- From: Roy T. Fielding <fielding@liege.ICS.UCI.EDU>
- Date: Wed, 31 Jul 1996 21:50:32 -0700
- To: http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
I don't consider anything in the HTTP specification to be optional
reading. Any implemenenter of any client or server software, cache or
no cache, must understand the entire document in order to implement
HTTP correctly. I know this may be difficult to do, but it is certainly
possible to do and a hell of a lot easier than fixing the constant
stream of bugs generated by people who don't.
The reason HTTP/1.1 is much bigger (in specification) than HTTP/1.0
is simply because HTTP/1.0 does not work in the presence of caches
(UA or proxy or server-based) or proxies. It may seem like it works,
simply because they do exist, but the vast majority of perceived transfer
problems with the Web are due to the interaction between caching
and intermediary applications that don't behave like browsers, and
browser implementions that can't even parse the basic message syntax
let alone determine if a message has been truncated in transit.
All implementers must be aware of those interactions, regardless of what
they are implementing with HTTP, because they won't know what they need
to implement (and why) until they do understand all those interactions.
...Roy T. Fielding
Department of Information & Computer Science (fielding@ics.uci.edu)
University of California, Irvine, CA 92697-3425 fax:+1(714)824-4056
http://www.ics.uci.edu/~fielding/
Received on Wednesday, 31 July 1996 22:15:15 UTC