I'm seeing a regression failure on the horology test on two different machines. I'd venture a guess that it is related to this change:
http://archives.postgresql.org/pgsql-committers/2003-02/msg00166.php
Joe