| 13 Apr 2026 |
 Mic92 | I am now monitoring: https://staging-hydra.nixos.org/eval/1 | 19:33:38 |
 John Ericson | ok cool | 19:34:05 |
| John Ericson | Mic92: do you think we could write hydra regression test for some of these? | 19:34:23 |
| John Ericson | I am surprised the fixed output paths were busted an the systems tests didn't fail, for example. that is weird | 19:34:50 |
| Mic92 | John Ericson: I think so. I have one regression test already in the upstream crate, but there might be no e2e test in hydra itself. | 19:36:00 |
| John Ericson | do we just need to build a fixed ouput derivation? | 19:36:38 |
| Mic92 | yes, file:// might work for this, otherwise a small webserver | 19:38:17 |
| John Ericson | OK working on a test | 19:45:26 |
| Mic92 | John Ericson: maybe have a look around what other kinds of derivations we are not covering | 20:00:46 |
| John Ericson | ok | 20:01:30 |
| John Ericson | also heard you | 20:01:34 |
| John Ericson | FYI I added a new check derivation to make sure the perl schema is up to date | 23:58:43 |
| 14 Apr 2026 |
| Mic92 | I discovered this class of reproducibility issues while looking at staging-hydra: https://github.com/NixOS/nixpkgs/pull/509752 | 10:05:24 |
| 15 Apr 2026 |
 dramforever | interesting, did we replace fetchpatch with fetchpatch2? i thought that was only a problem with fetchpatch2? | 01:04:25 |
| dramforever | (.patch and .diff differing is expected. the patch hunks are split per commit in .patch but not .diff) | 01:06:54 |
|  拆铁道拔线杆 joined the room. | 11:07:44 |
| 16 Apr 2026 |
 linj | Mic92 changed most part of fetchpatch to fetchpatch2 in the doc, but not everyone agrees with that change, see discussion after the merge of https://github.com/NixOS/nixpkgs/pull/406131 | 09:41:21 |
 vcunat | Over longer term, I think we need to carefully create a fetchpatch3 which won't be worse than either. | 09:45:36 |
| vcunat | * Over longer term, I think we need to carefully create a fetchpatch3 which won't be worse than either 😅 | 09:46:19 |
 connor (burnt/out) (UTC-8) | fetchpatch' so there’s no ambiguity as to which to choose ;) | 14:12:45 |
 figsoda | fetchpatch4FinalVersionUseThisInstead | 14:14:00 |
| 20 Apr 2026 |
|  siraben joined the room. | 19:12:22 |
| siraben | Hi, I'd like to get hydra build logs and metadata in bulk, for analyzing regressions on Nixpkgs over time, who can I talk to for this? Happy to pay as well to offset the infra/compute cost. | 19:13:20 |
| siraben | data is used for non-commercial purposes | 19:14:13 |
 hexa | repeatedly or one-off? | 19:15:20 |
| siraben | one off | 19:18:29 |
| 21 Apr 2026 |
|  gigacode joined the room. | 01:37:32 |
| vcunat | You can get logs from the CDN, e.g.:
- get .drv path /nix/store/6xvsgsnlap9vhc4xhnzfyax9rz69n7f1-filterpath-1.0.4.drv
- get https://cache.nixos.org/log/6xvsgsnlap9vhc4xhnzfyax9rz69n7f1-filterpath-1.0.4.drv
| 05:53:59 |
| vcunat | Some metadata is in .narinfo, too. (the same CDN again)
$ nix eval -f. nix.outPath
"/nix/store/bnrjki4vg8lw5k8zcfvk9sshda2k1ydf-nix-2.34.6"
https://cache.nixos.org/bnrjki4vg8lw5k8zcfvk9sshda2k1ydf.narinfo
| 05:57:36 |
| 24 Apr 2026 |
|  zowoq joined the room. | 06:42:00 |
