- From: Dave Kristol <dmk@allegra.att.com>
- Date: Fri, 26 Apr 96 09:42:03 EDT
- To: fielding@avron.ICS.UCI.EDU
- Cc: http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
"Roy T. Fielding" <fielding@avron.ICS.UCI.EDU> wrote: > > [DMK] > > I said "particularly for an origin server". I believe an *origin > > server* sees either a relativeURI or an absoluteURI in the form of an > > http_URL on the request line. I'm trying to nail down what an origin > > server should expect to see. It should be specified somewhere. > > No, it doesn't need to be specified anywhere beyond the syntax > limitations already specified. The server is capable of controlling > its own namespace, period. The protocol does not make any additional > limitations where none are needed, particularly when they would hinder > use of HTTP outside the typical Internet browser realm. Okay, I'll accept your arguments. However, there is still a piece I feel is missing: 1) We want clients to send, and servers to accept, an absoluteURI. 2) Eventually we want to obviate the need for Host:. 3) The purpose of absoluteURI and/or Host is to specify the virtual host for (at least) an http_URL. The reason I've been trying to restrict the absoluteURI syntax is that then it becomes easy to identify the virtual host. I believe somewhere in the spec. there should be an explanation of how to identify the virtual host of "the resource identified by the request". Dave Kristol
Received on Friday, 26 April 1996 06:57:25 UTC