|
From: Barton W. <wi...@un...> - 2016-01-27 18:36:22
|
> I seem to remember that ceil() and floor() did use bfloats internally - For constant expressions, floor and ceiling attempts to determine the value using big floats. The code tries to be cautious and return a nounform when it is too close to call. The method isn't perfect. But the float((exp(10000)) --> big float misfeature is unrelated to the the way floor and ceiling work (or fail to work). --Barton (author of floor and ceiling) |