I've inherited a project written in PHP which has a double entry style cashbook, which requires that the amount going in is equal to the amount going out.
When validating figures entered via a form (to work out if it equals zero), it runs the code below which works fine until there is a decimal number, which seems to make it randomly fall over and return an incorrect value.
In the example below, it returns a really tiny number from 53.87 - 53.87, which should be zero.This has been stripped back to eliminate other things from causing problems, so I've removed a lot of input validation etc.
<?php $inputs = array(
array(
"in" => '',
"out" => '249.6',
),
array(
"in" => '',
"out" => '396',
),
array(
"in" => '554.4',
"out" => ''
),
array(
"in" => '145.07',
"out" => ''
),
array(
"in" => '',
"out" => '53.87',
),
);
$fTotal = 0;
echo "Start at 0: ";
foreach($inputs as $key=>$sRef) {
$itemValid = true;
$aItem = array();
$amountIn = $inputs[$key]['in'];
$amountOut = $inputs[$key]['out'];
if ($itemValid) {
echo "'".$fTotal."'";
$is_in = 0;
if ($amountIn > 0.0) {
$is_in = 1;
echo "+";
$amount = $amountIn;
$fTotal = $fTotal + $amountIn;
} else {
$is_in = 0;
echo "-";
$amount = $amountOut;
$fTotal = $fTotal - $amountOut;
}
echo "'".$amount."'=";
echo "'".$fTotal."' | ";
$aItem["is_in"] = $is_in;
$aItem["amount"] = $amount;
$aItems[] = $aItem;
}
}
You can run this on a sandbox here.
Here is the expected output:
Start at 0: '0'-'249.6'='-249.6' | '-249.6'-'396'='-645.6' | '-645.6'+'554.4'='-91.2' | '-91.2'+'145.07'='53.87' | '53.87'-'53.87'='0' |
Here is the actual output:
Start at 0: '0'-'249.6'='-249.6' | '-249.6'-'396'='-645.6' | '-645.6'+'554.4'='-91.2' | '-91.2'+'145.07'='53.87' | '53.87'-'53.87'='-4.9737991503207E-14' |
What is wrong here?
Update
Following help below, here's the working code for anyone who stumbles on this in the future.
This is happening because of floating point (decimal) maths where computers like binary maths. Sometimes decimal numbers don't have a nice representation in binary so these tiny differences happen.
If you can state that every number can be rounded to 2 decimal points, and based on your limited dataset, wrapping your 'sums' in number_format($fTotal + $amountIn , 2)
etc can sort this out for you.
Alternatively for a "more accurate", just wrap the final one eg. echo "'".number_format($fTotal, 2)."' | ";
(or 0 or whatever)