| 02:15 | <Dominic Farolino> | I guess we never figured out the WHATWG spec issue where the style sheets randomly fail to load huh? |
| 02:17 | <Dominic Farolino> | It just happened to me and I see a 'The FetchEvent for "https://resources.whatwg.org/standard.css"; resulted in a network error response: an "opaque" response was used for a request whose type is not no-cors' warning in DevTools (3 times for 3 different scripts) |
| 03:17 | <Domenic> | Yep. It honestly seems like a browser bug. There's no way that <link rel=stylesheet crossorigin> should ever result in an opaque response. |
| 07:31 | <annevk> | The only way that can happen is if there's some whatwg.org website where we link to that style sheet without crossorigin |
| 07:31 | <annevk> | Perhaps historical versions end up poisoning the cache? |
| 07:32 | <annevk> | One thing we haven't tried is renaming the style sheet and ensuring all new places that link to it have the attribute set, but then we have versioning :/ |
| 08:27 | <annevk> | dlrobertson: found some more issues, now also in the spec; out-of-range start/end values are normalized by the File API, but then result in an inaccurate content-range header |
| 08:27 | <annevk> | dlrobertson: they probably need to be normalized upfront; the default for end is also wrong, it should be total - 1 |
| 08:29 | <annevk> | I'll work on some patches |
| 10:15 | <Noam Rosenthal> | annevk, Domenic can either of you review https://github.com/whatwg/fetch/pull/1614? It's been pre-reviewed by Jeffrey Yasskin and others already thanks! |
| 13:09 | <dlrobertson> | annevk: Yeah it looks like we don't check the end value of the range. The start value we use for the content range comes from parsing a single-range-header, and if I'm reading things right would mean we'd return failure from the parser and return a network error in fetch |
| 13:10 | <dlrobertson> | Good catch! |
| 13:21 | <annevk> | dlrobertson: I would not have discovered this if I hadn't tried to implement it; that was a very humbling experience (and continues to be) |
| 13:22 | <dlrobertson> | hahaha I know the feeling all too well... I've technically picked up the bug to fix this in gecko |