Re: infinite result sets?

"Bob Badour" <bbadour_at_golden.net> wrote in message news:R-OdnYzwFoIEzrPdRVn-jA_at_golden.net...
> "Graeme Newlands" <graeme_at_inorbit.com> wrote in message
> news:aecd37bf.0402140841.49e8a115_at_posting.google.com...
> > Just a random thought over the weekend ...
> >
> > Is there any join operation that could possibly return an infinite
> > result set in set theory?
>
> Yes, assuming at least one operand is an infinite set.

Or the join is deliberately recursive/cyclic in nature such that its end is to invoke its beginning.

> > Is this operation definable in SQL?
>
> Actual dbmses are finite; thus all operands are finite. However, one could
> achieve a similar effect by evaluating the cartesian product of two very
> large finite tables. By exceeding the capacity of the system, the large
> finite result might just as well be infinite.

But at some time machines will be built to evaluate such and thus this result will change, for although large it is not really infinite.

Better to specifically engineer a loop within the view by which the data will be pumped out so long as the processor keeps working.

Pete Brown
Falls Creek
Oz Received on Sun Feb 15 2004 - 05:28:51 CET