rational

• Explicitly throw an ArgumentError when a Rational is created with a zero denominator.

• Add Rational.tryParse.

The goal of this version is to move several decimal methods back to the decimal package and to have sharper types on the API.

It introduces several breaking changes.

• ~/, round(), floor(), ceil(), truncate() now return a BigInt. If you need a Rational you can convert the BigInt to Rational with bigint.toRational().
• Removal of toDecimalString(). You can replace it with rational.toDecimal(scaleOnInfinitePrecision: 10).toString() using the decimal package.
• Removal of isNaN getter. It was always returning false.
• Removal of isInfinite getter. It was always returning false.
• Removal of isNegative getter. You can replace it with rational < Rational.zero.
• Removal of roundToDouble(), floorToDouble(), ceilToDouble(), truncateToDouble(). You can replace them by round().toDouble(), floor().toDouble(), ceil().toDouble(), truncate().toDouble().
• Removal of toInt(). You can replace it with rational.toBigInt().toInt().
• Removal of hasFinitePrecision getter. This getter (provided as extension on Rational) is now part of the decimal package.
• Removal of precision and scale getters. They was only available on rational returning true on hasFinitePrecision (ie. decimal numbers). You can replace them with rational.toDecimal().precision and rational.toDecimal().scale using the decimal package.
• Removal of toStringAsFixed, toStringAsExponential and toStringAsPrecision. You can replace them with rational.toDecimal(scaleOnInfinitePrecision: xxx).toStringAsFixed,rational.toDecimal(scaleOnInfinitePrecision: xxx).toStringAsExponential and rational.toDecimal(scaleOnInfinitePrecision: xxx).toStringAsPrecision using the decimal package.

Other changes:

• Add extension method toRational() on int.
• Add extension method toRational() on BigInt.

• Improve performance of several methods by working around the issue BigInt.gcd() is really slow with a custom gcd implementation based on Euclidean algorithm.

• Improve parsing of number with big exponent part. However the exponent part must now be parsable as an int.

• Allow negative value as exponent of pow.

• Fix the doc of pow.

• Stable null safety release.

• Migrate to nullsafety.

• Improve pub score.

• Replace gcd implementation with dart:core's.

• add Rational.pow.

• add Rational.zero and Rational.one.
• add Rational.inverse.

• allow numbers starting with dot.

• fix issue with signnum.

• fix issue with toStringAsPrecision.

• migration to Dart 2.

• allow parsing of 1..
• make Rational.parse a factory constructor.

• move to Dart SDK 2.0
• remove BigInt class
• use BigInt provided by dart:core

• add types.

• fix bug on operator % with negative values.

• fix bug on operator % with negative values on browser.

• fix a bug on BigInt.toDouble.

• fix bugs with dart2js
• add Rational.signum
• add Rational.hasFinitePrecision
• add Rational.precision
• add Rational.scale

• fix a bug for Rational.precision on negative number.

• Rational.parse accepts strings in scientific notation (eg. 1.5e-3).

• BigInt.parse accepts an optional prepending + for positive integers.