Re: Motion P1788/M007.01_NaI: Discussion period begins
As Arnold just reminded us, NaI should carry a payload -- and we should
even standardize names for some essential kinds (if not actual values),
together with whatever special propagation properties they entail. If
actual values are specified they should be specified as small integers.
One nice property of Vincent's suggestion
NaI: ?NaN,non-NaNÙ
is that the payload could be carried in the non-NaN. This gets around
the difficulty that (unlike DFP) binary NaN payloads are incompletely
specified -- and that, since the truncation/padding rules of the bit
pattern ARE defined (on the "right", i.e. low-order fraction bits), and
are just the opposite of what was defined for DFP (on the "left", i.e.
high-order trailing-significand digits), if payload-preserving format
conversions are to be supported, the conversion rule has to be defined
explicitly. IBM's conversion instructions interpret BFP payloads as
the integer resulting from bit-reversal of the raw bitstring starting
immediately after the QNaN/SNaN indicator bit. I don't know what other
platforms do. IBM's rule has the advantage that small integers are
preserved through narrowing, widening and radix conversions, as long
as they don't exceed what can be expressed in the narrowest format
(e.g. 999999, max payload for DFP32, or 511 if BFP16 is supported.)
Michel.
---Sent: 2009-08-06 14:24:25 UTC