the approach (discussion topic for con call)

[James Tauber <JTauber@bowstreet.com>]

There seems to be some agreement and some open questions regarding the
approach I have recently been proposing, namely that DSML 2.0 should be
presented as (at least):

1. an abstract data model based on section 3.2 of RFC2251 (and probably RDF
too)
2. an abstract API for querying and manipulating anything that follows the
abstract data model
3. an XML serialization of the abstract data model (possibly but not
necessarily based on DSML 1.0)
4. a mapping of the abstract API to actual LDAPv3 operations
5. a binding of SOAP calls to the abstract API
6. possible other bindings

I would imagine proceeding in roughly the above order. One advantage of this
is that it delays decisions on the more debatable topics like surface syntax
of the operations.

I believe that each of the approaches we've touched on (SOAP, Andy
Harjanto's recent post, Chris's approach outlined on a couple of calls)
would actually agree on 1-4 and only diverge on the last two points.

I know a number of people feel a decision would be easier to make if they
had some examples, so I suggest that over the next couple of weeks we
produce:

1. a draft abstract data model (I will endeavour to do this today as it
won't be long)
2. a draft list of operations described in terms of the data model

It doesn't look like I'm going to be able to make the call today so I hope
that this provides enough fuel for discussion.

James