| 07:12 | <annevk> | Domenic hsivonen: I've now rebased https://github.com/whatwg/url/pull/905 and this changes the IDNA section a bit more. Notably the domain parser now handles all validation errors upfront (so we always report beStrict=true validation errors, despite parsing as beStrict=false), making the remainder of the code easier to read. There's no longer an as-nice early return for beStrict=true, but I've tried to compensate for it with the strictResult variable name. |
| 07:15 | <annevk> | Hopefully that's the last change in this area for a while. |
| 08:19 | annevk | wonders if nicolo-ribaudo wants to become editor of Web IDL |
| 08:21 | <Ms2ger> | No |
| 08:21 | <annevk> | Meanwhile, maybe Ms2ger can stamp https://github.com/whatwg/webidl/pull/1612 then |
| 08:21 | <Ms2ger> | Walked straight into that one |
| 08:26 | <Ms2ger> | Which reminds me, I was trying to find the .length of an async iterator's return method recently and didn't find anything that was completely convincing |
| 08:27 | <Ms2ger> | Oh, it's clear now. Did you change that recently? |
| 08:29 | <nicolo-ribaudo> | I was actually looking at WebIDL and thinking that all the recently updated PRs would benefit from me reviewing them 😅 |
| 08:30 | <Ms2ger> | Perhaps Apple would benefit from freeing up some budget for contracting out WebIDL work :) |
| 08:32 | <nicolo-ribaudo> | I have a question about Exposed/Global. Some interfaces are exposed in Worker, while others in (DedicatedWorker,SharedWorker) |
| 08:32 | <nicolo-ribaudo> | Is there any difference? |
| 08:32 | <Ms2ger> | Service workers? |
| 08:33 | <nicolo-ribaudo> | Oh I didn't realize that ServiceWorker is a Worker and not something special |
| 08:33 | <nicolo-ribaudo> | Thanks :) |
| 08:34 | <annevk> | They are also something special 😛 |
| 08:34 | <Ms2ger> | Was about to make a comment to that effect :) |
| 08:42 | <annevk> | smaug: did you see https://github.com/whatwg/html/pull/10995#discussion_r3353691486? |
| 08:47 | <annevk> | Hmm, AI still finds many trivial issues in that PR. 🙃 |
| 08:49 | <smaug> | I would expect source to set the internal relatedTarget, but let me check quickly what browsers do with FocusEvent as an example |
| 09:05 | <annevk> | Really need some kind of constant "what is the precedent?" reminder somewhere. Art Barstow was good at that. |
| 09:07 | <smaug> | AI is good at writing simple tests 🙂 |
| 09:08 | <smaug> | Now, we need a tool for github issues where we can just select a question from the issue/pr and the tool would create a testcase and run it in different browsers |
| 09:10 | <annevk> | I guess we really need to fix event constructors. Perhaps we should do a thing where you can use an IDL attribute to get the default behavior, but if you want custom logic (such as setting relatedTarget) you need to write (part of) it out. |
| 09:28 | <Jake Archibald> | annevk: I updated https://github.com/whatwg/fetch/issues/1936#issuecomment-4802023793 with something that wasn't clear from the WHATNOT meeting. Interested in your thoughts. |
| 09:37 | <Luke Warlow> | Should we remove the Twitter updates link from the html spec? It's not twitter anymore and the account has posted in 3 years? |
| 10:03 | <hsivonen> | annevk: With the changes, the text still says "A domain is a non-empty ASCII string that identifies a realm within a network.", which doesn't make it look like a domain is a strongly-typed output concept of an algorithm and string-ish implementation-wise. The definition still looks vague of whether an input string as a domain. |
| 10:05 | <hsivonen> | I don't particularly like having to work out the correct as-iffing for implementing the browser display stuff without logically unnecessary passes over the string, but it's permitted for specs to leave as-if exercises like this to implementors. |
| 10:10 | <annevk> | Luke Warlow: I don't object to removing it, but then we should remove it everywhere. |
| 10:16 | <annevk> | Jake Archibald: I was thinking that we'd require the crossorigin attribute which means option 1. And yeah, I guess that would allow for cleaning up some of the Range logic, especially the no-CORS logic, though we still need to set the header and such. |
| 10:22 | <annevk> | hsivonen: I agree it's a bit vague. I haven't really thought of anything better thus far. I suppose we could make it a struct that has a value that holds the underlying string and then define a serializer. But that seems a lot less elegant. |
| 14:20 | <Luke Warlow> | https://github.com/whatwg/html/pull/12623 - HTML PR and I've also made a PR to remove it from the bikeshed boilerplate which should cover the rest of the specs. I guess there's probably an instruction file somewhere that needs updating to no longer go through the twitter registration process? |
| 15:04 | <annevk> | Yeah in whatwg/sg. And we have to remove some code in whatwg/participate.whatwg.org |
| 21:42 | <smaug> | Shouldn't https://html.spec.whatwg.org/#process-internal-resource-link step 3 use exclude fragments true? Noam Rosenthal |