Re: Attribute subtyping, clarification request

["Kurt D. Zeilenga" <Kurt@OpenLDAP.org>]

At 07:44 PM 5/19/2005, Rici Lake wrote:

>On 19-May-05, at 9:22 PM, Kurt D. Zeilenga wrote:
>
>>
>>Not sure I am parsing this correctly, but basically,
>>        displayName;lang-es;lang-fr is a subtype of
>>                displayName;lang-es
>>                displayName;lang-fr
>>                displayName
>
>Yes, but a single-valued attributetype can still be present with multiple tags, as long as no two assertions have the exact same tags.
>Consequently, an entry with:
>
>displayName: Joe English
>displayName;lang-es: Jose Castellano
>displayName;lang-fr: Jo Francais
>displayName;lang-es;lang-fr: J Roman
>
>would be valid

As this entry has four distinct single-valued attributes.


>>So, if (displayName;lang-es;lang-fr=*) is TRUE then logically
>>(displayName;lang-es=*), (displayName;lang-fr=*) and
>>(displayName=*) would also be TRUE.
>
>>And displayName;lang-es;lang-fr
>>may be returned if the client requested any of the attribute
>>descriptions listed above (or all user attributes).
>
>Right, that was the question. So although displayName is a single-valued attribute type, a search request might return several different displayName's for the same DN, even if it specified that the attributetype to return is displayName;lang-es. Presumably it would not even be legitimate for the server to only return the exact match.
>In other words, if I'm writing a client, can I trust SINGLE VALUE to mean single valued?

In the sense that an attribute whose type is marked allowing
a single value will have only one value.  But this in no way
preclude the return of attributes, each having only a single
value, which whose description is a subtype of that type.

>My conclusion is that I cannot, and that the client needs logic to figure out which of the returned displayName's is the one to actually display.

Clients should expect return of attributes whose descriptions
are subtypes of descriptions requested.

For instance, cn may be returned when requesting name.

>I can live with it; I just wanted to be clear on the spec.
>
>Thanks for the prompt response.
>
>Rici