| 22 Apr 2026 |
 hexa | ok, everything prepared | 20:08:07 |
| hexa | will postpone until tomrrow evening ig | 20:08:12 |
| 23 Apr 2026 |
 tfc | vcunat: regarding this PR:
https://github.com/NixOS/nixpkgs/pull/510385
the fix will be to use hostPkgs.writeText instead of pkgs.writeText. i am going to create a PR to fix that. i am however a bit confused why this is blocking nixpkgs-unstable when i don't see the testdriver-config change. does "is blocking the nixpkgs-unstable channel" mean that this blocks the changes from going into nixpkgs-unstable branch?
i guess i would still open the PR normally against staging-nixos is that right?
| 08:24:19 |
 Vladimír Čunát | It fails the release-checks job which is critical for that channel (but it's confusing, as multiple issues were there recently). It only affects cases when darwin is included, so the Linux-only channels are not affected fortunately (which is why it took longer to notice, too). | 08:27:19 |
| Vladimír Čunát | Will the fix rebuild many NixOS tests? | 08:28:01 |
| tfc | only on macos if we even build them there | 08:28:18 |
| Vladimír Čunát | Ideally we'd merge it sooner than through staging-nixos. | 08:28:18 |
| tfc | on linux hostPkgs == pkgs so i think no derivation will change | 08:28:31 |
| Vladimír Čunát | OK, then we should target master with this. | 08:28:42 |
| tfc | i will ping you once the PR is open. takes a bit to download everything on the mac, don't have too fast internet this week | 08:29:24 |
| Vladimír Čunát | I can run it on a decent machine locally, say if you open a draft PR. | 08:35:03 |
| Vladimír Čunát |
[queue-runner] will postpone until tomrrow evening ig
Maybe we'll even have post-staging* nixos-unstable by then. All blockers have passed, so it's mostly just a matter of Hydra crunching through those thousands of NixOS tests.
| 08:39:19 |
| Vladimír Čunát | *
[queue-runner] will postpone until tomrrow evening ig
Maybe we'll even have post-staging* nixos-unstable by then. All blockers have passed, so it's mostly just a matter of Hydra crunching through those thousands of NixOS tests (but 24h is roughly the time required for that).
| 08:39:45 |
| tfc | vcunat: https://github.com/NixOS/nixpkgs/pull/512733 | 09:57:13 |
|  Yifei Sun changed their profile picture. | 22:01:51 |
| 24 Apr 2026 |
|  zowoq joined the room. | 03:42:44 |
| Vladimír Čunát | nixpkgs-unstable updated 🎉 It had almost 8 days of lag at that point, sadly, partially because of getting multiple blockers. | 05:47:49 |
 matthewcroughan | What is the etiquette of merging things that cause large amounts of rebuilds into staging? | 14:09:30 |
| matthewcroughan | I want to merge https://github.com/NixOS/nixpkgs/pull/511517 but don't know if it'll cause people to yell at me for doing the wrong thing | 14:10:14 |
 leona | merging large rebuilds to staging is the intended thing. It would be good if you could test a few dependent bulids if they could fail. Otherwise you can likely just merge | 14:11:45 |
| matthewcroughan | I believe it to be the right fix, yeah, and I have tested it on a pi zero, so full deps really | 14:12:09 |
| matthewcroughan | i.e a toplevel containing it as a closure | 14:12:17 |
| matthewcroughan | * i.e a toplevel containing it in the closure | 14:12:38 |
| matthewcroughan | An app that depends on it was failing due to it, so that's how I tested it | 14:12:51 |
| leona | imo fine to merge it then. (ideally a maintainer approves as always) | 14:13:06 |
| matthewcroughan | It's kind of been sitting there for a bit, I wanted to speed it up, I can almost guarantee that if I do nothing it'll still be there in a week, and nobody will review it. | 14:13:51 |
| matthewcroughan | Which would be fine if it weren't such a trivial change | 14:15:05 |
 Alyssa Ross | I think I need to start being less indirect when I'm writing commetns on your PRs :)
In this case, what I meant is "if you confirm that your best guess is that upstream would probably consider it sensible to list this in their pyproject.toml dependencies, I would merge the PR if it were changed to do that" | 14:16:36 |
| matthewcroughan | Rather my understanding is that this mistake of putting pybind11 in nativeBuildInputs seems to be common throughout nixpkgs | 14:17:16 |
| Alyssa Ross | I wonder why that specifically | 14:17:52 |
