Re: back to the roots - a simpler standard?
On 2013-07-03 08:32:21 +0100, John Pryce wrote:
> - If we had not had strong voices requesting support for mid-rad and other kinds of representation, we could have avoided "explicit" and "implicit" interval types, restricted ourselves to inf-sup, and made it even simpler.
I don't think that implicit types bring much complexity.
> - We might have restricted ourselves to systems whose underlying floating point is 754-conforming, which would have made it simpler again.
I'm not so sure. One would have to take care systems with and without
subnormals specifically. This would have also rejected some systems
including some implementations of native C types (e.g. long double on
PowerPC).
--
Vincent Lefèvre <vincent@xxxxxxxxxx> - Web: <http://www.vinc17.net/>
100% accessible validated (X)HTML - Blog: <http://www.vinc17.net/blog/>
Work: CR INRIA - computer arithmetic / AriC project (LIP, ENS-Lyon)