| 31 Aug 2025 |
 aloisw | Either that or XDG_MENU_PREFIX=plasma-. I don't know which one is expected to cause less problems. | 13:37:16 |
| aloisw | nix-locate only finds one in libsForQt5.kservice.bin (!). | 13:38:25 |
 K900 | Yaaaay | 13:38:34 |
| K900 | What the fuck does GNOME do then | 13:38:40 |
| aloisw | XDG_MENU_PREFIX=gnome- if the Debian VM is to be trusted. | 13:40:01 |
 Sandro 🐧 | FYI https://github.com/NixOS/nixpkgs/issues/379307 | 14:03:28 |
| Sandro 🐧 | https://github.com/NixOS/nixpkgs/issues/24913 | 14:03:34 |
| Sandro 🐧 | and the dolphin module PR https://github.com/NixOS/nixpkgs/pull/438914 | 14:04:36 |
 Herz uwu | In reply to @k900:0upti.me
But generally no Well heck | 14:09:01 |
| Herz uwu | I did manage to add nixGL as an overlay and making nixGLDefault available, but I didn't get the flake to run this way
Only if I nix run --impure the result binary manually with nixgl does it work | 14:11:31 |
| K900 | Yeah it's a problem | 14:13:19 |
| K900 | That we don't really have a good solution for | 14:13:25 |
| K900 | Maybe one day someone will make libcapsule actually usable | 14:14:28 |
| aloisw | From a look at the XDG menu specification it seems that the XDG_MENU_PREFIX is indeed supposed to be desktop-specific. Of course this will not help at all the folks building their own mini-desktops out of window managers and helper applications. Maybe the best solution would be to write a minimal flat menu for such use cases. | 14:22:42 |
| K900 | Yeah I think that's the way to go if anything | 14:23:51 |
 emily | won't work because libc, right? | 15:08:09 |
| K900 | I mean theoretically | 15:37:28 |
| K900 | If we adjust Mesa and such | 15:37:32 |
| K900 | To behave in that situation | 15:37:36 |
| emily | to behave when there are two totally different libcs? | 16:07:05 |
| emily | that sounds hard | 16:07:06 |
| emily | my idea sounds simpler :P | 16:07:17 |
| K900 | Which one | 16:09:29 |
| emily | well the same "stub libglvnd that talks to host libglvnd" thing for i686 would work for other distros right | 16:11:40 |
| K900 | If you do IPC across processes, yes, theoretically | 16:12:22 |
| K900 | But that's still significantly slower | 16:12:28 |
| emily | it's going to be lower overhead than, like, virgl at least | 16:14:01 |
| emily | though also you could just load the two processes in the same address space very far away from each other right. I guess that's what libcapsule tries maybe… | 16:14:26 |
| emily | but you wouldn't have to think about weird linker stuff beyond that if it's just single-process "IPC" | 16:15:13 |
| emily | maybe. | 16:15:16 |
