Re: Fw: OpenLDAP Search bug?

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

At 12:46 PM 2002-09-09, Banzaitron wrote:
>I tried this with the latest CVS engineering release and the same thing
>happens.  I have rebuilt the entire database, rebuilt indices, etc and
>nothing seems to make a difference.
>Does this sound like a valid bug to open a report on?

Well, it does seem a bit odd.  If you do two search
operations which only differ by the filters:
  1.  (&)                        [always TRUE]
  2.  (objectClass=myorgservice) [depends on entry]

then not only should the set of entries returned by
the second search be a subset of the entries returned by
the first search, the candidate set for the second
search should be a subset of the candidate set of the
first.

However, both sets of entries may be same (if all
entries match the filter) and both sets of candidates may
be the same (if indexing for the second cannot eliminate
any candidate of the first).

Logging will help you determine the size of the candidate
sets and to see if they are somehow out of wack.

Kurt


>It definitely should
>not be searching out of the scope I pass it right?

slapd(8) should return entries which don't match your
search criteria.  slapd(8) uses indexing to avoid
checking all entries in most cases, but there certainly
are cases where slapd(8) may need to check all entries,
including ones which aren't under the baseObject or
with the specified scope parameter.