| 06:59 | <annevk> | Yeah, that seems to have worked. Thanks! |
| 07:01 | <annevk> | Note that OP had already left the channel. |
| 08:07 | <annevk> | Domenic: https://github.com/whatwg/html/pull/9538#discussion_r1328593134 still seems concerning with the unsafe parser methods |
| 08:07 | <annevk> | Domenic: is there test coverage for that state? In particular because implementations also are not aligned on HTMLDocument vs XMLDocument vs Document and such |
| 08:07 | <Domenic> | annevk: I agree with Joey here that everything is initialized and all spec code that can be shared, is shared. |
| 08:07 | <annevk> | Domenic: so you're saying type should be "xml"? |
| 08:07 | <Domenic> | There is test coverage for that state. |
| 08:07 | <Domenic> | No, type gets set to html elsewhere |
| 08:08 | <annevk> | Domenic: note that type and content type are distinct |
| 08:08 | <Domenic> | https://whatpr.org/html/9538/dynamic-markup-insertion.html#parse-html-from-a-string |
| 08:10 | <annevk> | Domenic: I see, so URL will be about:blank meaning relative URLs will break? |
| 08:10 | <Domenic> | Yes, this is intentional. |
| 08:10 | <Domenic> | We discussed it upthread in the PR |
| 08:10 | <Domenic> | If people want URL inheritance we can add a { url } option in the future. |
| 08:13 | <annevk> | https://github.com/whatwg/html/pull/9538#discussion_r1293136059 is hard to find and the commit message doesn't describe it, but okay. |
| 08:15 | <annevk> | Anyway, I guess that's fine. Though new Document() does inherit the origin. Hmm. Maybe it shouldn't have? |
| 09:24 | <Domenic> | Yeah the inheritance seems all very magic to me, I say, if you want it, we should make that the non-default. |
| 09:26 | <Domenic> | I recorded that we need tests for document.origin in Joey's CL, I missed that one, thanks. |
| 09:51 | <annevk> | Domenic: not sure what I want, but I can imagine web developers wanting it for URL at least, having <img> not work has been an issue of sorts before I vaguely recall |
| 09:51 | <annevk> | (work in the sense of .href not returning something useful) |
| 09:52 | <annevk> | zcorpan: https://github.com/whatwg/html/pull/9689 should be good now, but happy to wait a bit to see if more people want to chime in |
| 11:54 | <sideshowbarker> | annevk: about https://bugs.webkit.org/show_bug.cgi?id=192000, would a patch that conforms WebKit to the Content-Type parsing requirements in the Fetch spec actually be accepted? Or instead not — because it would break on too many existing sites? |
| 12:58 | <Domenic> | Domenic: not sure what I want, but I can imagine web developers wanting it for URL at least, having In reply to @annevk:matrix.org The simplest workaround is for them to inject a |
| 13:03 | <annevk> | sideshowbarker: not sure all of it can be fixed in WebKit, but I think it's an area we'd be willing to clean up, might need some kind of preference so it's easy to revert for select sites? |
| 13:05 | <annevk> | Domenic: I guess Google should know since iirc Google mainly pushed for this method? |
| 13:07 | <Domenic> | Yeah we haven't heard any, that's why I promoted the simpler option. |
| 14:02 | <annevk> | I wonder if it's too late to change new Document(), but I also don't really want to work on it myself |