| 4 Jul 2026 |
 emily | are my Nixpkgs contributions that bad? | 14:31:25 |
| emily | so can we eat ldacbt sized rebuilds on Linux? | 14:31:57 |
 K900 | It's Pipewire | 14:32:05 |
| K900 | So like | 14:32:06 |
| emily | it's like 4k | 14:32:08 |
| K900 | 30k-ish? | 14:32:10 |
| K900 | I think? | 14:32:15 |
| emily | I really don't want to have to bifurcate the entire build system | 14:32:15 |
| K900 | Yeah | 14:32:19 |
| emily | https://github.com/NixOS/nixpkgs/actions/runs/26743154227?pr=514052? | 14:32:21 |
| K900 | Hm OK | 14:32:41 |
| K900 | That's not nearly as bad | 14:32:55 |
| K900 | I should do libpipewire treewide | 14:33:02 |
| K900 | Upstream first | 14:33:08 |
| emily | hmmmm | 14:38:50 |
| emily | why do we build this into one single library? | 14:38:54 |
| emily | given that we need separate .pc files anyway | 14:39:16 |
| emily | oh okay cool, so the immediate reason it's failing on Darwin is that libldac-dec just doesn't build on Darwin | 14:40:11 |
| K900 | Yay | 14:52:41 |
| K900 | Because like 90% of the code is shared | 14:52:53 |
| emily | but we're building it in a way that produces a library that nothing else expects or produces, right? | 14:53:22 |
| emily | like, we're merging in the abr/* stuff which is not what the Android build system does either | 14:53:33 |
| K900 | abr stuff is technically supposed to be another entry point | 14:54:50 |
| K900 | But I don't think anything actually uses that one | 14:54:54 |
| emily | # LDAC bitstream format assumes LE; source has endian checks
platforms = lib.platforms.littleEndian;
| 15:09:14 |
| emily | wat? | 15:09:15 |
| emily | cc Puna? | 15:09:34 |
| emily | https://github.com/open-vela/external_libldac/blob/6e0474c954103ff7aefb7908b97e65522c9208fa/src/ldacBT_internal.c#L256 oh, that's fun. | 15:09:45 |
| emily | so the header it includes that doesn't exist on Darwin is just there to error out. | 15:09:57 |
| emily | Alyssa Ross: …why on earth does our Meson hook not default to release builds? | 15:31:45 |
