I'd like to propose a new standard for creating algebraic data types. Consider the following.
const Maybe = {}; // Static Land Canonical Module
const Nothing = (() => {
function Nothing() {}
Nothing.prototype["static-land/canonical"] = Maybe;
return new Nothing; // Singleton Pattern
})();
function Just(value) {
if (this instanceof Just) this.value = value;
else return new Just(value); // Hack for calling the constructor without `new`.
}
Just.prototype["static-land/canonical"] = Maybe;
The primary advantage of defining algebraic data types like we did above, is good console logs.
> Nothing
Nothing {}
> Just(10)
Just { value: 10 }
Pattern matching is also standardized. You can use pattern matching with built-in types too.
Maybe.map = (mapping, maybe) => {
switch (maybe.constructor.name) {
case "Nothing": return Nothing;
case "Just": return new Just(mapping(maybe.value)); // Using `new` for performance.
}
};
We can also create a utility function which makes defining new data constructors less verbose.
const data = (type, name, ...keys) => {
const {length} = keys;
function Data(...vals) {
if (this instanceof Data)
for (let i = 0; i < length; i++)
this[keys[i]] = vals[i];
else return new Data(...vals);
}
Object.defineProperties(Data, {
name: { value: name },
length: { value: length }
});
Data.prototype["static-land/canonical"] = type;
return length > 0 ? Data : new Data;
};
This makes it easy to define new data constructors for algebraic data types.
const Maybe = {};
const Nothing = data(Maybe, "Nothing");
const Just = data(Maybe, "Just", "value");
I would love to hear your thoughts on this, and continue the discussion in #45 here.
I'd like to propose a new standard for creating algebraic data types. Consider the following.
The primary advantage of defining algebraic data types like we did above, is good console logs.
Pattern matching is also standardized. You can use pattern matching with built-in types too.
We can also create a utility function which makes defining new data constructors less verbose.
This makes it easy to define new data constructors for algebraic data types.
I would love to hear your thoughts on this, and continue the discussion in #45 here.