W3C home > Mailing lists > Public > ietf-http-wg-old@w3.org > May to August 1998

RE: ISSUE: Protection space

From: Paul Leach <paulle@microsoft.com>
Date: Fri, 7 Aug 1998 11:17:15 -0700
Message-Id: <CB6657D3A5E0D111A97700805FFE65875D743A@red-msg-51.dns.microsoft.com>
To: 'Dave Kristol' <dmk@bell-labs.com>
Cc: http-wg@hplb.hpl.hp.com

> -----Original Message-----
> From: Dave Kristol [mailto:dmk@bell-labs.com]
> Sent: Friday, August 07, 1998 11:04 AM
> To: Paul Leach
> Cc: http-wg@hplb.hpl.hp.com
> Subject: Re: ISSUE: Protection space
> 
> 
> Paul Leach wrote:
> > [...]
> > > [DMK]
> > > Since all URLs on a server are implicitly
> > > descended from "/"
> > > (no?), wouldn't it be easier just to say that relative URLs
> > > are taken to
> > > be relative to "/"?
> > 
> > The list allows absolute URIs with host names other than 
> that of the server
> > sending the "domain" directive.
> 
> Are we talking about two different things?

Maybe. See next comment.

  I'm not concerned with
> absolute URLs.  For them the protected set of URLs is obvious.

I think of two kinds of "relative URLs"  -- "dir/foo.html" and
"/dir/foo.html". The latter is relative to (e.g.) http://www.xxx.com, the
former to the URL of page in which it appears (typically). I don't think the
former belong in a domain list. 

> Here's the wording at issue (Sect. 3.2.1):
> If a URI is relative, it is relative to [the] canonical root 
> URL of the
> server being accessed.
> 
> My notion of a relative URL is one that does not begin with '/'.  For
> such a URL, wouldn't it make sense to give them an implicit 
> '/' prefix?

How about I say that URI in "domain=URI..." must be an "http_UTL" or
"abs_path" as defined in section 3.2.2 of the HTTP/1.1 spec? 
The former is the usual "http://www.xxx.com:port/dir/foo.html" type; the
latter is "/dir/foo.html".

Paul
Received on Friday, 7 August 1998 11:18:58 EDT

This archive was generated by hypermail pre-2.1.9 : Wednesday, 24 September 2003 06:33:20 EDT