| 00:27 | <jmdyck> | In https://html.spec.whatwg.org/#parsing-main-inhead, under A start tag whose tag name is "template", step 8 is Let document be intendedParent's node document.. However, that's now a pointless step because there's no use of document in the algorithm. (There used to be, but the recent commit re fragment parsing algorithm changed the algo's only step that used document.) |
| 00:28 | <jmdyck> | (And if you delete step 8, then intendedParent becomes unused, and if you then delete step 7, adjustedInsertionLocation becomes unused.) |
| 00:59 | <Shannon Booth> | yeah i am looking at that |
| 01:00 | <Shannon Booth> | https://github.com/whatwg/html/pull/12666 |
| 07:44 | <annevk> | I think I found a more difficult case for the realm of nodes issue. Remove a subtree from a nested document. Then remove the nested document. Now insert the subtree. |
| 08:11 | <annevk> | Got quite a few more tests that fail in Firefox because of that realization. Good welcome back from vacation present for Olli. 😛 |
| 08:11 | <annevk> | (Of course, it's not like other browsers don't have work to do here.) |
| 11:38 | <Dominic Farolino> | What's the intent of https://html.spec.whatwg.org/C#concept-sharedworkerglobalscope-constructor-origin? It's set, but only read in a note. I think it should be read again in all the worker equality checks above that note, right? |
| 12:04 | <annevk> | Dominic Farolino: looks like it regressed with 5aa1a598 |
| 12:07 | <Luke Warlow> | I'm assuming this work is relevant to the interop issue we found when working on trusted types. TypeErrors are sometimes from different realms between WebKit/Blink and Firefox. |
| 12:10 | <annevk> | I suspect that's a separate, but related issue. At least when creating exceptions we don't take the realm from the node I think. |
| 12:12 | <Luke Warlow> | It specifically was showing up in the tests where we moved nodes from an iframe, so I thought it might be. I'll admit I didn't look too far into it and we ended up just changing most of the tests to be realm agnostic. |
| 12:13 | <annevk> | It seems constructor origin should have become constructor storage key to keep data: URL workers working. I'll write a patch. |
| 12:14 | <Dominic Farolino> | I just pushed one annevk |
| 12:14 | <Dominic Farolino> | will request your review |
| 12:17 | <Dominic Farolino> | Although I don't think we have to store the constructor storage key on the SWGS, right? (I guess if we want to future-proof it for future things other than origin, but at least it doesn't seem strictly necessary). |
| 12:21 | <annevk> | Dominic Farolino: if we don't store it, where do we get it from? |
| 12:23 | <Dominic Farolino> | We just get it dynamically https://html.spec.whatwg.org/#shared-workers-and-the-sharedworker-interface:obtain-a-storage-key-for-non-storage-purposes-2 but yeah now I'm realizing that's the worker's live one, not the snapshot of its creator... |
| 12:28 | <annevk> | Here's my change: https://github.com/whatwg/html/pull/12680 |
| 12:32 | <annevk> | Dominic Farolino: the real problem is that you cannot derive the storage key from the origin (it's an opaque origin) or the URL (it's a data: URL). That makes these workers rather unique. |
| 12:32 | <annevk> | We have a related problem with dedicated workers that assume they are in their parent's agent cluster but a data: URL dedicated worker arguably should not be. |
| 12:38 | <annevk> | I hope you kept some! Always good to document interop issues even if it takes us a while to get to them. |
| 13:44 | <Luke Warlow> | There's definitely some testing this scenario but I think we seperated them from trusted types itself as the bug was distinct. There's lots of discussion in https://github.com/web-platform-tests/wpt/issues/45405 |
| 20:05 | <Yoav Weiss> | What's the best way for me to take over Luca Casonato 's https://github.com/whatwg/html/pull/10327? Should I open a new PR? |
| 20:32 | <TabAtkins> | I presume so, yes; taking it over directly would require Luca to give you edit rights to their fork. |
| 20:34 | <saliak> | Hey guys, I have been following the discussions here for some time. and I dont seem to understand anything. I come from a computer science background, but I havent done any major dev work. right now I work on a web personalization project, thats how I stumbled upon WHATWG. I want to learn more about how the web works so that I can contribute back to the community that build the web. If anyone could guide me on this it would have been really helpful. Sorry if this isnt the right place to ask this : ) |
| 20:39 | <jmdyck> | "how the web works" is a wide and deep topic. This is probably a reasonable place to ask about some of it. |
| 20:40 | <TabAtkins> | The only issue is that this room, particularly, is mostly used for technical discussion of the actual work behind the HTML specification, rather than general-purpose discussion about web topics. |
| 20:41 | <TabAtkins> | Not that we won't answer questions, of course. |
| 20:46 | <saliak> | Johnny Stenback: How can I get invited for the triage call on 16th July ? |
| 20:52 | <Johnny Stenback> | saliak: Sure can. Preferred email? |
| 22:14 | <jmdyck> | After https://html.spec.whatwg.org/#dom-canvaspattern-settransform, there's an algo introduced by "When a pattern is to be rendered within an area, the user agent must run the following steps to determine what is rendered:". Are there any references/uses of this algo (in HTML spec or elsewhere)? |