Executing bcsacle() will change the scale value of fpm.conf, not only the current process.
(PHP 4, PHP 5, PHP 7, PHP 8)
bcscale — Set or get default scale parameter for all bc math functions
Sets the default scale parameter for all subsequent calls to bc math functions that do not explicitly specify a scale parameter.
Gets the current scale factor.
scale
The scale factor.
Returns the old scale when used as setter. Otherwise the current scale is returned.
Version | Description |
---|---|
8.0.0 |
scale is now nullable.
|
7.3.0 |
bcscale() can now be used to get the current scale
factor; when used as setter, it now returns the old scale value.
Formerly, scale was mandatory, and
bcscale() always returned true .
|
Example #1 bcscale() example
<?php
// default scale : 3
bcscale(3);
echo bcdiv('105', '6.55957'); // 16.007
// this is the same without bcscale()
echo bcdiv('105', '6.55957', 3); // 16.007
?>
Executing bcsacle() will change the scale value of fpm.conf, not only the current process.
These functions DO NOT round off your values. No arbitrary precision libraries do it this way. It stops calculating after reaching scale of decimal places, which mean that your value is cut off after scale number of digits, not rounded. To do the rounding use something like this:
<?php
function bcround($number, $scale=0) {
$fix = "5";
for ($i=0;$i<$scale;$i++) $fix="0$fix";
$number = bcadd($number, "0.$fix", $scale+1);
return bcdiv($number, "1.0", $scale);
}
?>
Simple, easy way to remove excess trailing zeros using rtrim:
<php>
// $total is the result of a bcmath calculation
if ( strpos($total, '.') !== false ):
$total = rtrim($total, '0');
$total = rtrim($total, '.');
endif;
</php>
edited bcround function with negative number support:
<?php
function bcround($number, $scale=0) {
if($scale < 0) $scale = 0;
$sign = '';
if(bccomp('0', $number, 64) == 1) $sign = '-';
$increment = $sign . '0.' . str_repeat('0', $scale) . '5';
$number = bcadd($number, $increment, $scale+1);
return bcadd($number, '0', $scale);
}
?>
Does not work with big scale values, but accept INT as type, but can not process the max INT value, by example for calculate the cube root:
<?php
echo pow(1331, 1/3); // 11 (11^3=1331)
echo bcpow(1331, bcdiv(1, 3, PHP_INT_MAX)); // 1 ??
echo bcdiv(1, 3, PHP_INT_MAX); // 0 ??
To remove trailing zeros when using large bcscale number can be done by casting to float when ready to display the number.
<?php
bcscale(15);
$a=123.456;
$b=0.123;
$_ab=bcadd($a,$b);
echo $_ab;
// 123.579000000000000
$ab = (float)$_ab;
echo $ab;
// 123.579
?>
If you don't set the default scale, be careful when you're chaining together several BC math functions - since by default, these functions will round off your values, losing accuracy very quickly:
<?php
$a = 1.234
$b = 2.345
$c = 7.890
$ab = bcmul($a,$b); // 2
$abc = bcmul($ab,$c);
echo $abc; // 15
?>
... compare with the answer you get when you use more decimal places:
<?php
$a = 1.234
$b = 2.345
$c = 7.890
bcscale(15);
$ab = bcmul($a,$b); // 2.893730
$abc = bcmul($ab,$c);
echo $abc; // 22.83152970
?>
Previously stated one liner trailing 0 removing using rtrim() works fine except following class of subcases:
<?php
echo rtrim('100.0000', '0.'); // 1 instead 100
echo rtrim('1230.00000000', '0.'); // 123 instead 1230
?>
What seems to be working to me is using regular expression replacement. Althought it`s probably not that fast as rtrim(), I hope it should be more reliable:
<?php
function removeFloatNumStringZeroTrailer($input) {
$patterns = array('/[\.][0]+$/','/([\.][0-9]*[1-9])([0]*)$/');
$replaces = array('','$1');
return preg_replace($patterns,$replaces,$input);
}
echo removeFloatNumStringZeroTrailer('100.0000'); // says 100
echo removeFloatNumStringZeroTrailer('1230.00000000'); // says 1230
?>