W3C home > Mailing lists > Public > w3c-dist-auth@w3.org > July to September 1998

RE: Hierarchical URLs and Collections

From: Jim Davis <jdavis@parc.xerox.com>
Date: Tue, 18 Aug 1998 17:08:08 PDT
Message-Id: <3.0.3.32.19980818170808.007c9100@mailback.parc.xerox.com>
To: "WebDAV" <w3c-dist-auth@w3.org>
Cc: "Larry Masinter" <masinter@parc.xerox.com>
At 01:01 PM 8/18/98 PDT, Larry Masinter wrote:
> [I suggest] that there
>be a DAV:parents property and that interoperable clients ... use
> PROPFIND of DAV:parents in order
>to walk up the hierarchy, rather than just trying to parse the URL
>tree. It's simple, backward compatible, doesn't cost much in terms of
>server implementation, and much more resiliant.

This proposal resembles some proposals that were discussed in the context
of so-called 'strong' references, and which were found problematic then.

1) for the common case (internal membership) it adds no value, because one
can find the parent container by mere syntax.

2) for the uncommon case (referential membership), it's problematic, because:
 1. Security: When I create a referential resource whose target is another
resource I may not have the access to the target resource to set the back
pointer.  
 2. Privacy: I may not want the target to know I am linking to it
 3. Scalability: If the target is very very popular it won't scale.  

Also, I don't see that it addresses the actual concerns raised by the DMS
community, as opposed to those that came up as side issues in the long
discussion.  In my opinion, referential resources suffice to completely and
fully support such DMS systems.

Furthermore, if it turns out that supporting such a property is a good
idea, then I suggest the right thing is for the DMS vendor to implement it,
gain some experience, then propose it for a future version of WebDAV.  In
fact, we should look closely at the DMA API, which defines two models of
containment, Direct and Referential.  WebDAV has the former, and your
proposal is like the latter.  To support Referential containment requires
general graph manipulation (as Y. Goland pointed out) and/or transactions
(which DMA has, sort-of).

But anyway, unless someone can show why referential resources do not solve
the DMS vendor's problems fully, we should close this issue.  

Jim




------------------------------------
http://www.parc.xerox.com/jdavis/
650-812-4301
Received on Tuesday, 18 August 1998 20:08:48 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 2 June 2009 18:43:47 GMT