Re: commit: ldap/servers/slapd filterentry.c schema_init.c

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

Pierangelo Masarati writes:
>kurt@OpenLDAP.org wrote:
>> Update uniqueMemberMatch to be consistent with latest
>> 	draft-ietf-ldapbis-syntaxes
>> Add uniqueMemberMatch approximate support
>> Add uniqueMemberMatch indexing support
>
> As a consequence of this change, test026 is failing when the server is
> searched with "(uniqueMember=dc=example,dc=com)", i.e. the asserted
> value is missing the bitString portion.  Since this option appears to be
> absent from draft-ietf-ldapbis-syntaxes, i.e. there's no mention of the
> possibility of having the asserted value without the bitString portion,

Yes there is.  Section 4.2.31 (uniqueMemberMatch) says:

   The rule evaluates to TRUE if and only if (...) and either, the
   <BitString> component is absent from both the attribute value and
   assertion value, or (...)

If you find that somewhat complex sentence ambiguous or too hard to
parse, maybe you should suggest a better wording to
ietf-ldapbis@openldap.org.

> if my interpretation is correct the test is malformed and should be
> removed, since it seems to imply that an absent bitString is equivalent
> to a bitstring of "#'0'B".

I don't see how you get that from the draft.  Also note that a bit
string is just that - a bit string, not a representation of an integer.
So if anything #''B would be a more natural "default".

-- 
Hallvard