27 Feb 2024 |
K900 | Always | 19:57:23 |
K900 | But if the dict has startCommand , it'll use that | 19:57:33 |
K900 | And if it has anything else, it'll use THAT THING | 19:57:41 |
@qyliss:fairydust.space | Then I'm not sure this deprecation warning counts? If there was no possible way to get that functionality without using startCommand… | 19:58:18 |
@qyliss:fairydust.space | What I mean is that, I don't think we can expect users to have followed an unactionable deprecation warning. | 19:58:46 |
K900 | No | 19:59:40 |
K900 | I mean | 19:59:43 |
K900 | If you do create_machine({"startCommand": "..."}) | 19:59:52 |
K900 | You don't get the warning | 19:59:55 |
K900 | And you get the right thing done | 20:00:01 |
@qyliss:fairydust.space | Oh, I see. | 20:00:15 |
K900 | And after my PR you just need to change that to create_machine(start_command="...") now | 20:00:20 |
@qyliss:fairydust.space | That makes more sense —-I thought I'd have seen the warning. | 20:00:26 |
@qyliss:fairydust.space | Right | 20:00:30 |
K900 | But if you do create_machine({"hda": "foo.qcow2"}) or whatever | 20:00:37 |
K900 | That would get you the warning | 20:00:44 |
@qyliss:fairydust.space | Right | 20:00:48 |
K900 | And now it would blow up | 20:00:49 |
@qyliss:fairydust.space | Yeah I agree the non-startCommand keys don't need a deprecation period then. | 20:01:06 |
@qyliss:fairydust.space | The reason to do a deprecation period is usually to make it possible to have some code that works with both stable and unstable. | 20:01:34 |
K900 | I guess that's true | 20:01:45 |
K900 | I can add a hack for that | 20:01:48 |
@qyliss:fairydust.space | And then remove it after branch-off. | 20:01:59 |
@qyliss:fairydust.space | Usually we silently add the new API, wait a release, add a warning, wait a release, remove the old one. | 20:02:17 |
@qyliss:fairydust.space | But I don't think that's necessary here because warnings from the test driver are not nearly as disruptive as warnings from Nix. | 20:02:31 |
@qyliss:fairydust.space | So adding a warning straight away should be fine. | 20:02:49 |
K900 | @Alyssa Ross can you test your stuff against my branch? | 20:15:28 |
K900 | It should warn now | 20:15:31 |
K900 | And hopefully catch a whole bunch of stupid cases | 20:15:40 |
K900 | (thanks mypy) | 20:15:42 |