Against PEP 240

[Roman Suzi]

On Thu, 31 May 2001, Laura Creighton wrote:

> "Nick Perkins" <nperkins7 at home.com>:
> >On the other hand, the current system has never bothered me.  I enjoy having
> >tiny inaccuracies hidden from me.  It all depends on coming to peace with
> >the idea that a float is just a number that is very close to the number that
> >I want it to be.
> 
> The only thing necessary for the triumph of evil is for good men to do nothing.
>     - Edmund Burke (1729-1797)
> 
> (Not to pick on Nick, you do have to in some way get peace with the floating
> point to use it, though _my_ _word_ we are different people tempermentally.
> Hidden inaccuracies are what keeps me up at night, sweating.  No
> hyperbole.  But I think doing nothing rather than fixing the default
> behaviour of python on treatment of numbers that have a period in them
> <And thanks to more conversation with Remco Gerlich I think that it
> is decimal, not rational that is the way to go, somebody try to talk us
> out of it>  is exactly what Burke was talking about.

Financial applications better do in integers (long integers
if you work for large enough company ;-)

But in all other cases (save rocket science, some areas of physics and
number theory) even float accuracy is more than enough for most
applications. Noise usually have higher value than 7-th digit. If your
algorithm diverges due to inaccuracies, choose another one.

Why do you want to have more accuracy than double float?

> Laura 

Sincerely yours, Roman A.Suzi
-- 
 - Petrozavodsk - Karelia - Russia - mailto:rnd at onego.ru -