Re: SQL1999-standard makes no reference to term "relational" - why?

>> although SQL is widely recognized as the international "relational"
database standard, [the SQL1999 standard] does not describe itself as such; in fact, it never uses the term "relation" at all! (Indeed, it does not use the term "database" either <<

We were defining a programming language, not a mathematical model. You might also look at the fact that WHERE clauses use <search condition>s and not logical predicates. The idea is to keep all the definitions "inside" the standard as best we can and tread a line between a purely physical description of the implementation (say, with something like "FLOAT means IEEE floating point numbers and nothing else!") and something totally abstract (say, "Relational means Dr. Codd's Version 2.0 model and nothing else!").

You look at the early Cobol, Fortran, etc. standards, you will find that they were very physical -- arrays were defined as having row-major or row-minor order, the size of numerics was given in bytes along with their representation (two complement, etc.), variables had to be allocated in contigous storage for REDEFINES, etc. Received on Sun Jul 21 2002 - 23:12:17 CEST