Re: matched values 05

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

>It is perfectly reasonable
>for the MVO control to extend search operations extended by a sort
>control and to depend the semantics upon sequencing of controls
>(it is a sequence after all).

The problem with this statement is that nothing in RFC 2251 ties the order in which controls are processed to the ordering of the controls in the controls sequence. Thus, either 
1) 2251 needs to be fixed, 
2) every control spec needs to specify it's relationship with all other related controls (which is impossible), or
3) we just settle for interoperability problems.

Jim




>>> "Kurt D. Zeilenga" <Kurt@OpenLDAP.org> 12/19/00 7:37:43 PM >>>
At 03:33 PM 12/19/00 -0800, Bruce Greenblatt wrote:
>If you process sort then MVO:  You sort entries based on certain attribute values which may be subsequently deleted by the MVO control processing.

Yes.

>If you process MVO the sort: The sorting is always done on the remaining entries.

Yes.

>It appears to me that there is the possibility that the search results will come back in a different order depending on how the MVO and sort controls are processed.

Yes.

>Section 4 states: "This control acts independently of other LDAP controls such as server side sorting".  To me if the controls were independent, I would get exactly the same data back in exactly the same order.  So, I don't think that MVO and the sort control are "independent".

I would say that each control acts independently upon a set of inputs
provided to it but that the input to the control is dependent on
what factors proceed it in the processing of the operation. 

>Note that RFC 2891 requires the following processing method for multi-valued attributes: when an entry happens to have multiple values for that attribute and no other controls are present that affect the sorting order, then the server SHOULD use the least value (according to the ORDERING rule for that attribute).

This issue can be resolved by simply stating that the MVO control
may affect sort ordering in the presence of sort controls as it
may alter results after the sort has been completed.

>So, I would require the MVO to be processed BEFORE the sort control.  Otherwise, the sorted entries may turn out to be no longer sorted.

I disagree.  The sort control, like any other control, is subject
to other factors including extension.  It is perfectly reasonable
for the MVO control to extend search operations extended by a sort
control and to depend the semantics upon sequencing of controls
(it is a sequence after all).

Kurt