| 21 Sep 2025 |
 emily | can you hold off on it? I spent quite a while looking into and addressing this and think you are adjusting the knobs in the wrong place here | 20:16:30 |
| emily | (btw, this type of issue affects many packages before even getting to Rust, fwiw, though I have addressed a significant amount of it locally) | 20:19:05 |
 Winter | unfortunately ~4 months ago me could not see into the future, but yeah, sure ig | 20:23:17 |
| Winter | what are the right knobs to adjust then? | 20:23:47 |
| Winter | * | 20:25:41 |
| emily | I didn't mean it like that, sorry :( | 20:34:08 |
| emily | I didn't see your PR before writing stuff myself | 20:34:14 |
| emily | heading out the door but can share my patches in a couple hours | 20:34:31 |
| Winter | it’s fine, i don’t mind as long as it gets fixed | 20:35:06 |
| Winter | thanks, sounds good | 20:35:12 |
| emily | basically Cargo has native support for this upcoming but in the meantime I think it probably makes more sense to do it in the rustc wrapper. there's also an explicit flag to the rustc bootstrap we can use, and some misc. stuff like installation logs that need cleaning up | 20:35:20 |
 Grimmauld (any/all) | emily You have a good idea about how to disable this test (https://gitlab.gnome.org/GNOME/librsvg/-/blob/7a19a9216b4d81ba48c834351ab1e75793ae80f7/rsvg/tests/errors.rs#L17) in librsvg (https://github.com/NixOS/nixpkgs/blob/8eaee110344796db060382e15d3af0a9fc396e0e/pkgs/by-name/li/librsvg/package.nix#L50) ? I tried checkFlags = [ "--skip=too_many_elements" ]; but the test is still being run. Its meson shelling out to cargo, so our tooling is wonky... You know our rust/llvm stack, so maybe you have a good idea? If not i'll just substitute it away (https://gitlab.gnome.org/GNOME/librsvg/-/blob/7a19a9216b4d81ba48c834351ab1e75793ae80f7/rsvg/meson.build#L383). | 21:19:04 |
| emily | this is cursed | 21:24:54 |
| Grimmauld (any/all) | well yes XD | 21:25:07 |
| emily | @qyliss:fairydust.space turns out people nix Meson and Cargo after all | 21:25:11 |
| Grimmauld (any/all) | here i hoped you'd have some ace up your sleeve, but i guess if "what the hell is this" is your reaction, substituting it away will be my approach for now | 21:26:06 |
| Grimmauld (any/all) | its part of the libxml bump, only cursed things can ever come of this | 21:27:31 |
| emily | I don't see an obvious way to pass args through to https://gitlab.gnome.org/GNOME/librsvg/-/blob/7a19a9216b4d81ba48c834351ab1e75793ae80f7/meson/cargo_wrapper.py | 21:27:44 |
| emily | you can patch the test to be should_panic I suppose | 21:28:09 |
| Grimmauld (any/all) | then it would break again if at some point the test starts succeeding again? | 21:30:58 |
| emily | which is good because you know to drop the patch | 21:31:58 |
| Grimmauld (any/all) | thats a good point | 21:32:07 |
| emily | how does it fail anyway? | 21:32:38 |
| Grimmauld (any/all) | i have no clue yet lol | 21:36:15 |
| Grimmauld (any/all) | i mean it fails, it expects to hit a limit and just doesn't | 21:36:32 |
| emily | so it succeeds unexpectedly? | 21:36:44 |
| Grimmauld (any/all) | libxml changed how it counts xml objects, so either their reproducer is now too small or libxml broke something | 21:36:51 |
| emily | that's a bit concerning if the limit is there for security | 21:36:58 |
| Grimmauld (any/all) | if libxml broke something, then at least arch will have billion laughs once again lol | 21:37:10 |
| emily | I see | 21:37:01 |
