On Mon, 2005-12-19 at 11:13 +1300, Mark Kirkwood wrote: > > My understanding: Teradata and DB2 use this. > > FWIW - I think DB2 uses the successive fact RID buildup (i.e method 2), > unfortunately I think you're right; I was thinking about that point too because DB2 doesn't have index-organised tables (well, sort of: MDC). I was confused because IBM seem to have a patent on (1), even though it seems exactly like the original NCR/Teradata implementation, which predates the patent filing by many years. Wierd. It's a minefield of patents.... > I haven't got a working copy of DB2 in front of me to test. True, not all copies work :-) Best Regards, Simon Riggs
