| 29 Jun 2025 |
 emily | and at the point where it is just a very simple layer for the most basic networking configuration I don't think it would be meaningful tech debt at all, but you still avoided breaking "every config that sets up the most basic network stuff" | 02:42:35 |
| emily | * anyway I think most of the friction of the deprecation here just comes from breaking something ~every single system has for really basic needs | 02:42:42 |
 hexa | networkd has no ovs support, that's why that part is hacky | 02:42:47 |
| emily | right | 02:42:57 |
| hexa | any ovs support should go in a services.openvswitch module though | 02:42:59 |
| hexa | but the ovs package is barely maintained as well | 02:43:05 |
| emily | well substitute any of the "fancy" stuff in there as an example :p | 02:43:05 |
| emily | if you are configuring big brain stuff then little basis to object to switching to the native options | 02:43:21 |
| emily | it's the "I just want DHCP" or "I just want this IP" usecase where it is annoying to break every single user | 02:43:40 |
| hexa | fair | 02:44:03 |
| hexa | so the networking options should mostly just be simplified to the basics | 02:44:18 |
| hexa | with the complicated stuff being deprecated | 02:44:26 |
| hexa | and then we back it all with networkd instead | 02:44:38 |
| hexa | and by simple I mean … static ip addressing and dhcp | 02:44:54 |
| hexa | if you need anything else … go, use the real interface | 02:45:06 |
| emily | I think that would be reasonable yeah. but I also think we could just default useNetworkd tomorrow probably, before the slimming | 02:45:09 |
| emily | like, people have been using it for a very long time now | 02:45:14 |
| hexa | probably should | 02:45:28 |
| hexa | and then kill the scripted networking code | 02:45:32 |
| emily | I don't know if there's any blockers or if it's just "someone needs to open the PR and someone needs to hit merge" | 02:46:00 |
| hexa | it's so great to have you and me as commiters 😆 | 02:46:14 |
| emily | :D | 02:46:20 |
| emily | one specific thing | 02:46:28 |
| emily | is that scripted initrd only has scripted networking | 02:46:34 |
| emily | and initrd networking is pretty common | 02:46:39 |
| emily | so I think the actual thing blocking it is just systemd stage 1 by default | 02:46:46 |
| * hexa screams into the void | 02:46:51 |
| emily | which does have a small number of actual blockers | 02:46:58 |
| emily | pushing things over the finish line is hard | 02:47:07 |
| emily | once scripted initrd is gone though I see no reason to keep scripted networking around | 02:47:25 |
