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

RE: Client Option for Strong References

From: Slein, Judith A <JSlein@crt.xerox.com>
Date: Thu, 8 Oct 1998 11:14:38 -0400
Message-ID: <201BB34B3A73D1118C1F00805F1582E8B76C9C@x-wb-0128-nt8.wrc.xerox.com>
To: "'WebDAV'" <w3c-dist-auth@w3.org>
I have no problem with adding a requirement that a client be able to
request, when creating a reference, that referential integrity *not* be
enforced for that reference.

We already have a requirement that a client be able to request, when
creating a reference, that referential integrity be enforced for that
reference.  The proposed requirement is just the mirror of this existing
requirement.

The scenarios described in support of the new requirement seem reasonable.

However, I would like us to stand by our earlier position that we will not
implement anything related to referential integrity in the first release of
the Advanced Collections protocol.  

I think that the debate about the proposed requirement has been confused
with rehashing the discussions about the difficulties of implementing strong
references.  The reason for this is probably that we are also involved in
trying to add direct references to the protocol spec and in discussing
whether to add backpointers.

I contend that the weak / strong distinction is orthogonal to the redirect /
direct distinction, and that our decision to include direct references in
the protocol should not push us to do anything about referential integrity.

What about backpointers? Can we introduce backpointers without implementing
strong references?  The integrity of the backpointers has to be maintained
in order for them to be useful at all.  But that's still a separate issue
from strong references.  A server could maintain backpointers on target
resources, but not attempt to insure that references are always valid.

So I'd just like to separate the 3 debates about strong / weak references,
direct / redirect references, and backpointers.  And also to separate
questions about whether there should be a requirement for any one of these
from questions about implementation.  Then I think it's clearer that we can
support the proposed requirement that a client be able to request, when
creating a reference, that referential integrity *not* be enforced for that
reference.

Judith A. Slein
Xerox Corporation
jslein@crt.xerox.com
(716)422-5169
800 Phillips Road 105/50C
Webster, NY 14580
Received on Thursday, 8 October 1998 11:14:19 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 2 June 2009 18:43:48 GMT