Re: Motion P1788/M0029.01: Level-3-interface-only
Baker, Lee and P1788
On 19 Nov 2011, at 03:15, Ralph Baker Kearfott wrote:
> Since Motion 29 has been made by John Pryce and
> seconded by Lee Winter, the discussion period now
> begins, and will end after Friday, December 9, 2011.
> I append the motion, along with John's original short
> comment.
A change and a few clarifications, following comments so far:
1. Friendly amendment. I agree with Dan Zuras: change the motion so that *every* (supported or available) 754-conforming type shall have an interchange format as described. This applies in particular to any multi-precision interval system each of whose precisions is 754-conforming.
2. Some comments indicate I may not have been clear enough. The format I'm proposing is
- public
- platform-independent
(isn't that what "interchange format" means?) and
- almost a straight binary dump of any reasonable way of storing the interval internally.
E.g. an implementation might store [xlo,xhi] as the pair (-xlo,xhi) with Empty stored as (NaN,0). The map from that to the proposed interchange bitstring is fairly trivial.
3. In case Ulrich is wondering: his Operation Tables (motion 5) will be in Level 3, but as informative text, not mandatory.
John