IEEE 754

[John W. Baxter]

In article <8liiba$1mh$1 at agate.berkeley.edu>, ejr at lotus.CS.Berkeley.EDU 
(Edward Jason Riedy) wrote:

> And John W. Baxter writes:
>  - 
>  - You mean I may someday be able to put aside my answers to the "why 
>  - doesn't 2/10 print as 0.2 as we all know it should?" sorts of 
>  questions?
> 
> I wish.  IEEE 854 is a radix-independent FP standard.  It could be
> implemented for base-10 in hardware, but only some hand calculators 
> have done that.  There are many libraries for BCD-style arithmetic, 
> but I don't know how good or fast they are.  I don't understand all 
> the issues involved.
> 
> Way off-topic:  Besides, not all numbers are natural for base-10.  
> One example is pi.  It strongly appears that the only natural base 
> for pi is 16 (pi may be base-16 normal and not base-10 normal).  Not 
> proven yet, but looking likely.  See 
> http://www.nersc.gov/~dhbailey/dhbpapers/bcrandom.ps for some neat
> pi results.
> 
> Back on topic:  Python's answer is even more painful to explain:
> > bash-2.03$ python -c 'print 2/10'
> > 0
> 
> Jason

I have no problem with 2/10 being zero.  If one does integer arithmetic, 
one gets integer arithmetic results.

People who initially learned C with its silly standard conversions and 
different operator definitions have other biases...but I've been around 
computers nearly twice as long as C has (as of sometime around 1994 I 
had to stop saying "more than twice as long").

   --John

-- 
John W. Baxter   Port Ludlow, WA USA  jwbnews at scandaroon.com