TODO


1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21

GMP has mpz_divexact(q,n,d), we should have zdiv_exact(q,n,d).
It uses optimised division algorithm that requires that d|n.

Add zsets_radix
Add zstr_radix
Add zranddist value based on % for fitness to bound

Test big endian
Test always having used > 0 for zero
  Test negative/non-negative instead of sign

Test optimisation of zmul:
  bc = [(Hb * Hc) << (m2 << 1)]
     + [(Hb * Hc) << m2]
     - [(Hb - Lb)(Hc - Lc) << m2]
     + [(Lb * Lc) << m2]
     + (Lb * Lc)

Would zmul be faster if we split only one of the
factors until they are both approximately the same
size?