[SOLVED] Problem with negative numbers

[micmania1]

-54162.76 + 54162.76 = -1.45519152284E-11

 

and number_format(-1.45519152284E-11, 2) = -0.00

 

Why is PHP doing this?

[kickstart]

Hi

 

Think what is happening is that php is using binary arithmatic, and a number ending .76 cannot be represented in binary. Suspect that a +ve and -ve version of the number land up represented with a very slightly different binary number.

 

All the best

 

Keith

[micmania1]

Thanks for the reply.

 

Is there a way of solving the problem?

[kickstart]

Hi

 

Suspect not really avoidable short of writing custom functions for such calculations. It is a pretty common issue with computer systems that do not support decimal arithmatic (old mainframe langauges such as PL/1 and Cobol supported decimal formats to avoid issues like this).

 

If you really wanted to then you could process the number as an array of single digit decimal numbers and do all the addition / subtraction / multiplication / division a digit at a time.

 

All the best

 

Keith

[micmania1]

function decimalAddition() {
$args = func_get_args();
$number = array();
$decimals = array();
foreach ($args as $v) {
	$exp = explode('.', $v);

	$number[] = $exp[0];
	if ($exp[0] < 0) {
		$decimals[] = 0 - $exp[1];
	} else {
		$decimals[] = $exp[1];
	}

}

if (!empty($number)) {
	$number = array_sum($number);
} else {
	$number = 0;
}

if (!empty($decimals)) {
	$decimals = array_sum($decimals) / 100;
} else {
	$decimals = 0;
}

return $number + $decimals;
}

 

Problem Solved.

 

Thanks for your time.

[micmania1]

Don't use above code - its got bugs.

 

function decimalAddition() {
$args = func_get_args();

$number = array();
$decimals = array();
foreach ($args as $v) {
	$round = number_format($v, 0);

	$number[] = $round;
	$decimals[] = ($v - $round);
}

if (!empty($number)) {
	$number = array_sum($number);
} else {
	$number = 0;
}

if (!empty($decimals)) {
	$decimals = array_sum($decimals);
} else {
	$decimals = 0;
}

return $number + $decimals;
}

[Mark Baker]

use bcmath if its available