Juergen Schoenwaelder wrote:
On Mon, Oct 16, 2006 at 12:06:38PM -0700, Andy Bierman wrote:
IMO we want the simplest approach possible.
Without specifying how you measure simplicity, your statement is
rather pointless.
This does not impact <filter type="xpath"> in PDUs.
It is a simple canonical format for identifying ONE instance of an
element.
It is not intended to be some sort of filtering tool.
A framework which allows to identify instances in some language X (and
uses this language for say error reporting and access control) and has
a required to implement a filter language which supports a filtering
language Y where the expressive power of Y is a true subset of X seems
a bit, lets say, surprising. Once again, I like X and I do believe Y
will be overcome anyway if netconf is going to be successful. (So I
better stop here since I believe the issue is pretty clear.)
We don't have 2 languages X and Y.
We have a constrained algorithm for using Xpath to represent
an Instance Identifier.
I didn't think we needed to discuss all the functional requirements
for an Instance Identifier, but let's try:
- non-ambiguous, deterministic, canonical format
- identifies a single instance in a static manner
(Note that all filter expressions can yield different results
based on when they are applied to the underlying datastore.
We want an Instance ID to have 0 or 1 matches
- An instance ID could be thought of as just an Xpath filter
that selects a single instance, based solely on the data model
naming components and element position.)
- persistence:
We would like an Instance ID 'match' at time T0 and later time T1
to represent the same conceptual instance. (Insert long debate
about discontinuity markers here.)
/js
Andy
--
to unsubscribe send a message to netconf-request@ops.ietf.org with
the word 'unsubscribe' in a single line as the message text body.
archive: <http://ops.ietf.org/lists/netconf/>