Few Questions

[SourceCode]

Hi I have a function that reads a double and I'm wondering how can i check for overflow and underflow? Like say I have double x; and theres a value stored in x that I want to check. I tryed comparing it to DBL_MAX and DBL_MIN but it doesn't work.

Also another question
I ran into this

const char * const  x;
and
char const *x;

I read that these both declare the pointer x to be a constant and that the top one declares the value pointed to by x as constant. I understand the part about the value being constant but the part about the pointers being constant confuses me. What does that mean, that the pointer is constant and that its memory address doesn't change? If so is it good to declare pointers like this if I don't plan on changing their address? Just seems weird cause I've never seen that before.

[Vber]

about your double question, use float.h to discover DBL_MAX;

#include <float.h> /* for DBL_MIN and DBL_MAX */
#include <stdio.h>

int main( void ) {
	double db1 = 2500.2000;
	int i;

	for( i=0; i<50; i++)
	{
		db1 *= 5000000000.22;
		if (db1 >= DBL_MAX)
			puts( "overflow" );
		else
			puts( "still ok" );
	}

	return 0;
}

Sorry about the code, is dumbass but DBL_MAX in my system is gicantic huhauah, same thing with DBL_MIN;

char* const declares a constant pointer which has both read and write access to a character (or character array). The pointer itself is a constant and you can not change it. Like all other constant variables, you must initialize it with a constant value at the same time when it is declared:

const char* declares a pointer to a constant character (or a constant character array). The pointer can be changed, but the character (or array) to which it points can not be changed.

[SourceCode]

Hey thanks alot for the explanation on constant pointers, great stuff. I played with strtod and checked errno for ERANGE and it works, and I also did it the other way with just comparing DBL_MAX and DBL_MIN. And yea Vber its a HUGE number on my computer also lol. About the overflow thing well I think it's something else and not overflow. Cause if I enter this

12345678901234567

I get this when I print it out.

12345678901234568.000000

It's always the 17th digit consistently that the discrepancy begins at. Overflow or Underflow doesn't happen until I enter a ton of stuff, so that works, but the above must not be considered that. Are floating point numbers just not exact after a certain amount of digits?

This is what I was using to test.

#include <stdlib.h>
#include <stdio.h>
#include <float.h>
#include <errno.h>
#include <string.h>

static int readdouble(double *number)
{
    int value;
    int ch;
    
    if ((value = scanf("%lf", number)) == 1 && (ch = getchar()) == '\n') {
	ungetc(ch, stdin);
	if (*number <= DBL_MAX && *number >= DBL_MIN) {
	    return 1;
	}
	else {
	    return 0;
	}
    }
	    
    if (value == EOF)
	return EOF;
	
    if (value == 0)
    {
	ungetc(ch, stdin);
	return 0;
    }
}

int 
main(void)
{
    double number; 
    int value;
    char array[BUFSIZ];
    char *replace;
    char *p;

/*
    if ((value = readdouble(&number)) == 1)
	printf("Valid number %f\n", number);
    else if (value == EOF)
	printf("end-of-file\n");
    else if( value == 0)
	printf("Invalid number, overflow, or underflow: %f\n", number); 
*/   
 
    if (fgets(array, sizeof array, stdin) != NULL) {
	if ((replace = strchr(array, '\n')) != NULL) {
	    *replace = '\0';
	    number = strtod(array, &p);
	    if (array[0] != '\0' && *p == '\0') {
		if (errno != ERANGE)
		    printf("Valid number finally ! %f\n", number);
		else
		    printf("We overflowed or underflowed: %f\n", number);
	    }
	    else {
		printf("Only a newline was entered or we didn't read a number!\n");
	    }		    
	}
	else {
	    printf("Newline wasn't read!\n");
	} 
    }
    else {
	printf("end-of-file encountered\n");
    }
    
    return (0);
    
}

[stumon]

>>Are floating point numbers just not exact after a certain amount of digits?

Yes, with only 8 bytes available on most systems, there is some exponential equation, lol, i dont know it off hand but you only get about 17 precise number places.

--edit--
Did a search on google and found the exponential form: 1.7E+308. A good explination of the data types can be found here.

[SourceCode]

Hey stumon thanks

[stumon]

You are welcome.

[stumon]

>>12345678901234567
>>I get this when I print it out.
>>12345678901234568.000000

He is having precision problems.....I shouldn't have been confusing him.