- From: Bjoern Hoehrmann <derhoermi@gmx.net>
- Date: Tue, 25 Feb 2014 18:33:29 +0100
- To: Cory Benfield <cory@lukasa.co.uk>
- Cc: Roberto Peon <grmocg@gmail.com>, HTTP Working Group <ietf-http-wg@w3.org>
* Cory Benfield wrote: >That's not entirely true. The difference between HTTP/1.1 and HTTP/2.0 >is that in HTTP/1.1 if I send `Accept-Encoding: identity', the server >must not return me a compressed response. In HTTP/2.0, however, the >server may always return a response compressed with gzip or deflate. >This means all HTTP/2.0 user-agents must allow decompression of gzip >and deflate encoded responses, because we cannot reliably inform >upstream that we don't want them. > >Jesse and I want to step that back to say that servers may always >return a response compressed with gzip only: any other compression >must be explicitly requested. In this sitatuion, if I then send >`Accept-Encoding: gzip' I'll only ever get gzip back, even if the >origin of the entity body is a HTTP/1.1 host. This would allow me to >rip my deflate/zlib code out as I can ensure that it won't be sent to >me. The Accept-Encoding header indicates preferences and honouring them is a SHOULD, not a MUST. It is not unusual to ignore it, e.g. typical Apache configurations would send .gz files with `Content-Encoding: gzip` and would not fail content negotiation and would not decompress the file automatically when the client is not prepared to accept `gzip`, and some deliberate configure it this way, e.g. to conserve resources. -- Björn Höhrmann · mailto:bjoern@hoehrmann.de · http://bjoern.hoehrmann.de Am Badedeich 7 · Telefon: +49(0)160/4415681 · http://www.bjoernsworld.de 25899 Dagebüll · PGP Pub. KeyID: 0xA4357E78 · http://www.websitedev.de/
Received on Tuesday, 25 February 2014 17:34:09 UTC