math

⚡️ Performance improvements

• BigNumber::of() is now faster, thanks to @SebastienDug in #77.

💥 Breaking changes

• Minimum PHP version is now 8.1
• RoundingMode is now an enum; if you're type-hinting rounding modes, you need to type-hint against RoundingMode instead of int now
• BigNumber classes do not implement the Serializable interface anymore (they use the new custom object serialization mechanism)
• The following breaking changes only affect you if you're creating your own BigNumber subclasses:
  • the return type of BigNumber::of() is now static
  • BigNumber has a new abstract method from()
  • all public and protected functions of BigNumber are now final

💥 Breaking changes

• Minimum PHP version is now 8.0
• Methods accepting a union of types are now strongly typed*
• MathException now extends Exception instead of RuntimeException

* You may now run into type errors if you were passing Stringable objects to of() or any of the methods internally calling of(), with strict_types enabled. You can fix this by casting Stringable objects to string first.

👌 Improvements

• BigRational::toFloat() now simplifies the fraction before performing division (#73) thanks to @olsavmic

✨ New features

• BigInteger::gcdMultiple() returns the GCD of multiple BigInteger numbers

💥 Breaking changes

• Minimum PHP version is now 7.4

🚀 Compatibility with PHP 8.1

• Support for custom object serialization; this removes a warning on PHP 8.1 due to the Serializable interface being deprecated (#60) thanks @TRowbotham

🐛 Bug fix

• Incorrect results could be returned when using the BCMath calculator, with a default scale set with bcscale(), on PHP >= 7.2 (#55).

✨ New features

• BigInteger::not() returns the bitwise NOT value

🐛 Bug fixes

• BigInteger::toBytes() could return an incorrect binary representation for some numbers
• The bitwise operations and(), or(), xor() on BigInteger could return an incorrect result when the GMP extension is not available

🐛 Bug fix

• BigInteger::toBytes() could return an incorrect binary representation for some numbers
• The bitwise operations and(), or(), xor() on BigInteger could return an incorrect result when the GMP extension is not available

👌 Improvements

• BigNumber::of() now accepts .123 and 123. formats, both of which return a BigDecimal

💥 Breaking changes

• Deprecated method BigInteger::powerMod() has been removed - use modPow() instead
• Deprecated method BigInteger::parse() has been removed - use fromBase() instead

🚑 Critical fix

• This version reintroduces the deprecated BigInteger::parse() method, that has been removed by mistake in version 0.8.9 and should have lasted for the whole 0.8 release cycle.

✨ New features

• BigInteger::modInverse() calculates a modular multiplicative inverse
• BigInteger::fromBytes() creates a BigInteger from a byte string
• BigInteger::toBytes() converts a BigInteger to a byte string
• BigInteger::randomBits() creates a pseudo-random BigInteger of a given bit length
• BigInteger::randomRange() creates a pseudo-random BigInteger between two bounds

💩 Deprecations

• BigInteger::powerMod() is now deprecated in favour of modPow()

🐛 Fixes

• added missing ext-json requirement, due to BigNumber implementing JsonSerializable

⚡️ Optimizations

• additional optimization in BigInteger::remainder()

✨ New features

• BigInteger::getLowestSetBit() returns the index of the rightmost one bit

✨ New features

• BigInteger::isEven() tests whether the number is even
• BigInteger::isOdd() tests whether the number is odd
• BigInteger::testBit() tests if a bit is set
• BigInteger::getBitLength() returns the number of bits in the minimal representation of the number

🛠️ Maintenance release

Classes are now annotated for better static analysis with psalm.

This is a maintenance release: no bug fixes, no new features, no breaking changes.

✨ New feature

BigInteger::powerMod() performs a power-with-modulo operation. Useful for crypto.

✨ New feature

BigInteger::mod() returns the modulo of two numbers. The modulo differs from the remainder when the signs of the operands are different.

⚡️ Performance improvements

A few additional optimizations in BigInteger and BigDecimal when one of the operands can be returned as is. Thanks to @tomtomsen in #24.

🐛 Bug fixes

• BigInteger::toBase() could return an empty string for zero values (BCMath & Native calculators only, GMP calculator unaffected)

✨ New features

• BigInteger::toArbitraryBase() converts a number to an arbitrary base, using a custom alphabet
• BigInteger::fromArbitraryBase() converts a string in an arbitrary base, using a custom alphabet, back to a number

These methods can be used as the foundation to convert strings between different bases/alphabets, using BigInteger as an intermediate representation.

💩 Deprecations

• BigInteger::parse() is now deprecated in favour of fromBase()

BigInteger::fromBase() works the same way as parse(), with 2 minor differences:

• the $base parameter is required, it does not default to 10
• it throws a NumberFormatException instead of an InvalidArgumentException when the number is malformed