Re: Unique identifiers for LDAP attributes

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

>>> "David Chadwick" <d.w.chadwick@salford.ac.uk> 7/13/00 11:02:47 AM >>>

>There is another interesting problem that you may be interested in 
>related to the non-use of OIDs. The matching rule used to select a 
>subschema definition is, wait for it....
>
> objectIdentifierFirstComponentMatch
>
>Thus the client needs to know the OID of the schema definition it 
>needs to selectively fetch it. But if LDAP never passes an OID to 
>the client, how does the client know which subschema definition it 
>needs? In order to solve this, it means we really need  a 
>"nonUniqueStringSecondComponentMatch" matching rule to be 
>defined for LDAP.

Though I agree with the general notion of moving toward the use of unique OIDs, there's a minor flaw with this statement.

objectIdentifierFirstComponentMatch uses the OID syntax which can either be a numericoid (1.2.3.4) or the descr form (cn), so it's still usable with short names as it stands today.

Jim