Re: draft-ietf-boreham-numsubordinates-00.txt

[dboreham@netscape.com (David Boreham)]

Bruce Greenblatt wrote:
> 
> I would like for it to be the case that if a server supports this
> attribute, then it MUST support it for all entries in the DIT on that
> server.  Is there a problem with adding this conformance requirement?

I flipped and flopped back and forth on this one.
Our implementation _does_ support the attribute 
on every entry in the DIT, but I wondered whether
this might be too tight a restriction.
One reason being that some naming context
the server hosts might be implemented in a
database which can't support the count
efficiently. Another reason would be that
the DIT is distributed between multiple
servers, and hence some levels in the tree
might have the attribute and others might not.

So I don't like the idea of specifying
that any entry returned by a server must
have the attribute.

I also don't like the idea that clients
can't be sure whether it's supported
or not on a given entry. e.g. Clients performing
searches of the form "numSubordinates>=1"
ought to get sane behavior from a server.