| 09:53 | <sideshowbarker> | I somehow only recently realized that the ES spec rightly uses real typographical (“curly”) quotation marks — rather than the ASCII character. |
| 09:53 | <sideshowbarker> | I wish all other specs would follow that example. |
| 09:55 | <sideshowbarker> |
|
| 09:57 | <nicolo-ribaudo> | The only downside is that it makes typing spec text more annoying 😅 |
| 09:58 | <sideshowbarker> | It’s a good learning experience for everybody 😀 |
| 09:58 | <nicolo-ribaudo> | Btw, you only noticed it now because we only did that change a few months ago! https://github.com/tc39/ecma262/pull/3861 |
| 09:58 | <sideshowbarker> | aha! |
| 09:58 | <sideshowbarker> | bravo Richard Gibson |
| 09:59 | <sideshowbarker> | I noticed because I was looking at verbatim spec text copied from the spec into a C++ code comment. |
| 09:59 | <sideshowbarker> | And the copy/paste preserved the typographical quotes. |
| 10:00 | <sideshowbarker> | …which I think is also great — because in 2026 there’s no good reason for all of us to be avoiding the use of non-ASCII text in code comments |
| 10:02 | <sideshowbarker> | It’s a great test for any broken tooling that might be processing the source files in some way — and for people’s IDEs and whatever. |
| 10:03 | <sideshowbarker> | Anyway, I wish the ES spec would also go further and also use real typographical apostrophes, rather than the ASCII thing. |
| 10:04 | <sideshowbarker> | And real em dashes too, while we’re at it (if it’s not already). |
| 10:23 | <annevk> | Shannon Booth: when I reviewed your javascript: URL, AI found more, as it does. I created https://github.com/whatwg/html/pull/12978 to address that. Would appreciate it if you could take a look when you have time. |
| 10:24 | <annevk> | sideshowbarker: I would take patches for WHATWG specifications to fix their typography. I thought we mostly did this correctly, but maybe we stopped caring. We do use straight quotes to delimit strings though. |
| 10:25 | <sideshowbarker> | Yeah, using straight quotes for string literals makes sense. |
| 10:30 | <Shannon Booth> | will take a look after work! we indeed also have a workaround for the null client case, i didnt research yet whatever client we picked actually matches other implementations. and yeah the spec also doesnt match wpt for how load event works, so all of those changes sound positive |
| 11:14 | <zcorpan> | sideshowbarker: thoughts on https://github.com/whatwg/html/issues/12951 ? |
| 11:16 | <sideshowbarker> | No strong opinion from the POV of conformance checking. But otherwise, last I understood, no implementors were interested (due to the perf issue). |
| 11:17 | <sideshowbarker> | Or wait, implementations already have to handle this, right? |
| 11:17 | <sideshowbarker> | So I guess the “performance issue” problem is, do we want to encourage author-developers to do something that’s gonna result in bad performance. |
| 11:19 | <sideshowbarker> | So then my opinion from the POV of conformance checking is: If it’s bad for performance, then we are right to continue disallowing it (and for the HTML checker to report it). |
| 11:21 | <sideshowbarker> | https://validator.w3.org/nu/about.html#why-validate |
| 11:21 | <sideshowbarker> |
|
| 11:25 | <sideshowbarker> | https://html.spec.whatwg.org/multipage/introduction.html#:~:text=Errors%20that%20result%20in%20disproportionately%20poor%20performance |
| 12:00 | <zcorpan> | sideshowbarker: with the requirement that a selector that matches anything before the style element's parent in tree order is "should not" or "must not", there's no perf issue |
| 12:26 | <sideshowbarker> | OK, so I could make that checker check specifically for that, and not report any message in that case. |
| 12:27 | <sideshowbarker> | I now vaguely recall that I may have already made it do something like that |
| 15:44 | <Richard Gibson> | not if you continue to type ASCII quotes and rely upon the format script to convert them |
| 16:38 | <bakkot> | I've been looking more seriously at how we can use AbortController in JS specs; we want to have cancelable APIs and I really don't want to have a different cancelation mechanism in the language. This would be a bit weird because it would basically require adopting a subset of the DOM spec, but could hopefully be done without requiring changes in implementations (except we need a replacement for would appreciate any thoughts and feedback on https://docs.google.com/presentation/d/1iRpJ3ADc50ZRUsmwYi9Yp3qjixBq3wMWWx_jw7cFleU/edit |
| 18:03 | <Shannon Booth> | I found an issue while taking a look, raised https://github.com/whatwg/html/pull/12981 (initialInsertion is the thing I mentioned about spec not matching WPT) |
| 18:14 | <Shannon Booth> | (I am not 100% sure in those changes by the way, but it's definitely not aligned) |