Re: The danger of (dropping) num2interval
On 2012-02-28 17:56:39 +0100, Marco Nehmeier wrote:
> Dear P1788,
>
> most of the interval libraries have the constructor num2interval and
> therefore people are used to it.
IEEE 1788 is for the future. One of its goals is to remove defects
from existing libraries. There will also be new users (more and
more?), who will not be used to this constructor.
> So if we drop it there is the danger that num2interval will be provided
> anyway without warnings for the user.
The standard could have a footnote or examples to explain how to
convert a point value to an interval and the problems related to
this.
> Therefore it is maybe an option to recommend num2interval together with a
> clear documentation.
IMHO it shouldn't be recommended or even suggested. If an
implementation provides num2interval as an extension, it's at its
own risks. And it's up to the implementation to provide a clear
documentation (the standard won't be able to change that anyway).
Also remember that most users don't read the documentation. So,
it is better to avoid traps.
--
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)