RE: I-D ACTION:draft-ietf-ldapext-ldapv3-dupent-02.txt

["Salter, Thomas A" <Thomas.Salter@unisys.com>]

Ok, but in that case, I think you should include a description of behavior
when the control returns an error but the search succeeds.   Is the result
the same as if the control was not specified?  What if the control is
critical?  Maybe another way of asking this question is, 'Of what use is
this information to the client?'  Why not just combine this error status in
the search result status?  

I can certainly see an advantage in returning extra error information if the
control is ignored because of the error.  Then there is clearly two pieces
of information to convery: the reason the control was ignored and the status
of the search.  On the other hand, if the operation is rejected because of
the control error, then why not put the status directly in the search
result.

If server side sort and virtual list view controls also do this, then these
same questions apply to those drafts.

Suppose all three (sort, vlv, and dupent) controls are specified on a
search.  What is the procedure the client app uses to test all the command
statuses and decide how to proceed?


 > -----Original Message-----
 > From: Jim Sermersheim [mailto:JimSe@novell.com]
 > Sent: Monday, October 18, 1999 12:26 PM
 > To: ietf-ldapext@netscape.com; Thomas.Salter@unisys.com
 > Subject: RE: I-D ACTION:draft-ietf-ldapext-ldapv3-dupent-02.txt
 > 
 > 
 > Right, I included those errors for cases where a search may 
 > succeed, but the use of the control causes a failure. I 
 > tried to limit the error types to those that could be caused 
 > by servicing the control.  I believe the server side sort 
 > and virtual list view controls do this as well.
 > 
 > Jim
 > 
 > >>> "Salter, Thomas A" <Thomas.Salter@unisys.com> 10/18/99 
 > 08:17AM >>>
 > Why does this control define its own error results?  Can't 
 > these error
 > conditions all be handled in the error result of the search 
 > operation?  If
 > not, how do the two error conditions interact?  Is it 
 > possible for the
 > search to succeed, but the control to fail?
 > 
 > 
 > 
 >