Skip to content

keithamus/ecmascript-operator-overloading-proposal

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

1 Commit
 
 
 
 

Repository files navigation

Symbol.operator

This proposal introduces the Symbol.operator and Symbol.unaryOperator functions to the ECMAScript Standard Library. The Symbol.operator and Symbol.unaryOperator functions can be used to extend prototypes to overload operator behaviours, with a future scope of being able to easily add new operators.

Symbol.operator and Symbol.unaryOperator behave a lot like Symbol.for in that they search for existing symbols in a runtime-wide symbol registry with the given key and returns it if found. However, these registries exist separately from Symbol.for - so calling Symbol.for('+') will return a different symbol to Symbol.operator('+') which will return a different symbol to Symbol.unaryOperator('+').

Where an operator is used with two operands of the same type, a lookup begins for <rho>.__proto__.constructor[Symbol.operator(<operator>)] (for example, the + operator will look up <rho>.__proto__.constructor[Symbol.operator('+')]). If the property exists and is a function, then the function is called, with the left hand operand as the first argument and the right hand operand as its second argument. The function's return value is what will be used as the result for the operarion.

Where an operator is used with one operand, for example the ! operator, the single operand will be evaluated to check the existance of the <operand>.__proto__.constructor[Symbol.unaryOperator(<operator>)] property (for example the ! operator will look up <operand>.__proto__.constructor[Symbol.unaryOperator(!)]). If the property exists and is a function, then the function is called with the operand as the first and only argument. The function's return value will be used as the result of the operation.

Operators may additionally coerce return values for comformity across the platform. For example the ! operator will coerce any return value ToBoolean. As another exampe, the unary + operator will always coerce ToNumber to maintain compatibility with existing expecation in areas like asm.js.

Supported Operators

The proposed supported operators are:

Exceptions

The following operators are not supported:

  • Loose inequality Comparison: != (Symbol.operator('!=')) (coerces return value ToBoolean) This is supported through negating the value of the loose equality comparison (==) for consistency.

  • Assignment operator: =. This is not supported in any way and behaviour will remain the same as older versions of the spec.

  • Strict equality/inequality comparison: ===, !==. This is not supported in any way, and behaviour will remain the same as older versions of the spec.

  • Any compound assignment operator, e.g. +=, -=. These are supported through their non-compound equivalents, for example a += b will desugar to a = a + b which will in turn run the steps for the + operator.

  • The delete operator. This is already dealt with via Proxies and defineProperty

  • The void operator. This discards the value and so is pointless to implement.

  • Logical && or || operators. These are not supported in any way and behaviour will remain the same as older versions of the spec.

Examples

Example 1

class Thinger() {
  constructor(...things) {
    this.things = things
  }
  
  static [Symbol.operator('+')](leftHandOperand, rightHandOperand) {
    return new Thinger(...[...leftHandOperand.things, ...rightHandOperand.things])
  }

}

const thingerA = Thinger('a')
const thingerB = Thinger('b')
const thingerAB = thingerA + thingerB
console.assert(thingerAB instanceof Thinger, 'the returned value from the `+` operator is a new Thinger instance')
console.assert(thingerAB.things[0] === 'a' && thingerAB.things[1] === 'b', 'the returned instance is created from the logic of the static method against both operands')

console.assert((thingerA + 1) === 1)

Future: Extending of operators

Should one want to add an operator to the language, it could become as simple as implementing a new symbol a prototype; for example the bind operator :: could simply be polyfilled with the following:

Function[Symbol.operator('::')] = function (lho, rho) {
  return rho.bind(lho)
}

How could this even work?

When parsing a piece of javascript, when the tokenizer reached an unrecognised token between one or two operands (with, perhaps a whitelist of certain allowed operator characters, for example any Unicode character betwixt U+2000–U+218F), it could search in the operator registry (perhaps at runtime? during evaluation?) for the operator, and if not found could raise a SyntaxError.

About

A proposal for operator overloading in ECMAScript

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published