Is ColumnSlice range inclusive or exclusive?

When the iterator has to update the range to get additional columns,
the column name of the preceding queries last column is used as the
start point for the new query.  On line 96 of ColumnSliceIterator, the
logic consumes the first column returned by the new query.  The basis
of the logic is that the start and end points of the query are
inclusive.  This no longer appears the be the case.  Can anyone shed
light on the expected behavior (I'm using cassandra 1.0.9)?  The only
reference I've been able to find with respects to the start and end
points being inclusive is
http://mail-archives.apache.org/mod_mbox/cassandra-user/200912.mbox/%3Ce06563880912211858l798fb57cpf39404807ae6efa7-JsoAwUIsXosN+BqQ9rBEUg <at> public.gmane.org%3E.

I am willing to fix and submit a patch, but I'd like to know what the
actual rule is about range inclusion.

Thanks,

Shane