On Tue, Oct 26, 2004 at 11:35:50PM -0400, Pierre A. Humblet wrote: >Question: Are there any objections to using NULL instead to indicate >that the tokens do not exist? Personally, I like NULL tests because they're cheap so no objections from me. In Corinna's absence, you're our remaining expert in this area, so please just check in what you think is best. cgf