21 Jul
2010
21 Jul
'10
10:13 p.m.
On Wed, Jul 21, 2010 at 08:56:20PM +0200, JohnLM wrote:
I think this should reduce problems with two
conversions to/from
floats, when both inputs and outputs are of integer type.
It doesn't and there are no such problems. Once you have a float,
multiplying or dividing it by 2^N just means changing the exponent.
It's exact. Your 'original' floats are no more precise than those
that have been reduced to [-1..+1].
Ciao,
--
FA,
There are three of them, and Alleline.