How to define constant in floating point double precision in c89

Question

I have code with a constant declared in a representation that is not valid in c89 (compiler option on historic project).

#define K_MAX_KCG_REAL 0x1.FFFFFFFFFFFFFp1023

I am looking for a solution valid in c89

I have tried

#define K_MAX_KCG_REAL 0x7FEFFFFFFFFFFFFF

but is is interpreted as an integer with a float value of approx. 9.22e18. Far from 1.79e308 that I need. What is the best way to declare a coonstant with max value for double precision ?

Use type-punning, possibly through a union or a byte buffer? Or why not translate it to the actual `double` value it represents? — Some programmer dude, Jul 08 '19 at 12:54
There is a macro DBL_MAX defined in : https://stackoverflow.com/questions/5834635/how-do-i-get-double-max — Dali, Jul 08 '19 at 12:56
Basically the best way is to ditch that compiler. Or ditch that job. Perhaps even in reverse order! C99 is *only* 20 years old! — Antti Haapala -- Слава Україні, Jul 08 '19 at 13:20
thanks for the link but this is critical embedded code, I can't add external header files — Marine Jouanneau, Jul 08 '19 at 13:28

score 2 · Answer 1 · answered Jul 08 '19 at 18:41

The number you are looking for is 179769313486231570814527423731704356798070567525844996598917476803157260780028538760589558632766878171540458953514382464234321326889464182768467546703537516986049910576551282076245490090389328944075868508455133942304583236903222948165808559332123348274797826204144723168738177180919299881250404026184124858368.

If DBL_MAX is defined in <float.h>, you should use that.

With many compilers, 1.7976931348623157e308 would suffice.

If the compiler fails to parse those correctly, you could try (9007199254740991. * 1073741824. * 1073741824. * 1073741824. * 1073741824. * 1073741824. * 1073741824. * 1073741824. * 1073741824. * 1073741824. * 1073741824. * 1073741824. * 1073741824. * 1073741824. * 1073741824. * 1073741824. * 1073741824. * 1073741824. * 1073741824. * 1073741824. * 1073741824. * 1073741824. * 1073741824. * 1073741824. * 1073741824. * 1073741824. * 1073741824. * 1073741824. * 1073741824. * 1073741824. * 1073741824. * 1073741824. * 1073741824. * 2048.)

score 0 · Answer 2 · answered Jul 08 '19 at 12:54

0

You'll need to create a union containing a unsigned long long and a double and initialize the former.

const union {
    unsigned long long i;
    double d;
} K_MAX_KCG_REAL = { 0x7FEFFFFFFFFFFFFF };

answered Jul 08 '19 at 12:54

dbush

205,898
23
218
273

2

`unsigned long long` would be a non-standard, extended type in c89/c90 though. – Ian Abbott Jul 08 '19 at 13:04
1

I don't think `unsigned long long` is standard per C89. It's mentioned as an extension under [**A.6.5 Common extensions**](https://port70.net/~nsz/c/c89/c89-draft.html). – Andrew Henle Jul 08 '19 at 13:06

How to define constant in floating point double precision in c89

2 Answers2