Re: (ITS#4086) rwm-mapping problem

[hyc@symas.com]

ando@sys-net.it wrote:
> I've committed a fix that temporarily solves your problem - i.e. no more
> core dumps.
>
> However I think that part of the code needs some review.  In fact, if
> you map some attributes to nothing, and they're used in a filter, it is
> unclear (to me) what would be the most appropriate behavior.  If the
> related assertion is (silently) removed from the filter, the semantics
> of the filter is changed; so it may seem appropriate to reject the
> operation.  The fix just committed returns LDAP_OTHER since the current
> behavior reflects the fact that it is the result of an implementation
> specific error condition.
>
> Another approach would be to turn that assertion into an undefined
> filter, letting the operation to continue accordingly.  I currently
> favor this second option, but it may require a bit more work.
>   
Yes, setting that filter component to undefined that would be more in 
keeping with normal behavior.

-- 
  -- Howard Chu
  Chief Architect, Symas Corp.  http://www.symas.com
  Director, Highland Sun        http://highlandsun.com/hyc
  OpenLDAP Core Team            http://www.openldap.org/project/