| 07:09 | <hsivonen> | TIL: Edge encodes the Yen sign as ASCII-state backslash instead of using the Roman state of ISO-2022-JP |
| 07:12 | <annevk> | hsivonen: do other browsers do that? |
| 07:14 | <hsivonen> | annevk: Chromium and Gecko use the Roman state like the spec says |
| 07:14 | <annevk> | hsivonen: that seems more sensible |
| 07:15 | <hsivonen> | annevk: when encountering Pile of Poo in the Roman state, Chromium behaves per spec (stays in the Roman state) and Gecko switches to the ASCII state for the NCR |
| 07:15 | <hsivonen> | Edge encodes overline as an NCR instead of switching to the Roman state |
| 07:16 | <hsivonen> | additionally, searching Duckduckgo for the overline character gives Emmanuel Macron as a related topic |
| 07:17 | <annevk> | Machine learning is a great thing |
| 07:22 | <hsivonen> | Well, I guess I'm not going to complain about the spec, since it matches Chromium. (Not switching from Roman to ASCII as part of encoder internals is quite a footgun, though, is someone wants to write replacement with \u escapes) |
| 07:22 | <hsivonen> | s/is someone/if someone/ |
| 07:22 | <hsivonen> | I guess I'm going to put that in documentation |
| 07:22 | <hsivonen> | encoding to non-UTF-8 is a bad idea anyway |
| 07:39 | <annevk> | hsivonen: I'd consider changing that if you think it's worth it |
| 07:40 | <annevk> | hsivonen: although it's not immediately obvious me how to change that around, might be non-trivial |
| 08:05 | <hsivonen> | annevk: let's not try to polish ISO-2022-JP for the benefit of people writing CSS/JS serializers for non-UTF-8 |
| 08:05 | <annevk> | hsivonen: will you review my patch today? |
| 08:05 | <hsivonen> | annevk: the ISO-2022-JP one? |
| 08:06 | <annevk> | hsivonen: yup |
| 08:06 | <hsivonen> | annevk: Implementing it right now. Will comment after implementation experience. |
| 08:06 | <annevk> | sg |
| 08:58 | <hsivonen> | I wonder what the history of the code point order for half-width katakana is. |
| 08:59 | <hsivonen> | like how did one of the sound marks not get grouped together with the other two |
| 08:59 | <hsivonen> | it's not combining, for one |
| 08:59 | hsivonen | knows nothing about the expected sort order |
| 09:05 | <annevk> | If only Unicode was developed in HTML and in version control |
| 09:05 | <annevk> | Would be much easier to answer those questions |
| 09:06 | <annevk> | hsivonen: do you mind if I change the ID for <dfn>index</dfn> btw? |
| 09:07 | <annevk> | If anyone depends on that working... |
| 09:07 | <hsivonen> | annevk: I don't recall having dependencies on that |
| 09:08 | <hsivonen> | annevk: the half-width order is older than Unicode |
| 09:08 | <hsivonen> | (full-width order, too) |
| 09:08 | <annevk> | Unicode could have changed the order though, no? |
| 09:09 | <annevk> | Or did you just import whole character sets wholesale, preserving order? |
| 09:09 | <annevk> | I don't actually know, though I always imagined the reason we have all these conversion tables is because they fiddled with things |
| 09:10 | <hsivonen> | annevk: it looks like Unicode imported ancient half-width katakana wholesale and then imported full-width katakana in the JIS X 0208 order |
| 09:16 | <hsivonen> | the annoying part about ordering is that Unicode, JIS X 0208 and GB 2312 all have sections that are supposed to be sorted first by radical and then by stroke count, but the Unicode order doesn't agree with the JIS X 0208 or GB 2312 orders for those bits anyway |
| 10:49 | <annevk> | Domenic: FWIW, I'm no longer convinced Window object tearoff objects are a good idea |
| 10:50 | <annevk> | Domenic: coupling everything to the Document object which isn't reused between different origins seems a lot safer |
| 12:10 | <Domenic> | annevk: it's crucial for customElements since it is intimately tied the per-Window prototypes/constructors. |
| 12:10 | <Domenic> | Maybe can change for others, although it seems strange. |
| 12:11 | <annevk> | It seems really weird that we'd reuse the same Window object across two origins |
| 12:11 | <Domenic> | I.e. It's unclear why window.history should change when window.History stays the same |
| 12:11 | <annevk> | I'm probably missing something |
| 12:12 | <Domenic> | Yeah I'm not thinking about it too hard today, on a vacation (and traveling tomorrow) |
| 13:46 | <wanderview> | Domenic: you are responding to github quickly for being on vacation :-) |
| 18:42 | <KP-> | does Ben Millard come here often? |
| 19:01 | <TabAtkins> | Don't know who that is, so... maybe? |
| 19:08 | <KP-> | https://projectcerbera.com/me/ |
| 19:08 | <KP-> | > BenMillard in Freenode |
| 19:08 | <KP-> | #whatwg |
| 19:26 | <hsivonen> | KP-: I don't recall seeing BenMillard here in the recent years. Not in the last 8 years, I'd estimate. |
| 19:29 | <KP-> | :O |
| 19:45 | <gsnedders> | Probably longer, I think. |
| 23:15 | <reggna> | How do you unset the src of an img, video, iframe, whatever? |
| 23:16 | <reggna> | x = new Image(); x.src = "foo" ; delete x.src ; x.src |
| 23:16 | <reggna> | Ah, x.removeAttribute("src") maybe.. |
| 23:19 | <reggna> | That seems to work. |
| 23:21 | <reggna> | x = new Image(); x.src = "foo" ; x.removeAttribute("src") ; x.src |
| 23:24 | <reggna> | That would have been an interesting #HTMLQuiz. |
| 23:27 | <reggna> | I'll do it in a few years when we've all forgotten the answer. |
| 23:55 | <jyasskin> | I'd like to have a set of guidelines for github issue metadata like labels, milestones, and projects. https://w3c.github.io/spec-labels.html had a start, so I've sent https://github.com/w3c/w3c.github.io/pull/27 to add milestones and projects. Do folks here have opinions? e.g. tobie, TabAtkins |