Re: [boost] Infinite precision integer draft

From: Carlo Wood (carlo_at_[hidden])
Date: 2006-05-30 21:19:49

On Tue, May 30, 2006 at 09:20:59AM -0700, Geoffrey Irving wrote:
> >   Z<r> x = N;
> >   Z<s> y = N;
> 
> As pointed out by someone else already, what if r = 2^30402457-1?
> 
> More fundamentally, encryption usually fails if the secret prime numbers are
> hard coded into the encryption code as template arguments.  Unless you want
> the encryption code to recompile itself whenever it generates a new random
> key, that is.

Okay. I agree that it should be possible to have dynamic (or non-builtin
integral) numbers for the modulo (say 'p'). One could still argue that in those
cases it probably isn't necessary to work in Z_p, but well. A general library
shouldn't post restrictions on practical use like that, I guess-- and you
are right that a template argument would be limited to int or long long at
most.

The only real advantage of using a template argument is probably that you
can make things faster during runtime - but as always, the algorithm used
is much more important than some constant factor that one might gain there.

So, I'm convinced ;). I'm okay with a library that allows the modulo
to be set during runtime.

-- 
Carlo Wood <carlo_at_[hidden]>