RE: WG last call on "duplicate entries"

["Jim Sermersheim" <jimse@novell.com>]

FWIW, Kurt's example below reflects the original intent. Any attribute you specify represents that attribute and any subtypes (whether by "natural" attribute subtyping or via attribute description option(s)). If an attribute is specified that has subtypes, it is analogous to specifying that attribute and all its subtypes (as a sequence of attribute descriptions). I believe this is most consistent with the way attribute subtypes are handled in other scenarios (especially when requesting attributes to be returned from search).

That said, the draft is weak on specifying this behavior, and I can fix. 

I don't know how to best resolve with 2891.

Jim


>>> "Kurt D. Zeilenga" <Kurt@OpenLDAP.org> 3/27/01 2:50:05 AM >>>
At 09:48 AM 3/27/01 +0100, Dave Watts wrote:
>Kurt, Jim,
>
>"Both" is a fine answer, but I'm not sure Kurt's suggested minor editorial
>change clarifies things explicitly enough. We could add something like the
>following to 4.1.1 (or made into a 4th example in section 5).
>
>...
>
>For the purposes of determining multi-valuedness, values of subtypes are
>treated as if they were values of the supplied attribute.  eg 'cn;lang-en'
>is a subtype of 'cn' is a subtype of 'name', so an entry with
>  cn: Johann
>  cn;lang-en: John
>is considered to have two values of 'name'.

That's not how I believe it should be viewed.

If you dupliate 'name' and you have:
        cn: Johann Smith
        cn: John Smith
        sn: Smith

You should get back two entries, one with
        cn: Johann Smith
        sn: Smith

and one with
        cn: John Smith
        sn: Smith

that is, requesting duplication of 'name' implies requesting
duplication of 'name' and its subtypes.