Re: subschema semantics (Was: Models: Schema references to undefined entities)

[Hallvard B Furuseth <h.b.furuseth@usit.uio.no>]

Kurt D. Zeilenga writes:
> I assume you are asking to revert the I-D to "servers publish
> schema what they support" semantics.

No, this part of 4.4 (Subschema Discovery), which disappeared at the
same time.  Sorry, I should have tracked it down before posting:

  Clients SHOULD NOT assume that server supports all referenced elements
  of a particular definition.  For example, a client is not to assume
  the server supports the EQUALITY matching rule of a listed attribute
  unless the server publishes a definition for that matching rule.

It doesn't quite fit the messages below, though.

> Last spring we had a discussion of these issues last spring.
> Here is the start of the thread:
>   http://www.openldap.org/lists/ietf-ldapbis/200205/msg00034.html
> 
> Steven Legg raised a concern about this
>   http://www.openldap.org/lists/ietf-ldapbis/200206/msg00002.html
> 
> which lead to the I-D having its current wording.  I (as Editor)
> believe the current language in this area reflects WG consensus.
> Of course, the purpose of this WG Last Call is to confirm WG
> consensus. I suggest you review the above referenced messages
> as well as other messages you might find on this subject in the
> archives.  After doing so, if you have concerns, make a specific
> suggestion (change X to Y) for the WG to consider.

My concern, in light of that discussion, is that the current text can
lead a reader to assume that if a server only publishes schema elements
it supports.  If I understand that discussion correctly, adding this to
4.4 should cover it:

  Clients SHOULD NOT assume that server supports all published or
  referenced elements of a particular definition.

-- 
Hallvard