Re: bignum branch

[Paul Eggert]

Andy Moreton wrote:
> The size of a GMP limb is irrelevant - it is the API for getting native
> integer types into mpz_t values that cannot handle interger types wider
> than long.
>
> Wider mp_limb_t types improve efficiency, but do nothing about
> correctness on LLP64 systems.

You're right, of course. It would be helpful to have that two-line change 
anyway, for performance reasons on 32-bit systems; however, Eli seems to be 
thinking that we shouldn't worry much about performance on these obsolescent 
systems and I'm becoming more inclined to agree.