RE: ORDERPATCH in the Advanced Collections Specification

> At IETF 42, the question was raised:  What is the rationale for allowing
> only one change to an ordering per ORDERPATCH request?
>
> I believe that the rationale for this decision was that there may be
> existing DMSs that support only single changes per request, and that we
> should not require them to change in order to be compliant with the
> Ordering part of the Advanced Collections specification.

That these systems don't natively support multiple order changes per request
isn't necessarily sufficient reason to prevent multiple requests from being
submitted in an ORDERPATCH.  Since the semantics of ORDERPATCH are being
implemented by some DAV layer, rather than directly by the document
management system, it is quite possible that the DAV layer could implement
multiple order changes per ORDERPATCH even if the document management system
can't perform this functionality in a single operation.

In fact, since WebDAV servers support locking a collection (to implement
LOCK), a DAV layer could lock the collection, then perform multiple
individual order change operations, then unlock the collection.  It might be
difficult to implement all-or-nothing semantics with this approach (i.e.,
perform all these order changes or perform none of the order changes), which
is why I suggest making the semantics best effort.

- Jim

Received on Wednesday, 16 September 1998 14:04:17 UTC