| 22 Sep 2025 |
 nim65s | when they agree on something I fetchpath instead | 22:18:01 |
 Grimmauld (any/all) | there is a lot of dead upstreams on the way to things like freecad, in which case it makes sense fetching from gentoo (they did cmake 4 before we did and have patches already) | 22:18:25 |
| nim65s | but here you suggest doing writing a .patch in nixpkgs instead of fetchpatch ? | 22:18:33 |
| Grimmauld (any/all) | hm? no the patch file in nixpkgs is unused, we can just fetch all those patches from gentoo | 22:19:03 |
| nim65s | hum, they did that without opening a PR upstream ? I know it is dead, but that is still the best central place we have | 22:19:25 |
| Grimmauld (any/all) | in the end it is whatever, the output is basically the same (barring drv hashes) | 22:19:34 |
| Grimmauld (any/all) | oh wait i am dumb | 22:20:26 |
| Grimmauld (any/all) | wrong open | 22:20:29 |
| Grimmauld (any/all) | its late, i can't even read... ugh. Time for sleep | 22:20:41 |
| nim65s | OH | 22:20:48 |
| nim65s | yes | 22:20:50 |
| nim65s | same | 22:20:53 |
| Grimmauld (any/all) | i confused opencollada with openscenegraph, fucking hell | 22:20:54 |
| Grimmauld (any/all) | so only pystring, and there i prefer yours | 22:21:05 |
| Grimmauld (any/all) | anyways | 22:21:08 |
| Grimmauld (any/all) | how do people stay up until 4am doing nix things... | 22:21:35 |
| nim65s | when you dont need brain and patching cmake version numbers is only muscle memory, it's easier to stay late x) | 22:22:29 |
 Winter | there’s an answer but it’s one you probably won’t like | 22:22:33 |
| Grimmauld (any/all) | burnout-induced insomnia? | 22:23:02 |
| Winter | …something like that | 22:23:15 |
| Winter | i would know, a few weeks ago i spent a solid… 30h+ awake either doing or stressing about Nix things 😉 | 22:23:46 |
| Winter | can’t recommend | 22:23:53 |
| nim65s | (fun fact, cmake v4 fix for glew in staging is in #444660 and #444006) | 23:14:36 |
| nim65s | (hum, no, 445…) | 23:16:55 |
| 23 Sep 2025 |
| Grimmauld (any/all) | update re libxml2: the update is in a decent spot with few (now-fixed) known breaks. However, the "too many elements" test in rsvg is failing, suggesting it is again possible to build an xml bomb (see billion laughs CVE and such). The PR is waiting for someone to reproduce and/or report upstream (both in rsvg and libxml2), and will only move ahead once confirmed this is not actually a security regression (or the regression is fixed) | 06:54:20 |
| nim65s | https://github.com/NixOS/nixpkgs/pull/445433 | 07:19:57 |
 Vladimír Čunát | Why are you targeting staging and not staging-next? | 07:21:55 |
| Vladimír Čunát | cmake4 default is in staging-next already, since
https://github.com/NixOS/nixpkgs/pull/444862 | 07:22:27 |
| Grimmauld (any/all) | (you can probably also throw these against master, bumping cmake version to 3.10 does not depend on cmake 4. But i guess if hydra starts eval before automerge, then master would be slower to propagate through) | 07:25:56 |
| nim65s | 
Download image.png | 07:27:41 |
