[Date Prev][Date Next]
[Chronological]
[Thread]
[Top]
Re: Duplicate entries filter
At 11:30 PM 8/16/00 +0100, David Chadwick wrote:
>Date sent: Wed, 16 Aug 2000 10:31:18 -0600
>From: "Jim Sermersheim" <JIMSE@novell.com>
>To: <Kurt@OpenLDAP.org>,<d.w.chadwick@salford.ac.uk>
>Copies to: <ietf-ldapext@netscape.com>
>Subject: Re: Duplicate entries filter
>
>> These examples do reflect the original intent of the I-D. While
>> writing it, I had assumed the behavior previously discussed as a)
>> (filter, duplicate, return). Given the usefulness of Kurt's example, I
>> changed it to b) [filter, duplicate, filter, return (if your server
>> can do this more efficiently, more power to ya)]. Then the notion of a
>> flag was introduced.
>
>Jim,
>two points
>
>i) I note that in both a) and b) above you never mention (duplicate,
>filter, return) which was one of the earlier threads that was being
>discussed. Is this option finally being laid to rest? I hope so.
Please... Remember I only dared to suggest it. :-)
>ii) for option b) by combining duplicate entries and matched values
>we actually have (search filter, duplicate, match filter, return) which
>is more powerful than using the same filter in both instances. And
>as Kurt pointed out, we can alternate the ordering of the duplication
>and match filter (although this probably will not give different results
>except for the case of when an empty set of attributes is returned.
>With duplicate followed by match you could get identical entries
>returned several times, containing an attribute with no values).
This may be the case, it's just wasn't obvious to me (and "probably"
you as well). I'll have to chew on this to see if both orderings
are useful enough to warrant specification.
I namely brought the ordering of controls issue as it seemed relevant
to the "Extensions Guide" thread. That is,
Controls ::= SEQUENCE OF Control
implies that order may matter (and multiple controls of the
same type or even identical controls may be listed multiple times...
which may or may not make sense depending upon the semantics of
the control).
Kurt