| 8 Aug 2025 |
 aidalgol | K900: The steam container change seems to have caused a regression in heroic: https://github.com/NixOS/nixpkgs/issues/432037 | 19:16:22 |
| aidalgol | I haven't yet had a chance to look into it, but does anything leap out at you as a likely cause? | 19:16:38 |
 K900 | Hmm | 19:16:51 |
| K900 | Somehow dbus something? | 19:16:56 |
| K900 | I don't know how it implements single instance stuff | 19:17:02 |
| K900 | ...oh god it's not privateTmp is it | 19:17:09 |
| aidalgol | This gets weird. I can't reproduce the issue with heroic on master. It just runs a second instance. But if I override privateTmp to false, then I get the symptom described in the issue. | 20:48:19 |
| K900 | I think it's intended to not run a second instance? | 20:49:22 |
| aidalgol | correct | 20:49:35 |
| aidalgol | but before the steam wrapper change, trying to run a second instance would just raise the existing instance's window. | 20:50:10 |
| K900 | As in, the "runs two instances" behavior is wrong | 20:50:21 |
| K900 | And the expected behavior is to activate the existing one | 20:50:29 |
| aidalgol | actually, that might be an intentional change. | 20:50:40 |
| aidalgol | OK, how the hell do I build a Windows executable? I need to package this thing for GOG Galaxy support in Heroic to work properly.
https://github.com/imLinguin/comet/tree/main/dummy-service | 22:10:14 |
| aidalgol | None of the guides for normal cross-compilation in nixpkgs quite fit this odd scenario. | 22:10:54 |
| aidalgol | i.e. https://wiki.nixos.org/wiki/Cross_Compiling and https://nix.dev/tutorials/cross-compilation.html by my reading are describing setting up a dev environment for cross compiling using nixpkgs, not packaging, say, a Windows executable for use in an x86_64-linux NixOS package. | 22:15:33 |
| aidalgol | Comet's upstream releases include GalaxyCommunication-dummy.exe, but our comet-gog package does not. I think this .exe was added upstream sometime since the nixpkgs package was added. | 22:18:14 |
| aidalgol | * Comet's upstream releases include GalaxyCommunication-dummy.exe, but our comet-gog package does not because it only builds comet. I think this .exe was added upstream sometime since the nixpkgs package was added. | 22:18:27 |
| 9 Aug 2025 |
| aidalgol | SigmaSquadron: Is there anything we can do around the heroic package to get people to bother nixpkgs people first rather than going straight to upstream? This happens a lot. | 02:06:48 |
 SigmaSquadron | Could patch the sources so whenever we get an error message, it provides a link to Nixpkgs instead. | 02:22:45 |
| aidalgol | Yeah, maybe I should do that. I've been hesitant to modify the upstream source much. | 02:57:14 |
| aidalgol | Regarding the Windows helper executables, I have packaged them in PR#432183, and I'm not sure the way I have done so is acceptable in nixpkgs, so I would really appreciate a review on this. | 05:42:20 |
| aidalgol | It's a rather odd scenario because they're Windows dependencies for a Linux application, for games run under Proton. | 05:42:56 |
| K900 | @aidalgol pushed to your PR | 05:55:48 |
| K900 | Can you test | 05:55:49 |
| aidalgol | Yup, fix still works. | 05:59:40 |
| K900 | OK hold on I noticed a thing | 06:20:19 |
| K900 | Pushing a fix now | 06:20:21 |
|  seapat set a profile picture. | 14:23:03 |
| 10 Aug 2025 |
|  matthewcroughan changed their display name from matthewcroughan to matthewcroughan @ WHY2025 (DECT: 8793). | 11:43:22 |
