Re: 0.8.2 candidate

Well, why don't we then add another variant of normalization function which 
would find maximal component and divide by it before length calculation? 
That function could be used everywhere where it is known that vectors could 
be rather long.

Oleh Derevenko
-- ICQ: 36361783

----- Original Message ----- 
From: "Jon Watte" <[email protected]>
To: "Oleh Derevenko"


> If you do the analysis of how the bit errors propagate, the distortion in 
> direction is insignificant compared to the length problem.
>
>
> Oleh Derevenko wrote:
>> By introducing extra multiplication before length calculation you will
>> decrease error of addition operation but introduce more distortion to 
>> vector
>> direction because that multiplication will be a separate operation and 
>> its
>> result will be truncated to precision of float as well.
>>