- From: Roy T. Fielding <fielding@gbiv.com>
- Date: Thu, 15 Jul 2010 12:05:12 -0700
- To: Willy Tarreau <w@1wt.eu>
- Cc: Adrien de Croy <adrien@qbik.com>, HTTP Working Group <ietf-http-wg@w3.org>
On Jul 14, 2010, at 9:31 PM, Willy Tarreau wrote: > On Wed, Jul 14, 2010 at 07:53:37PM -0700, Roy T. Fielding wrote: >> On Jul 14, 2010, at 6:50 PM, Adrien de Croy wrote: >>> All entity headers are end to end are they not? >> >> Anything in the spec that says something is end-to-end or >> hop-by-hop will be rewritten soon. Fields are hop-by-hop >> when they are listed in Connection (or are Connection). >> All others are end-to-end. > > So probably the spec should enumerate the well-known headers > that cannot be hop-by-hop and that must not be listed in > Connection. At the moment the spec only says "All other > headers defined by HTTP/1.1 are end-to-end headers", which > is not much practical for implementers, considering the size > of the spec and the high risk of missing a few ones. > > And that leaves the question open about what to do if those > headers are found. The worst cases I thought about (Host, > Content-Length) don't work with Apache at least, but they > may cause real trouble on proxies which blindly apply the > spec. They should not cause any problems, at least no more so than receiving invalid messages. If the software can't handle weird things then it doesn't belong on the Internet. We don't need to protect implementations from their own stupidity. It is far better for the protocol to be self-descriptive, so that we can extend it over time. ....Roy
Received on Thursday, 15 July 2010 19:05:42 UTC