| 15 Sep 2021 |
 Vladimír Čunát | I didn't check. Were there some reverted merges? | 06:40:37 |
 lukegb (he/him) | Yeah, but I think we ended up in a good state: mostly it was down merges so we should be alright. We had some shenanigans between staging, python-unstable and staging-next | 06:41:41 |
 trofi | What would be a rule of thumb for a change to be considered big WRT caused rebuilds to go to staging instead of master branch? I was surprised to see 2.5K being not too big: https://github.com/NixOS/nixpkgs/pull/137973 | 08:40:45 |
| lukegb (he/him) | It doesn't get moved automatically | 08:43:12 |
| lukegb (he/him) | It depends a bit on the ecosystem as well, e.g. 1k *pure* Python builds might not be a huge amount | 08:43:57 |
| Vladimír Čunát | Importance also plays a role. If it's a significant (security) fix, the threshold would move higher. | 08:48:54 |
| trofi | From red color of 10.rebuild-linux: 501+ label I assumed 500 is big enough. But 10.rebuild-linux: 2501-5000 does not look as scary and now I'm confused :) | 09:02:07 |
| trofi | Say, for a general case of bumping a C library when you don't really know the impact and it's not super urgent. What would it be? 5k+? | 09:02:43 |
| Vladimír Čunát | I might put it lower if it's not urgent, perhaps 1-2k. (personally, I'm not aware of any hard rule) Especially if that includes rebuild of chromium and similar ones. | 09:18:18 |
 Sandro | except python and go everything over 1000 or 1500 but definitely over 2000 should go to staging especially if it also rebuilds chromium. | 18:12:10 |
| Vladimír Čunát | Kernel updates are another exception. | 18:25:43 |
| Vladimír Čunát | * Kernel updates are another exception. (labels put them at 1000--2500) | 18:26:26 |
| trofi | WDYT of something like https://github.com/NixOS/nixpkgs/pull/138052 | 20:37:27 |
| trofi | It's inspired by guix's rules (there staging is called core-updates) and gives nice initial impression of what goes where: https://guix.gnu.org/manual/en/html_node/Submitting-Patches.html | 20:38:55 |
| 16 Sep 2021 |
 hexa | let's all remember that the next staging-next run is reserved for systemd 249 | 03:18:12 |
| Sandro | So we skip another python unstable or what does this mean exactly? | 04:39:26 |
| Vladimír Čunát | One python dependency of all NixOS tests is broken in staging, by the way. | 07:49:42 |
| trofi | https://github.com/NixOS/nixpkgs/pull/137617 should fix it | 09:14:13 |
| trofi | * https://github.com/NixOS/nixpkgs/pull/137799 should fix it (amended link) | 09:14:48 |
| Sandro | cachix: CppStdException "Exception: path '/nix/store/q1lx81pcqj3030rv1hwyifixjh9cp5c7-python3.9-sh-1.14.2' is not valid; type: nix::InvalidPath"
| 09:35:43 |
| Sandro | Not sure what causes this | 09:35:57 |
 asbachb | Finally my bottles changes are in staging-next :D | 18:44:21 |
| 17 Sep 2021 |
| Vladimír Čunát | staging-next is contained in master now. | 07:02:03 |
| hexa | In reply to @hexa:lossy.network
let's all remember that the next staging-next run is reserved for systemd 249 I missed that systemd was rebased and merged into staging | 11:40:28 |
| lukegb (he/him) | so... does that me we can start the next staging-next cycle? | 11:56:20 |
| hexa | I believe so. | 12:12:13 |
| hexa | would you like to? | 12:12:18 |
| lukegb (he/him) | What's the process? Merge staging into staging-next and open a PR? | 12:15:09 |
| lukegb (he/him) | Oh, it's just a FF anyway | 12:15:41 |
| lukegb (he/him) | or not | 12:19:14 |
