| 15:27 | <smaug> | Luke Warlow: curious, which subtests are you looking at in https://github.com/whatwg/html/issues/12016 ? |
| 15:27 | <Luke Warlow> | "customElements.define must get callbacks of the constructor prototype" was I believe one of them, but basically all the customElements.define ones |
| 15:27 | <smaug> | And IIRC there is an old issue to make the spec bit more reasonable by relying on a webidl dictionary |
| 15:28 | <smaug> | (I'd be surprised if all the .define ones check the order) |
| 15:29 | <Luke Warlow> | https://github.com/web-platform-tests/wpt/blob/1b23e54b99/custom-elements/CustomElementRegistry.html#L279 |
| 15:30 | <smaug> | I can't remember having any issues with this when implementing, so I wonder if I just accidentally somehow got the same result as in wpt |
| 15:31 | <Luke Warlow> | I'm guessing you either looked at WPT for the ordering or put it in the logical place, and it happened to match WPT. |
| 15:31 | <smaug> | Luke Warlow: https://github.com/whatwg/html/issues/3580 is the other issue I'm thinking |
| 15:32 | <Luke Warlow> | I think it's just a simple switcheroo in the spec as a quick fix. Firefox and Chromium agree, and I've swapped my Servo PR to use what's in the test. |
| 15:34 | <smaug> | yeah, the order shouldn't really matter in practice |
| 16:10 | <Luke Warlow> | https://github.com/whatwg/html/pull/12017 - pushed up a quick fix hopefully it's okay. |