unable to use long double

[bostjanv]

Hello,
I'm having trouble using the long double type. So I wrote a small c program to output the max number representable by long double, and the result was that on my computer the max number for long double is approx zero! (see the attached .PNG) file. I have a Lenovo P71 computer and I'm using gcc. Does anyone have an idea why this is happening? Is it connected to my hardware?
Regards,
bostjanv

[laserlight]

You should use the corresponding format specifiers for the values that you are trying to print, e.g., the type of the result of sizeof is an unsigned integer, so you should either use %zu, or if you must compile with respect to an older version of the C standard, use %u then cast the result of sizeof to unsigned int. Likewise, for printing long double you should use %Le instead of %e:

#include <float.h>
#include <stdio.h>

int main(void) {
    long double r;

    printf("sizeof(float) %zu\nsizeof(double) %zu\nsizeof(long double) %zu\n",
        sizeof(float),
        sizeof(double),
        sizeof(long double));
    printf("FLT_MAX %e\nDBL_MAX %e\nLDBL_MAX %Le\n",
        FLT_MAX,
        DBL_MAX,
        LDBL_MAX);

    r = 1.1L + 1.2L;
    printf("r=1.1L+1.2L;\nr=%Le\n", r);
}

Notice that instead of using %s as a placeholder and then supplying the text as arguments to printf, I embedded the text directly in the format string. This is the usual approach for readability.

[bostjanv]

Thanks for your reply, and I appreciate your style suggestions; however, after making the changes you suggested I am still getting 3.302445e-317 as the value of LDBL_MAX, and the same as the value of r after the assignment r=1.1L+1.2L; (?)
Regards,
bostjanv

[Salem]

c - printf and long double - Stack Overflow
Type long double | Microsoft Docs
Gcc on windows uses the Microsoft run time library, which doesn't understand true long double data types.
You have to make it use the alternative mingw library.

[bostjanv]

Thanks. It turns out I needed two changes: gcc needs to be used with the flag -D__USE_MINGW_ANSI_STIO, and the program needs the macro
#define printf __mingw_printf

Regards,
bostjanv

[stahta01]

From memory "-D__USE_MINGW_ANSI_STIO" should be "-D__USE_MINGW_ANSI_STDIO" that is with a D added.

Tim S.

[bostjanv]

You are right; but the thing works even without the "D". I am not sure how gcc recognizes its options.
Regards,
bostjanv