>>and certainly RAND() cannot return 1.
>
>I'll address only this aspect of your message. A bunch of us fought to have RAND() return only one of 0 and 1. Otherwise, it's too hard to spread the return value evenly over a range.
>
>Before VFP 5, RAND() could return both 0 and 1, each of them rarely. It made turning random values into a range (m,n) very difficult. With no 1's, we can also do INT(n * RAND()) + 1 to get values from 1 to n and know they'll be properly distributed.
>
>Tamar

Of course.

Tamar, if this can interest you,
I have discovered this thing, this morning,
while I was using BINTOC(,"B") to produce a pseudocasual string
with a more rapid way in comparison to the banal solution CHR(RAND()*256).
It has immediately appeared that
only the first part of the mantissa was not zero.
The rest is obvious consequence...

Fabio