Re: ISSUE-126 (Revisit Datatypes): A new proposal for the real <-> float <-> double conundrum

On 4 Jul 2008, at 16:18, Boris Motik wrote:

> Well, this might make determining the number of floats in a range  
> easier. I still don't see the point in requiring people to do all
> that,

My point is that I'd rather not introduce a variance for the  
semantics of floats unless there is great benefit. Numerical methods  
is a difficult field. I'd prefer to be very careful before doing  
"obvious" simplifying moves.

In this case, since I believe it's no more difficult than for range  
of integers, thus it doesn't complicate the implementation  
"significantly", I think we should stick with the traditional semantics.


Received on Friday, 4 July 2008 16:27:08 UTC