21:26 | <TabAtkins> | zcorpan: ping on https://github.com/w3c/csswg-drafts/issues/9301 |
23:15 | <annevk> | TabAtkins: FWIW, it's really when you invoke https://url.spec.whatwg.org/#concept-url-parser that you don't pass a third argument |
23:19 | <TabAtkins> | ...oh, huh, I think I finally understand. I didn't realize URLs were, by definition, made of ASCII and percent-encoded bytes; I assumed that after parsing they were codepoints. |
23:20 | <annevk> | TabAtkins: they are code points, just all of them are ASCII :-) |
23:20 | <TabAtkins> | right, I mean non-ascii codepoints are turned into ascii codepoints via %-encoding, as part of the underlying value of the URL |
23:20 | <annevk> | TabAtkins: I kinda redundantly left that remark in the issue as well for other onlookers as well as that we're experimenting with this in WebKit |
23:20 | <TabAtkins> | rather than as a late "serialize this URL" step |
23:22 | <annevk> | Yeah makes sense. One of those things we might do differently if it were designed today. |
23:29 | <TabAtkins> | Hm, I might drop a PR to reword the note in the URL spec to make that clearer; I did not imagine it would be designed that way. |