early quit when replicating

[Matthieu Patou]

Hello metze & tridge,

I've pushed in my repo a branch called drs_earlyquit to address the 
problem of timeout when doing a getncchanges with highwatermark = 0 + 
replication_vector.

I'm still working on the timebased solution that tridge suggested me but 
there is a couple of patches that should make their way in the tree like 
for instance the use of Critical + ANC flags while replicating base_dn 
both for python join a C join ("becomedc") also the patch for reducing 
the number of  ldb_msg_find_attr_as_* calls.

Then we can discuss wether or not we need to always sort for base_dn.

Matthieu.

-- 
Matthieu Patou
Samba Team
http://samba.org