RE: back-dnssrv

["Howard Chu" <hyc@highlandsun.com>]

> -----Original Message-----
> From: owner-openldap-devel@OpenLDAP.org
> [mailto:owner-openldap-devel@OpenLDAP.org]On Behalf Of Pierangelo Masarati

> > back-dnssrv seems to be lately broken.
> >
> > First, I'm not quite sure what the purpose of the following
> > assertion is, but it crashes slapd for normal searches:
> >
> > 	assert( get_manageDSAit( op ) );
> >
> > Should it perhaps be inverted? This is in dnssrv_back_search()
> > and dnssrv_back_compare().

Well, that assert has been in search.c since rev 1.7, so this isn't a new
behavior. Odd that no one has tripped over it before now. Given that the only
entries that dnssrv_back_search generates are referral objects, I suppose it
makes sense that you must provide the manageDSAit control if you expect to
get any entries back from your search request. But probably it shouldn't be
an assert, maybe sending LDAP_UNWILLING_TO_PERFORM or LDAP_OPERATIONS_ERROR
in the absence of the control might make more sense. Or simply calling
send_search_reference instead of send_search_entry...

dnssrv_back_compare isn't even referenced, it's just an orphaned stub.

> > Secondly, rs->sr_ref is uninitialized, which causes a further
> > assertion failure in send_ldap_response().

This is very odd. None of the other backends have a problem with this, and
the SlapReply is exactly the same for all the code since it's only
initialized in one place, connection_operation().
>
> I'll have a look at it; probably the new ABI upgrade left over
> some typos/flaws.

  -- Howard Chu
  Chief Architect, Symas Corp.       Director, Highland Sun
  http://www.symas.com               http://highlandsun.com/hyc
  Symas: Premier OpenSource Development and Support