24 Aug 2024 |
emily | I genuinely think we should use libcapsule to obsolete nixGL. | 20:17:42 |
emily | and maybe even also to fix the whole running graphical programs from an old NixOS thing. | 20:17:56 |
emily | I just sort of bring it up periodically to try and nerd-snipe someone who isn't me into trying it. | 20:18:30 |
K900 | libcapsule is hard for arbitrary things | 20:24:15 |
emily | I did see that it apparently claims it doesn't handle separate libcs? | 20:38:18 |
emily | which is pretty sad. | 20:38:21 |
emily | (…couldn't we just load the host GPU drivers in a separate process and carry the glvnd ABI across IPC?) | 20:39:27 |
emily | (I guess it'd require a good amount of bookkeeping to share the right bits of memory) | 20:39:44 |
Ralith | feasible in principle | 23:40:33 |
Ralith | a custom vulkan loader, too | 23:41:23 |
emily | how hard can it be | 23:43:48 |
Ralith | mostly just tedious, I think | 23:52:03 |
Ralith | that's why you'd have to pay me :P | 23:52:09 |
25 Aug 2024 |
| @mall0c:matrix.org left the room. | 17:41:12 |
emily | https://notgull.net/announcing-dozer/ | 22:22:51 |
emily | maybe we can escape bootstrapping from old pre-C++ GCCs by writing a C++ compiler in Rust instead :) | 22:23:42 |
26 Aug 2024 |
Sashanoraa.gay (she/her, ze/zir) | Lol | 00:29:39 |
ghpzin | Saw this backlinked in tracking 1.80 issue:
https://github.com/rust-lang/cargo/pull/14452#issuecomment-2309726000 | 10:20:29 |
ghpzin | * Saw this backlinked in tracking 1.80 issue:
https://github.com/rust-lang/cargo/pull/14452#issuecomment-2309726000
(not sure what will happen if it gets merged, but I think nothing very good) | 10:22:38 |
ghpzin | * Saw this backlinked in tracking 1.80 issue:
https://github.com/rust-lang/cargo/pull/14452#issuecomment-2309726000
(not sure what will happen if it gets merged, but probably nothing very good) | 10:23:40 |
Alyssa Ross | oh no | 11:03:41 |
blitz | btw, is rust-hypervisor-firmware currently broken on unstable? | 13:25:55 |
emily | In reply to @9hp71n:matrix.org Saw this backlinked in tracking 1.80 issue: https://github.com/rust-lang/cargo/pull/14452#issuecomment-2309726000 (not sure what will happen if it gets merged, but probably nothing very good) oh this is a terrifying precedent to set | 14:31:14 |
emily | more motivation for the Rust rework, which would mitigate the damage | 14:31:40 |
emily | they should really really really just do some hack in the compiler | 14:32:19 |
emily | I really don't understand why Rust settled on defaulting to the only impl of a trait if one exists | 16:11:16 |
emily | it's such a compat hazard. Haskell doesn't do it | 16:11:26 |
emily | I mean I assume "ergonomics" but… | 16:11:34 |
@aloisw:kde.org | In reply to @emilazy:matrix.org I really don't understand why Rust settled on defaulting to the only impl of a trait if one exists Same, and I guess they could even change it under the same policy that allows all these breaking changes, so that the entire cost can be paid at once. | 16:53:49 |
emily | apparently it would break .as_ref() on Cow and stuff | 16:56:46 |