W3C home > Mailing lists > Public > ietf-http-wg-old@w3.org > January to April 1995

Re: Getting full URI to the server

From: Brian Behlendorf <brian@wired.com>
Date: Mon, 13 Feb 1995 11:54:44 -0800 (PST)
To: "Roy T. Fielding" <fielding@avron.ICS.UCI.EDU>
Cc: http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
Message-Id: <Pine.BSD.3.91.950213114718.18272D-100000@get.wired.com>
On Mon, 13 Feb 1995, Roy T. Fielding wrote:
> So, which one should we include in the 1.1 specification:
> 
>    Host: foo.bar.com
> 
> or
> 
>    Orig-URI: http//foo.bar.com:8001/home/is/where/the/wallet/is.html

Host:, as it's the only unique part of the request, and we should 
definitely try to keep the amount of stuff in the request headers down.

> Also, should it be:
> 
>    a) recommended for all requests
>    b) recommended only for requests to standard URLs like / and /site.idx

It should be recommended for all requests - otherwise collections of HTML 
pages will be nowhere nearly as portable.  

However, one way to reduce the number of times it has to stick Host: into 
the headers would be to do a reverse-lookup of the IP number associated 
with that Host, and if it returned the same hostname, then that must be 
the canonical hostname for that IP number and thus no Host: header is 
needed (presuming providers don't do something stupid like have multiple 
reverse DNS tables for the same IP numbers :)  Just a thought.
 
	Brian

--=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=--
brian@hotwired.com  brian@hyperreal.com  http://www.hotwired.com/Staff/brian/
Received on Monday, 13 February 1995 12:26:51 EST

This archive was generated by hypermail pre-2.1.9 : Wednesday, 24 September 2003 06:31:14 EDT