18 Sep 2025 |
Sergei Zimmerman (xokdvium) | In reply to @philiptaron:matrix.org The update from 2.28 to a newer version is blocked on some regressions Mic mentioned. I'm not tracking that closely. Is 2.32 on track to fix them? A preliminary fix has been backported to 2.30/2.31 | 17:50:44 |
Sergei Zimmerman (xokdvium) | As discussed with emily it might be good enough for now. | 17:51:18 |
Philip Taron (UTC-8) | Is that the only defect that would block the update? | 17:52:57 |
Sergei Zimmerman (xokdvium) | In reply to @philiptaron:matrix.org Is that the only defect that would block the update? Known - yeah. Hopefully not much else | 17:53:23 |
Philip Taron (UTC-8) | The only other one I know about is something that makes nom not understand that downloads are done | 17:54:05 |
Sergei Zimmerman (xokdvium) | In reply to @philiptaron:matrix.org The only other one I know about is something that makes nom not understand that downloads are done The fix is out for it | 17:54:18 |
Philip Taron (UTC-8) | is there something I could link so I could point https://github.com/maralorn/nix-output-monitor/issues/201 at the fix? | 17:54:46 |
Sergei Zimmerman (xokdvium) | Honestly that’s as much of a NOM issue | 17:54:58 |
Philip Taron (UTC-8) | yeah that makes sense, internal-json says it all in the name | 17:55:21 |
Philip Taron (UTC-8) | (hi @maralorn, appreciate you) | 17:56:38 |
maralorn | I agree. | 17:56:48 |
Sergei Zimmerman (xokdvium) | In reply to @philiptaron:matrix.org is there something I could link so I could point https://github.com/maralorn/nix-output-monitor/issues/201 at the fix? Commented on that issue | 17:57:32 |
maralorn | Although the situation got a little hairy now that there is not just different nix versions (which I’d prefer to support at the same time) but also lix and determinate nix. 😄 | 17:57:39 |
maralorn | They all have different internal-jsons at this point. 😄 | 17:58:04 |
Sergei Zimmerman (xokdvium) | In reply to @maralorn:maralorn.de Although the situation got a little hairy now that there is not just different nix versions (which I’d prefer to support at the same time) but also lix and determinate nix. 😄 When it comes to store references the syntax is the same. | 17:58:12 |
maralorn | Fair enough. | 17:58:40 |
maralorn | The true problem with nom is that I just can’t find the time for it. 😄 | 18:00:32 |
Sergei Zimmerman (xokdvium) | As John points out it’d be great to have this type and parser in hnix-store to avoid the need to reinvent the wheel | 18:00:26 |
maralorn | Anyway, this is sliding into offtopic. | 18:00:34 |
Philip Taron (UTC-8) | https://github.com/NixOS/nixpkgs/pull/444079 | 18:01:26 |
Sergei Zimmerman (xokdvium) | In reply to @maralorn:maralorn.de The true problem with nom is that I just can’t find the time for it. 😄 Hopefully Nix code will just get cleaned up enough to either have this directly in nix or expose a more structured and stable endpoint for logs and events that doesn’t require parsing MBs of json | 18:03:36 |
Philip Taron (UTC-8) | OK, I ran tests on Darwin and on Linux. I think we're good for merge of 2.31.2 after a ✅️ @Sergei Zimmerman (xokdvium) | 18:36:14 |
Sergei Zimmerman (xokdvium) | Philip Taron (UTC-8): btw, here's the metrics and the impact from the bindings change: https://hydra.nixos.org/job/nix/master/metrics.nixpkgs/metric/nix-env.qaAggressive.allocations | 20:02:08 |
Philip Taron (UTC-8) | very cool | 20:05:53 |
Philip Taron (UTC-8) | is that powered by pkgs/top-level/metrics.nix ? | 20:06:18 |
Philip Taron (UTC-8) | if so shift your bookmark to https://hydra.nixos.org/job/nix/master/metrics.nixpkgs/metric/nix-env.qa.allocations | 20:06:45 |
Philip Taron (UTC-8) | (qaAggresive and qa had been the same since 2019) | 20:07:21 |
Philip Taron (UTC-8) | https://github.com/NixOS/nixpkgs/pull/442477 | 20:07:37 |
Sergei Zimmerman (xokdvium) | Yeah, it's using a pinned nixpkgs checkout from a62e6edd6d5e1fa0329b8653c801147986f8d446 (23-11) | 20:08:49 |
Sergei Zimmerman (xokdvium) | Yay, finally ASAN https://github.com/NixOS/nix/pull/14016 | 22:41:45 |