Re: DELETE in WebDAV Collections

So far, it looks this way:

The cases we're concerned with in advanced collections are mostly ones where
there may be multiple bindings that eventually get you to the same
representation, but each goes through a separate resource.  So we've
mistakenly been drawing these as:

URI1       URI2     URI3
  |             |            |
  |             |            |
+------------------------+
|  Resource R           |
+------------------------+
                |
                |
+------------------------+
|  State                     |
+------------------------+

when we should have been drawing them as:

URI1                URI2                 URI3
  |                       |                       |
  |                       |                       |
+-------------+  +-----------------+  +------------------+
|Resource 1|   | Resource 2    |   | Resource 3     |
+-------------+  +-----------------+  +------------------+
       |                   |                         |
       |                   |                         |
+----------------------------------------------------------+
| State                                                               |
+----------------------------------------------------------+

So DELETE for these cases really does remove only one binding.  But there
could be cases like Roy's example of the content negotiation case where
DELETE would remove multiple bindings.  So it looks as if in order to be
sure of removing only one binding, we would have to define UNBIND.

Roy believes that the versioning case is *not* one where DELETE would end up
removing multiple bindings.  So DELETE would work fine for down-level
versioning clients when interpreted as deleting the resource.  (Confirm with
Geoff that this is true.)

Conclusion: DELETE deletes a resource, and we define a new UNBIND method for
removing a single binding.

Then what's left is the case where a client really wants to remove the state
from the Web by removing all the resources and all the bindings that provide
access to it.  Do we need to define a method for this (maybe that's what
we've been looking for with the proposed DESTROY method), and is advanced
collections the place to do that?

--Judy

Judith A. Slein
Xerox Corporation
jslein@crt.xerox.com
(716)422-5169
800 Phillips Road 105/50C
Webster, NY 14580

Received on Friday, 23 April 1999 12:02:06 UTC