Can I Rely on PHP php.ini Precision Workaround for Floating Point Issue?

Introduction

Floating-point arithmetic, a complex subject often misunderstood, pervades modern computer systems. As most fractional numbers lack an exact binary representation, rounding unavoidably occurs. Understanding the nuances of floating-point operations is crucial, as outlined in "What Every Computer Scientist Should Know About Floating-Point Arithmetic."

Questions and Answers

Question 1: Can I use the precision workaround for precise 2-digit calculations (e.g., money)?

Answer: No, you cannot rely on php.ini precision settings for consistent 2-digit accuracy, even for numbers below 10^6, because the precision length can increase during calculations.

Question 2: Provide a clear example where the precision workaround fails.

Answer: Consider the following example with a 10% discount applied only to the first kilogram:

ini_set('precision', 8); // Your precision
$a = 5.88; // Cost of 1kg
$q = 2.49; // User buys 2.49 kg
$b = $a * 0.01; // 10% Discount only on the first kg
echo ($a * $q) - $b;

Output: 14.5824

Question 3: Which php.ini.precision value is suitable for 2-digit money calculations?

Answer: PHP precision settings are not a reliable basis for financial calculations or floating-point precision. It's better to use specialized libraries or methods.

Example Calculations

Let's examine various examples to illustrate the accuracy limitations:

$a = 342349.23;
$b = 341765.07;

// Example A: precision = 20
echo $a - $b, PHP_EOL; // 584.15999999997438863
echo floatval(round($a - $b, 2)), PHP_EOL; // 584.15999999999996817 (rounding introduces instability)
echo number_format($a - $b, 2), PHP_EOL; // 584.16
echo bcsub($a, $b, 2), PHP_EOL; // 584.15

// Example B: precision = 14
echo $a - $b, PHP_EOL; // 584.15999999997
echo floatval(round($a - $b, 2)), PHP_EOL; // 584.16
echo number_format($a - $b, 2), PHP_EOL; // 584.16
echo bcsub($a, $b, 2), PHP_EOL; // 584.16

// Example C: precision = 6
echo $a - $b, PHP_EOL; // 584.16
echo floatval(round($a - $b, 2)), PHP_EOL; // 584.16
echo number_format($a - $b, 2), PHP_EOL; // 584.16
echo bcsub($a, $b, 2), PHP_EOL; // 584.00

// Example D: precision = 3
echo $a - $b, PHP_EOL; // 584
echo floatval(round($a - $b, 2)), PHP_EOL; // 584
echo number_format($a - $b, 2), PHP_EOL; // 584.16 (consistent)
echo bcsub($a, $b, 2), PHP_EOL; // 0.00

These examples demonstrate the inconsistencies in relying on PHP precision settings for accurate calculations.

Conclusion

For reliable financial calculations, forget about floating-point and use cents or consider the number_format function for consistent 2-digit accuracy.

Update

ini_set('precision', 8);

$a = 0.19;
$b = 0.16;
$c = 0.01;
$d = 0.01;
$e = 0.01;
$f = 0.01;
$g = 0.01;