gmp hello world

[luser droog]

It's alive! It's alive!

//number.c
//$make number LDLIBS='-lgmp'
#include <stdarg.h>
#include <stdint.h>
#include <stdio.h>
#include <gmp.h>

void nothing(mpz_t c, const mpz_t a, const mpz_t b){
mpz_set_ui(c,0);
}

int main(){
char op[2];
//scanf("%d %1s %d", &a, op, &b);
mpz_t az,bz,cz;
mpz_init(az);
mpz_init(bz);
mpz_init(cz);
gmp_scanf("%Zd %1s %Zd", &az, op, &bz);
( (*op=='+')?mpz_add:
(*op=='*')?mpz_mul:
nothing ) (cz, az, bz);
gmp_printf("%Zd\n", cz);
}

--
My creature is alive!

[luser droog]

On Friday, June 3, 2016 at 12:34:32 AM UTC-5, luser droog wrote:
> It's alive! It's alive!
>
> //number.c
> //$make number LDLIBS='-lgmp'
> #include <stdarg.h>
> #include <stdint.h>
> #include <stdio.h>
> #include <gmp.h>

> gmp_scanf("%Zd %1s %Zd", &az, op, &bz);

Don't need the `&`s. Since the `mpz_t` is an
array [1] of a struct type, (char*)(az) == (char*)(&az).
So it doesn't hurt anything, I think.

[Barry Schwarz]

The fact that the two expressions will compare equal after the cast is
irrelevant since nothing in your code will perform that conversion.

%Zd is not a standard format for scanf functions. Does it mean the
same as %zd? Whatever it means, are you sure your struct is properly
aligned for that data type? Is the first member of the struct a type
compatible with this format?

While we are at it, are you absolutely certain that a pointer to your
struct or a pointer to an array of your struct have the same size and
representation as a pointer to whatever the data type %Zd requires.

--
Remove del for email

[luser droog]

On Friday, June 3, 2016 at 3:46:12 AM UTC-5, Barry Schwarz wrote:
> On Thu, 2 Jun 2016 22:42:26 -0700 (PDT), luser droog
> <luser...@gmail.com> wrote:
>
> >On Friday, June 3, 2016 at 12:34:32 AM UTC-5, luser droog wrote:
> >> It's alive! It's alive!
> >>
> >> //number.c
> >> //$make number LDLIBS='-lgmp'
> >> #include <stdarg.h>
> >> #include <stdint.h>
> >> #include <stdio.h>
> >> #include <gmp.h>
> >
> >> gmp_scanf("%Zd %1s %Zd", &az, op, &bz);
> >
> >Don't need the `&`s. Since the `mpz_t` is an
> >array [1] of a struct type, (char*)(az) == (char*)(&az).
> >So it doesn't hurt anything, I think.
>
> The fact that the two expressions will compare equal after the cast is
> irrelevant since nothing in your code will perform that conversion.

That part was my (mistaken) rationalization for why the additional `&`s
still appeared to work correctly. The documentation does not use `&`.

> %Zd is not a standard format for scanf functions.

It's an extension implemented by the gmp_scanf and gmp_printf which
correspond to mpz_t parameters.

>Does it mean the
> same as %zd? Whatever it means, are you sure your struct is properly
> aligned for that data type? Is the first member of the struct a type
> compatible with this format?
>
> While we are at it, are you absolutely certain that a pointer to your
> struct or a pointer to an array of your struct have the same size and
> representation as a pointer to whatever the data type %Zd requires.

I'm confident that following the documentation will remove much of my
need for this level of scrutiny.

[luser droog]

On Friday, June 3, 2016 at 2:38:33 PM UTC-5, luser droog wrote:
> On Friday, June 3, 2016 at 3:46:12 AM UTC-5, Barry Schwarz wrote:
> > On Thu, 2 Jun 2016 22:42:26 -0700 (PDT), luser droog
> > <luser...@gmail.com> wrote:
> >
> > >On Friday, June 3, 2016 at 12:34:32 AM UTC-5, luser droog wrote:
> > >> It's alive! It's alive!
> > >>

> I'm confident that following the documentation will remove much of my
> need for this level of scrutiny.

A little more fleshed-out. 4-function calculator. Produces floating-point
results for division, but only accepts integer input.

//number.c
//$make number LDLIBS='-lmpfr -lgmp'

#include <stdarg.h>
#include <stdint.h>
#include <stdio.h>

#include <string.h>
#include <mpfr.h>
#include <gmp.h>

enum tag { NONE, Z, FR };

typedef union {
enum tag tag;
struct z {
enum tag tag;
mpz_t z;
} z;
struct fr {
enum tag tag;
mpfr_t fr;
} fr;
} u_number;
typedef u_number number[1];
typedef u_number *number_ptr;

void init_z(number_ptr z){
z->z.tag = Z;
mpz_init(z->z.z);
}

void init_fr(number_ptr fr){
fr->fr.tag = FR;
mpfr_init(fr->fr.fr);
}

void promote(number_ptr n){
mpz_t t;
memcpy(&t, &n->z.z, sizeof t);
init_fr(n);
mpfr_set_z(n->fr.fr, t, MPFR_RNDN);
mpz_clear(t);
}

#define op(func, C, A, B) \
if ((A)->tag==Z && (B)->tag==Z) { \
if (!strcmp(#func,"div")) { \
init_fr(C); \
promote(A); \
promote(B); \
mpfr_##func((C)->fr.fr, (A)->fr.fr, (B)->fr.fr, MPFR_RNDN); \
} else { \
init_z(C); \
mpz_##func((C)->z.z, (A)->z.z, (B)->z.z); \
} \
} else if ((A)->tag==FR && (B)->tag==FR) { \
init_fr(C); \
mpfr_##func((C)->fr.fr, (A)->fr.fr, (B)->fr.fr, MPFR_RNDN); \
}

void mpz_nothing(mpz_t c, const mpz_t a, const mpz_t b){
mpz_set_ui(c,0);
}

void mpfr_nothing(mpfr_t c, const mpfr_t a, const mpfr_t b, mpfr_rnd_t rnd){
mpfr_set_ui(c,0,rnd);
}

int main(){
number a, b, c;
char op[2];
init_z(a);
init_z(b);

gmp_scanf("%Zd %1s %Zd", a->z.z, op, b->z.z);
switch(*op){
case '+': op(add, c, a, b); break;
case '*': op(mul, c, a, b); break;
case '-': op(sub, c, a, b); break;
case '/': op(div, c, a, b); break;
default: op(nothing, c, a, b);
}
switch(c->tag){
case Z: gmp_printf("%Zd\n", c->z.z); break;
case FR: mpfr_printf("%Rf\n", c->fr.fr); break;
}
}

[Tim Rentsch]

Probably it will work, but technically it is undefined behavior,
because the type of pointer passed doesn't match the type
expected.

[Real Troll]

Is this no good for you these days? !!!!!!!!!!!!!!!!!

#include <stdio.h>
#include <conio.h>

int main()
{
printf("Hello World!!!");
printf("Press any key to continue ...");
getch();
}