W3C home > Mailing lists > Public > w3c-dist-auth@w3.org > October to December 2005

Re: Proposal for response URLs in Multi-Status: always based on Request-URI

From: Julian Reschke <julian.reschke@gmx.de>
Date: Mon, 05 Dec 2005 01:26:05 +0100
Message-ID: <4393891D.2090504@gmx.de>
To: Lisa Dusseault <lisa@osafoundation.org>
CC: WebDAV WG <w3c-dist-auth@w3.org>

Lisa Dusseault wrote:
> We've discussed the allowability of relative URLs in the 'href' element 

I think we should stick with proper terminology: there is no such thing 
as a relative URL, RFC3986 calls these things "relative references".

> in the 'response' elements of Multi-Status responses.  I'll call each of 
> these a "response URL" for now.  I believe our conclusion was that in 
> response to PROPFIND, response URLs which are relative URLs MUST be 
> relative to the Request-URI, and those which are absolute MUST begin 
> with the Request-URI (exactly the same scheme, host, port and path).


> For MOVE and COPY, one could consider relative URLs as being resolved 
> against the Destination header instead of the Request-URI, but I don't 
> believe that anybody does this.  One could also imagine seeing absolute 
> URLs that were part of the Destination namespace rather than the 
> Request-URI namespace, but again, I don't believe that anybody does this.
> Thus, are there any objections if we treat all Multi-Status responses 
> the same way -- for MOVE and COPY as well as PROPFIND and PROPATCH?  
> That the response URLs MUST always be in the Request-URI namespace, and 
> if relative, be resolved against the Request-URI?

No. Of course a multistatus upon COPY/MOVE can contain URLs below the 
destination URI, not the Request-URI. And other methods such as REPORT 
and SEARCH may return URLs completely independently of the Request-URI.

But yes, if an <href> is a relative reference, it is always relative to 
the Request-URI. That was certainly the consensus each time this was 

> Would making this requirement a MUST for all Multi-Status responses 
> break any extensions using Multi-Status -- or do we limit the 
> requirement to Multi-Status responses to methods defined in RFC2518bis 
> only?

See above, and please review 
(you do remember that we agreed on the conference call that I should 
make a proposal for spec text, right?).

Best regards, Julian
Received on Monday, 5 December 2005 00:27:21 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 15:01:34 UTC