- From: Boris Zbarsky <bzbarsky@MIT.EDU>
- Date: Tue, 30 Jun 2009 17:01:19 -0700
- To: Ian Hickson <ian@hixie.ch>
- CC: Anne van Kesteren <annevk@opera.com>, robert@ocallahan.org, David Singer <singer@apple.com>, "ietf-http-wg@w3.org" <ietf-http-wg@w3.org>, public-html@w3.org
Ian Hickson wrote: > On Tue, 30 Jun 2009, Boris Zbarsky wrote: >> Note that Gecko, at least, sniffs all HTTP responses without a >> Content-Type header, and after that treats them all effectively as >> metadata-provided. It sounds like doing that is fundamentally >> incompatible with this specification, though, unless we teach our >> sniffer to recognize video/audio types by data, not just by extension... > > Generally I would encourage you to sniff based on data and not extension > for anything sent over HTTP anyway. In that case we'd typically sniff the videos as application/octet-stream. Which raises an interesting question. If a <video> points to data that has Content-Type metadata that says it's application/octet-stream.... I'd say that the UA should ignore that and look at the actual data just like it would if the MIME type is not set. -Boris
Received on Wednesday, 1 July 2009 00:02:34 UTC