Re: relations aren't types?
Date: 30 Dec 2003 13:11:27 -0800
> > Types are not atomic, values are.
> Given a set of values of the same type, could this set have both atomic and
> nonatomic values present?
I cannot answer this question, because I cannot make sense of a non-atomic value. The notion of atomicity is relative to the operator being invoked. I might add that its not a very useful notion in the first place. It is interesting to note that the concept of atomicity is only mentioned in passing in the preliminary remarks of TTM. What benefit is acheived by classifying values into atomic and non-atomic? Why does it matter. For a given operation, an argument is atomic, whether an integer value or a relation value or any value at all. Whether it can be "decomposed" is only a function of other operators invoked on the value as a whole. We could even make the argument that the integer value 5 is not atomic, because it can represented as 2 + 3. Received on Tue Dec 30 2003 - 22:11:27 CET