On Mon, Aug 26, 2013 at 11:42 AM, Jonathan S. Shapiro <[email protected]>wrote:
> On Sun, Aug 25, 2013 at 11:34 AM, Florian Weimer <[email protected]> wrote: > >> I suppose the routines you look at can only correctly deal with a >> restricted set of inputs (probably those generated by the >> corresponding binary-to-decimal conversion). But they fail on >> arbitrary precision decimal input. >> > > Florian: stop looking at examples and start looking at the numerical > analysis of the actual conversion. This is a precise conversion. > Clarification: this is a conversion that takes an exact decimal input and correctly produces the closest floating point value. The floating point result is inexact in some cases because of the inherent limits of floating point representation, but no loss of exactness is happening in the conversion process itself.
_______________________________________________ bitc-dev mailing list [email protected] http://www.coyotos.org/mailman/listinfo/bitc-dev
