| 03:40 | <zoya> | https://mimesniff.spec.whatwg.org/#signature-for-webm Hello, I've been trying to rewrite the webm sniffing algorithm to js and encountering some weirdness. My first question was if anyone here has already implemented this for something so I can see how other people are interpreting it, and second, does anyone else agree that the wording is unclear? Especially within the "matching a padded sequence" algo, the phrase "...eventually preceded by bytes..." doesn't quite make sense. Do they mean "...evenly preceded by bytes", or perhaps "eventually followed by bytes"?. Also, the variable "end" in italics is mentioned nowhere else on the page. I'm assuming it's referring to the end of the sequence. Finally, I've been testing it and found that removing step 5 in the nested while loop ("if iter < length - 4...") will yield a correct result, but if that step is enabled, it will not detect some sample webm videos I've been testing it with |
| 03:42 | <zoya> | Also, I've been working on doing the same thing with rewriting the tagless mp3 detection algorithm in JS and it seems broken too... yeah |
| 06:26 | <annevk> | I think padenot worked on that, but I don't see him here. Maybe file issues zoya as a first step? |
| 20:47 | <Jake Archibald> | Is the behaviour of `input.value` considered good, or if we could do it all again, would simpler attribute reflection be preferred? |
| 20:56 | <Jake Archibald> | I'm referring to how `input.value` initially gets the attribute, but then switches to getting some internal value when the user interacts or `input.value` is set. Or would it be better to do something like `<details>`, where interaction is reflected to the attribute value? |
| 20:58 | <bkardell> | I think it would have been better to have an initialValue and no reflecting like that |
| 21:53 | <TabAtkins> | Yeah, the magic reflection is weird. |