Motion M0033.01 -- Number format: YES
I vote YES on Vincent's Motion 33.
I have a mild uneasiness about the interpretation that 754-conforming
implies that 1788 maintain the distinction between signed zeros. (B3)
It is of course possible to define the sign of a zero result for those
P1788 operations that return a number -- but the actual sign of zero
returned by the basic implementation might not always be as required,
so that an additional test might be needed. Is that justifiable?
I think I would prefer laxness in this case, and allow the sign of
zero to be unspecified, thus shifting the burden of testing it to
the rare applications that might care. (This is the Vienna way.)
Surely we know of no cases where the distinction would be meaningful
in the P1788 context? Or rather, surely we DON'T WANT TO INTRODUCE
such a distinction, e.g. width(singleton) = +0 but width(empty) = -0?
I also think that a clarification is needed for rounding direction.
The applicable rounding direction would be determined by the operation,
and never by 754's concept of "current rounding direction", right?
(The "current rounding direction" would of course apply to any numeric
operations on the results of P1788 operations -- but that's outside
the scope of P1788, as the motion makes clear already.)
Michel.
---Sent: 2012-05-29 20:06:34 UTC