| 13 Jan 2026 |
 K900 | OK I think I have everything building now | 17:45:17 |
| K900 | The next step is to see if it actually runs | 17:45:31 |
| K900 | And after that we can think about wiring up plasma-login-manager | 17:45:40 |
| K900 | OK plasma-login-manager works | 20:40:47 |
| K900 | https://github.com/NixOS/nixpkgs/pull/479797 | 20:40:48 |
| K900 | Have fun | 20:40:51 |
 emily | they made a greenfield login manager that supports X…? | 20:44:10 |
| K900 | No | 20:44:43 |
| K900 | They forked SDDM | 20:44:45 |
| K900 | And it only supports X sessions | 20:44:53 |
| emily | # don't install generic nixos display manager unit
systemd.services.display-manager.enable = false;
I don't think this is safe because we rely on conflicts/ordering stuff with it
| 20:45:29 |
| emily | maybe ElvishJerricco knows for sure though | 20:45:34 |
| emily | for example | 20:45:56 |
| emily | # We can't just rely on 'Conflicts=autovt@tty1.service' because
# 'switch-to-configuration switch' will start 'autovt@tty1.service'
# and kill the display manager.
systemd.targets.getty.wants =
lib.mkIf (!(config.systemd.services.display-manager.enable or false))
[
"autovt@tty1.service"
];
| 20:45:58 |
| emily | I think you'll need to at least make it an alias or whatever | 20:46:28 |
| K900 | Ughhhhhhhhhhh | 20:46:31 |
 ElvishJerricco | ugh | 20:46:41 |
| ElvishJerricco | ideally, the way that would work is that every display manager would provide their own service and alias themselves as display-manager.service and display-manager.service wouldn't actually exist | 20:47:04 |
| K900 | That is what p-l-m does and that is correct yes | 20:47:29 |
| emily | well, this module can start the trend by aliasing :P | 20:47:34 |
| K900 | UNFORTUNATELY | 20:47:38 |
| ElvishJerricco | this particular hack is, uh.... Not in line with that ideal :P | 20:47:47 |
| ElvishJerricco | but it was convenient at the time | 20:47:53 |
| emily | as in it aliases to display-manager.service upstream? | 20:48:08 |
| K900 | Yes | 20:48:12 |
| K900 | Which of course does not help this hack | 20:48:19 |
| emily | ok. not sure if there's a good way to declare that on the NixOS level I guess | 20:48:39 |
| ElvishJerricco | we could just make that hack use its own enable flag and then have the display manager modules in nixos set that flag | 20:49:02 |
| emily | I mean you could also just bodge this into the conditional there | 20:49:09 |
| ElvishJerricco | (I still contend that Conflicts=getty@tty1.service is the way to actually disable the getty on that tty, but I need to make stc less insane for that to work) | 20:50:11 |
