Path: utzoo!attcan!uunet!mcsun!tuvie!vexpert.dbai.tuwien.ac.at!mst From: mst@vexpert.dbai.tuwien.ac.at (Markus Stumptner) Newsgroups: comp.databases Subject: Re: Stoopid questions - One down, one to go Message-ID: <3696@vexpert.dbai.tuwien.ac.at> Date: 1 Oct 90 10:04:08 GMT References: Distribution: comp Organization: DB and ES Subdivision, TU Vienna Lines: 44 From article , by cimshop!davidm@uunet.UU.NET (David S. Masterson): > In article > Chuck.Phillips@FtCollins.NCR.COM (Chuck.Phillips) writes: > > Continuing the "value unknown" thread... > [...] > ergo, in three-value boolean algebra: > > TRUE OR UNKNOWN = TRUE > > ...but in relational algebra > > TRUE OR UNKNOWN = UNKNOWN (!) > > This is not what I remember seeing in Codd's book. Certain conclusions have > to occur in order to maintain transitive closure. I think relational algebra > goes along with boolean algebra. Unless this is some mix of three and four > valued logic?!? Codd definitely uses the standard TRUE OR NULL = TRUE. The real "loss of information" problems with his approach arise from straightforward use of this rule in tautologies: (birthdate < 1-1-66) OR (birthdate >= 1-1-66) results in the two halves of the disjunction being evaluated on their own (both to NULL), and the result of the whole expression is then NULL, even though it is clear that for any known value (which NULL stands for), at least one half and thus the whole expression would evaluate to NULL. There is an article in the ACM SIGMOD RECORD (Vol. 15:4, Dec. 86) where Codd writes on this and related topics. He concludes (not wholly without justification) that this is not a very pressing problem, since building tautologies into your conditions is a rather stupid thing to do, anyway. Still, it's annoying. In that paper, Codd also writes about four-valued logic based on "unknown" and "not applicable" values (though he doesn't call them that). I have not read his book, so I don't know whether this discussion is mentioned there. Markus Stumptner mst@vexpert.dbai.tuwien.ac.at Technical University of Vienna vexpert!mst@uunet.uu.net Paniglg. 16, A-1040 Vienna, Austria ...mcsun!vexpert!mst