- From: Poul-Henning Kamp <phk@phk.freebsd.dk>
- Date: Thu, 20 Feb 2014 20:04:50 +0000
- To: Bjoern Hoehrmann <derhoermi@gmx.net>
- cc: Martin Thomson <martin.thomson@gmail.com>, HTTP Working Group <ietf-http-wg@w3.org>
In message <dflcg9d9ugmc1m40cjca9npmnf84gel21j@hive.bjoern.hoehrmann.de>, Bjoer n Hoehrmann writes: >The gzip format is a container format and the container includes data >like filename, operating system, checksums, timestamps, comments, and >a DEFLATE stream with the compressed data. The overhead is not really >useful in this context, [...] I disagree. The overhead is minute, for instance the filename is optional and seldomly used in HTTP context. The CRC32 it brings is a very good integrity check which suprisingly often uncovers trouble. Gzip further has the advantage that there are plenty of command line tools available for it, whereas deflate is quite hard to produce with regular tools. After considerable research, I decided to only support gzip in Varnish, and I have only ever had one user ask for deflate, and that was based on somebody raising the same bogus "overhead" argument as you do above. If anything deflate should die. -- 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 Thursday, 20 February 2014 20:05:15 UTC