| 2 Oct 2025 |
 lovesegfault | https://github.com/NixOS/nix/pull/14138 | 03:52:59 |
| lovesegfault | the bikeshedding needed is on that isArchiveBased() name, idk what else to call it or if that's good | 03:54:37 |
 getchoo | opened https://github.com/NixOS/experimental-nix-installer/pull/49 so we can use builds with the previous installer fix in nix's main ci | 05:13:58 |
| getchoo | it does this by using action artifacts instead of the cache, and in the process of doing that, i ended up de-duplicating the small mess of workflow files | 05:17:21 |
 Sergei Zimmerman (xokdvium) | In reply to @lovesegfault:matrix.org
the bikeshedding needed is on that isArchiveBased() name, idk what else to call it or if that's good substitutatableFromNarHash? | 06:03:46 |
 Mic92 | getchoo: where you able to run https://github.com/NixOS/experimental-nix-installer/blob/main/assemble_installer.py btw? | 10:59:48 |
| lovesegfault | Nice | 18:56:51 |
| lovesegfault | why do we have this rev and lastModified stuff in fetchTree? | 19:15:53 |
| lovesegfault | it feels to me (naively) really weird that narHash is not sufficient to say "I already have this" reproducibly | 19:16:36 |
| Sergei Zimmerman (xokdvium) | I lack the full context why it's there in the first place. As Eelco mentions here https://github.com/NixOS/nix/pull/14138#pullrequestreview-3293258649 we can probably treat the fetchTarball as final at least. | 19:29:05 |
| lovesegfault | yeah, I'm working on that right now | 19:29:26 |
| lovesegfault | it's also weird to me that I can't explicitly set lastModified for a fetchTree { type = "tarball"; } and make it final | 19:30:00 |
| Sergei Zimmerman (xokdvium) | lovesegfault, Mic92 https://github.com/NixOS/nix/pull/14142. I think that was just an small oversight, right? | 20:14:42 |
| lovesegfault | btw Sergei Zimmerman (xokdvium) are we clear to merge the aws-crt-cpp changes? I've rebased the other work on top of it and made sure it all still works | 20:16:09 |
| lovesegfault | whoops, yes! | 20:16:54 |
| lovesegfault | John Ericson: you're currently the assigned reviewer | 20:33:24 |
 John Ericson | @lovesegfault:matrix.org: I am on a walk right now, but I should do it when I get back, and if you don't mind, make sure I don't forget, I got a lot on my mind right now :) | 21:43:54 |
| 3 Oct 2025 |
|  sinan changed their profile picture. | 02:43:04 |
| lovesegfault | next little bit of the curl-based-s3 work: https://github.com/NixOS/nix/pull/14144 | 03:47:54 |
| Mic92 | Merged. | 07:25:45 |
| getchoo | https://github.com/NixOS/nix/pull/14002 is now all green (consistently) | 08:00:40 |
| Mic92 | Thanks a lot. I also opened https://github.com/NixOS/org/issues/167 | 08:16:19 |
| Mic92 | so you can trigger ci runs etc. | 08:16:55 |
| Mic92 | @getchoo:matrix.org: i added you now to the installer team. There is a python script repo that uploads new releases from a triggered GitHub action that accepts a hydra eval. When i tested it once it failed but it seemed to have worked at least one time. It could be that we know have to set a user agent to get past the ai scraper shield that we added in front of hydra.nixos.org? I didn't had a closer look. | 14:09:15 |
| lovesegfault | thoughts on temporarily extending CI to also test withAWS = false; withCurlS3 = false and withAWS = false; withCurlS3 = true;? | 15:38:14 |
| lovesegfault | I just want to reduce the chance of messing something up while we get all these parts merged | 15:38:28 |
| lovesegfault | https://github.com/NixOS/nix/pull/14147 | 16:59:23 |
| lovesegfault | we can revert it once we're done messing with that codepath | 16:59:46 |
|  asoderlind joined the room. | 17:08:14 |
| lovesegfault | cc. Sergei Zimmerman (xokdvium) on the above, I think? | 17:18:34 |
