) or string at /nix/store/jvlbqylkvks50vfiqncbyxrgas38bb5z-nixos-channel-scripts/bin/.mirror-nixos-branch-wrapped line 200.
No product could be selected for nixos.iso_graphical.x86_64-linux, with type at /nix/store/jvlbqylkvks50vfiqncbyxrgas38bb5z-nixos-channel-scripts/bin/.mirror-nixos-branch-wrapped line 200.
| 10:51:56 |
| Vladimír Čunát | It's stuck, despite the result already being in cache. | 10:51:58 |
| Vladimír Čunát | The hydra job itself never finished. | 10:52:16 |
| Vladimír Čunát | (post-processing for hours) I already asked Mic directly with details from log. | 10:52:38 |
| hexa | I don't think this was caused by my change fwiw | 10:53:08 |
| hexa | but I was worried for a second there | 10:53:20 |
| Vladimír Čunát | And in the meantime I triggered another eval, as I really hate that nixos-unstable is on a 10 days old commit. | 10:53:17 |
| Vladimír Čunát | No, it is not. | 10:53:23 |
| hexa | eval seems good this time | 12:53:29 |
| Vladimír Čunát | Something else would block it, but I forced the channel update already
567a49d1913c..e73de5be04e0 nixos-unstable
| 13:10:43 |
 Mic92 | @hexa:lossy.network: would you be sad if I rewrite this perl thingie in python some day? | 14:24:00 |
|  zimward joined the room. | 14:27:54 |
| hexa | by which you mean … if Claude did the thing? 😜 | 14:37:24 |
| hexa | https://github.com/NixOS/nixos-channel-scripts/issues/59 | 14:38:00 |
| hexa | I'd like it if we had that | 14:38:04 |
| Mic92 | I did this once by hand for the nix release script back in the day but never upstreamed it. It had even a self-test with a local registry and minio. | 14:41:37 |
| hexa | so yeah, a job in nixpkgs that drops a manifest and we just upload what we just execute based on that | 14:42:27 |
| hexa | instead of piling on janky version dependent logic | 14:42:37 |
| Mic92 | * I did this once by hand for the nix release script back in the day but never upstreamed it. It had even a self-test with a local docker registry and minio. | 14:43:03 |
| hexa | Redacted or Malformed Event | 14:43:20 |
| hexa | and then, sure, it can be python for all I care | 14:43:32 |
| hexa | the code is so "complex" because we need to deal with nixpkgs changes | 14:44:09 |
| hexa | Redacted or Malformed Event | 14:44:21 |
| Mic92 | I am still in favour moving the iso image to a non-hydra CI. In nix we do that in GitHub actions with oidc Auth for aws | 14:44:32 |
| hexa | that's orthogonal, no? | 14:44:57 |
