Re: Thinking about MINUS
Date: 7 Jan 2007 11:49:45 -0800
Message-ID: <1168199385.871193.166460_at_v33g2000cwv.googlegroups.com>
Bob Badour wrote:
> I wonder whether one can
> define some reasonable extension to MINUS for relations of different
> types that gives us PROJECT and/or JOIN etc.
This is very ambitious. The least reduction I know of has 3 operators. Actually, it looks more symmetric to have 4:
- [Inner] join (the adjective conventionally understood to be redundant)
- Inner union (the only purpose of the adjective is to disambigute this operation from set union)
- Antijoin (aka generalization of MINUS)
- Relational division
Antijoin is related to union in a way similar to the '-' is related to
'+' in ordinary arithmetics.
Relational division is related to join in a way similar to the '/' is
related to '*' in ordinary arithmetics.
It is well known that relational division can be expressed via join, inner union and antijoin. Curiously, antijoin can be expressed via join, inner union and relational division too. In a word: "Pick up the any 3 basic operators out of these 4 and you have complete set of relational operators"... Received on Sun Jan 07 2007 - 20:49:45 CET