Re: [boost] Multiprecision Review

Subject: Re: [boost] Multiprecision Review
From: John Maddock (boost.regex_at_[hidden])
Date: 2012-06-25 06:51:15

> As I've said before, I imagine algorithms that operate on sequences of 
> "limbs" (as John calls them), e.g. (PSEUDO-CODE)
>
> template <typename ITER, typename OITER>
> void add(ITER begin1, ITER end1, ITER begin2, ITER end2, OITER result)
> {
>   bool carry = false;
>   while (begin1!=end1 && begin2!=end2) {
>     (carry,*result++) = add_with_carry(*begin1++,*begin2++);
>   }
> }

Without wanting to reignite this discussion too much, I'll merely point out 
that there are a number of situations where that would break - fixed 
precision integers being but one.  In most cases the fixes are trivial, as 
long as you don't mind the performance hit - except of course you do :-(

Cheers, John.