- From: Mark Nottingham <mnot@mnot.net>
- Date: Wed, 26 Feb 2014 07:37:34 +1100
- To: Jesse Wilson <jesse@swank.ca>
- Cc: Jeff Pinner <jpinner@twitter.com>, Zhong Yu <zhong.j.yu@gmail.com>, Roberto Peon <grmocg@gmail.com>, Patrick McManus <mcmanus@ducksong.com>, HTTP Working Group <ietf-http-wg@w3.org>, Martin Thomson <martin.thomson@gmail.com>
We clarified this in HTTPbis - see <http://tools.ietf.org/html/draft-ietf-httpbis-p1-messaging-26#section-4.2.2>. The downside of only doing gzip is that if a server *does* use deflate, a HTTP1->2 intermediary will have to re-encode responses. Not saying that's a showstopper, just noting it. On 26 Feb 2014, at 1:32 am, Jesse Wilson <jesse@swank.ca> wrote: > In Billy Hoffman's blog[1], and elsewhere, there's discussion over the fact that 'deflate' is used to describe two different algorithms. And several browsers and webservers got it wrong. > > "So, DEFLATE, and Content-Encoding: deflate, actually means the response body is composed of the zlib format (zlib header, deflated data, and a checksum)." > > [1]: http://zoompf.com/blog/2012/02/lose-the-wait-http-compression > > If the spec is going to require that clients support deflate compression, it should use strong language to remind implementors that 'deflate' means zlib. > > And if 'deflate' is implicit, then middleboxes that bridge HTTP/2 clients to HTTP/1.1 servers will need to transcode/decompress deflated data for the legacy browsers[2] that misinterpret the spec. > > [2]: http://www.vervestudios.co/projects/compression-tests/results > > Plus, middleboxes that bridge HTTP/1.1 servers to HTTP/2 clients may need to transcode/decompress deflated data from broken servers that also get the APEC wrong. > > Otherwise all HTTP/2 clients will need to use heuristics to guess which algorithm the peer is using. From Sam Saffron on Stack Overflow[3]: > > "So, over the years browsers started implementing a fuzzy logic deflate implementation, they try for zlib header and adler checksum, if that fails they try for payload." > > [3]: http://stackoverflow.com/questions/388595/why-use-deflate-instead-of-gzip-for-text-files-served-by-apache > > My preference is to avoid this rats nest and require gzip only. > -- Mark Nottingham http://www.mnot.net/
Received on Tuesday, 25 February 2014 20:38:09 UTC