Ciao,

Il Ven, 30 Novembre 2018 12:43 pm, Vincent Lefevre ha scritto:
> On 2018-11-30 07:43:40 +0100, Marco Bodrato wrote:
>> mpz_set_si (mpz_t r, signed long int x)
>> {
>>   if (x >= 0)
>>     mpz_set_ui (r, x);
>>   else /* (x < 0) */
>>     {
>>       mpz_set_ui (r, GMP_NEG_CAST (unsigned long int, x));
>>       mpz_neg (r, r);
>>     }
>> }
> 
> What is the purpose of GMP_NEG_CAST

The purpose of the macro?
Well, it's simple. When we write a piece of code for GMP, we do not want to take care of the best sequence to negate a negative number and cast it to unsigned, that's why we use a macro :-)

> #define GMP_NEG_CAST(T,x) (-((T)((x) + 1) - 1))
> instead of just a cast followed by a negation? i.e.
>   mpz_set_ui (r, - (unsigned long int) x);

I'm sure that the compiler can understand it anyway and do the right thing :-)
When we will agree that the way you propose is the best one, we will change the macro :-)

By the way, the code Paul wrote and I propose to simplify is the following.

/* MPZ assignment and basic conversions. */
/* function changed by PZ */
void
mpz_set_si (mpz_t r, signed long int x)
{
  if (x >= 0)
    mpz_set_ui (r, x);
  else /* (x < 0) */
    {
      if (x == LONG_MIN) /* special case for 2^-k */
        {
          mpz_set_ui (r, -(x + 1));
          mpz_add_ui (r, r, 1);
        }
      else
        mpz_set_ui (r, -x);
      mpz_neg (r, r);
    }
}

Following the flow Paul uses, maybe we should change the macro to

#define GMP_NEG_CAST(T,x) (((T) -((x) + 1)) + 1)

Personally, I have no opinion about that. I'll keep on using the macro in my code, however defined :-)

PS: are we sure that LONG_MIN has the form -2^k? Does it depend on the one-complement or two-complement representation?

Ĝis,
m

-- 
http://bodrato.it/