- From: Poul-Henning Kamp <phk@phk.freebsd.dk>
- Date: Mon, 30 Jul 2012 07:23:32 +0000
- To: "Adrien W. de Croy" <adrien@qbik.com>
- cc: "Larry Masinter" <masinter@adobe.com>, "ietf-http-wg@w3.org" <ietf-http-wg@w3.org>
In message <em704e4b8a-ca78-4787-810d-6b51e6587714@bombed>, "Adrien W. de Croy" writes: >Do we even need C-T if clients are sniffing anyway? Actually not. If we want maximum efficiency and lowest latency, the content should always use some kind of efficiently compressed format for content. Most A-V content do so already, and self-indentify as a result. In fact, about the only content which does not automatically compress is XML, HTML, CSS, ASCII and Javascript. Unfortunately the readily available general purpose compression standards only identify the compression algorithem and not the content in the compressed bitstream. But I think this would work: Remove the C-T header. Mandate that content be compresssed with a self-identifying compression. Mandate support for gzip compression of content, with an optional MIME type in the gzip "comment" header field. The important detail that would make this work in practice, is that GZIP has a no compression mode, where bytes are just framed by other bytes, so in the rare case of a precompressed but not self-identifying data type, gzip can provide the identification without the overhead of repeated compression. -- Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 phk@FreeBSD.ORG | TCP/IP since RFC 956 FreeBSD committer | BSD since 4.3-tahoe Never attribute to malice what can adequately be explained by incompetence.
Received on Monday, 30 July 2012 07:23:58 UTC