dynamic vs. static rounding

[David Stewart]

To throw my two-bits' worth as a newcomer to the NCEG group, I would favour
dynamic rounding.
	Why?
	Because without it things like interval arithmetic cannot be
properly implemented. While interval arithmetic will never be the be-all and
end-all of computer arithmetic as some thought it might
(see e.g. Triplex Algol) it's still worth having. Interval arithmetic and its
relatives for all their faults are still, as far as I can see, the only way
to get ** guaranteed ** bounds for general computations.
	And it needs dynamic rounding control as (e.g.) adding two intervals
requires one rounding up and one rounding down.
	Currently it is inmpossible to write ** portable ** interval
arithmetic libraries as there is no standard way of dealing with rounding.
					-- David Stewart
------------------------------------------------------------------------
David Stewart                   Program in Advanced Computation
desathrain.anu.edu.au           School of Mathematical Sciences
                                Australian National University