- From: Jake Archibald <notifications@github.com>
- Date: Thu, 31 May 2018 15:36:03 +0000 (UTC)
- To: whatwg/fetch <fetch@noreply.github.com>
- Cc: Subscribed <subscribed@noreply.github.com>
- Message-ID: <whatwg/fetch/issues/747/393572169@github.com>
Ok, I have done research! https://github.com/jakearchibald/accept-encoding-range-test I took all the URLs from HTTP Archive's latest crawl that end in `.mp3` or `.mp4` and tested them with various `Accept-Encoding`s. **Accuracy warning:** A lot of these URLs seem to return different results each time. I think it can depend on whether the resource is cached at the CDN or not. Repo: https://github.com/jakearchibald/accept-encoding-range-test. Results: https://jakearchibald.github.io/accept-encoding-range-test/ (Chrome only due to transform streams). 3.21% of requests fail to return a 206 specifically if there's a non-identity encoding specified. 0.36% of requests fail to return a 206 specifically if the identity encoding is specified, or no encoding is provided. A few of these look like one-time errors, but some are reproducible. Based on that, it seems like Safari, Firefox, & Chrome are doing something right. If we spec any of them, it should be `Accept-Encoding: identity`. It's the shortest, and most specific. The repo includes a command to rerun the test for a particular URL if you want. @annevk Should this kind of weirdness be written into the spec? 3% is a lot. -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/whatwg/fetch/issues/747#issuecomment-393572169
Received on Thursday, 31 May 2018 15:36:31 UTC