some linting

[fperrad]

No description provided.

[minad]

uL suffix is wrong for small digits (MP_16BIT).

[fperrad]

reworked

[czurnieden]

The chance is quite high that tmp * rand_int() overflows, which just happend in Travis. If you want to add parentheses it needs to be more something like

double dbl = ( (double)tmp * rand_int() ) + 1.0;

or maybe even cast rand_int() too and do a

double dbl =  (double)tmp * (double)rand_int() + 1.0;

[fperrad]

@czurnieden fixed with another way

[czurnieden]

fixed with another way

Travis moans: "warning: cast from function call of type ‘int’ to non-matching type ‘double’"

rand_int() -> rand_double()?

static double rand_double(void)
{
   uint64_t x;
  /* Doesn't work with feeding a float directly?  */
   if (s_mp_rand_source(&x, sizeof(x)) != MP_OKAY) {
      fprintf(stderr, "s_mp_rand_source failed\n");
      exit(EXIT_FAILURE);
   }
   return (double)x;
}

And then:

double tmp = rand_double();
double dbl = (tmp * rand_double()) + 1.0;

But there is a change that this produces NaN, +-inf, and denormal numbers, we would need to weed them out in rand_double().

[fperrad]

I think that a rand_double() could help.

But at this time, I'll remove this part of the commit.

[czurnieden]

But there is a chance that this produces NaN, +-inf, and denormal numbers, we would need to weed them out in rand_double().

No it wouldn't, because it is just a cast; we use the value not the bitpattern.
Sorry, my fault.