🚀 go-pugleaf

theo@rascal.UUCP (1472) writes:

>In article <1992Jul16.232926.4740@rossinc.com> ciaranst@rossinc.com (ciaran storan) writes:
>>
>>Hi,
>> Does anyone out there have a rounding function for ANSI 'C', I am
>>trying to do some financial applications, and badly need a rounding
>>function.
>>
>>Any HELP??.

>Me too please.

try the following:

---- included copy of message sent to original poster of this problem ----
this may not be exactly what you are looking for, but using casts you
can don something like what I have included below. I am not in any way
an ANSI 'C' programmer, K&R is more my style, but this might at least
give you some ideas....

I tested it quickly and it works quite well. ( but I didn't bother
doing efficiency tests though, so someone else might have a quicker
hack )

( in fact I just looked and we don't even have an ANSI compiler on
this machine.. oh well..the theory is the same! )

good luck!

-----------------------------------------------------------------
Stephen Riehm				   smr@pki-nbg.philips.de
not in any way talking on behalf of:
Philips Kommunikations Industrie	  Phone: +49 911 526 2975
8500 Nuremberg, Germany			    Fax: +49 911 526 2095
I come from the land down under, where women glow and men chunder

--- cut here --- file: /home/smr/bin/.s/round.c
/*-----------------------------------------------------------------------------

	File:		round.c
	Written by:	Stephen Riehm
	Date:		20/07/92 14:50
	Description:
		round numbers

*/

/*
 * ----- includes -----
 */
#include <stdio.h>

/*
 * externs
 */
float round();

/*
 * ----- functions -----
 */
main( )
    {
    float	orig = 12.65432;	/* original number to round	*/
    float	result;			/* orig rounded to 1 dec. place */
    int		i;			/* loop counter			*/

    for( i = 0; i < 5; i++ )
	{
	result = round( orig, i );	/* round to i decimal places	*/
	printf( "%f rounded to %d places = %.*f\n", orig, i, i, result );
	}
    }

/*-----------------------------------------------------------------------------

	round: basic idea:
		multiply the number by 10^( however many decimal
		    places you want +1 )
		cast the resulting number into an int ( this strips the
		    bits behind the . )
		add 5 ( because we shifted one dec. pos. too far this will
		    only effect the outcome if the last digit is 5 or
		    greater )
		divide the integer by 10. ( this loses the 5 from the end )
		convert the number back to a float and divide by 10 ^ 
		    ( the number of decimal places ) to get the true result.

	OK its not straight forward, but it works! and its the most
	reasonable way I know of actually rounding ( and not just
	truncating )

*/
float round( orig, exp )
float	orig;
int	exp;
    {
    int		tmp_int;	/* temporary integer	*/
    float	rtn;		/* return value		*/

    tmp_int = orig * ( float )power( 10, exp + 1 );
	/*  ^-- trunctation takes place here!	*/
    tmp_int += 5;		/* now round to the NEAREST number	*/
    tmp_int /= 10;		/* strip the 5 off	*/
    /* move the decimal point back where it belongs!	*/
    rtn = ( float )tmp_int / ( float )power( 10, exp );

    return( rtn );
    }

power( base, exp )
int	base;
int	exp;
    {
    int		rtn = 1;		/* return value	*/

    while( exp-- > 0 )
	{
	rtn = rtn * base;
	}

    return( rtn );
    }
--- cut here ---