/flyd

The minimalistic but powerful, modular, functional reactive programming library in JavaScript.

Primary LanguageJavaScriptMIT LicenseMIT

Flyd

The modular, KISS, functional reactive programming library for JavaScript.

Build Status Coverage Status Join the chat at https://gitter.im/paldepind/flyd

Table of contents

Introduction

Functional reactive programming is a powerful programming paradigm for expressing values that change over time. But existing libraries for JavaScript are huge, complex, have a high learning curve and aren't functional enough.

Flyd is simple and expressive. It has a minimal but powerful core on top of which new abstractions can be built modularly.

Features

Main features

  • Simple but powerful. Less is more! Flyd provides combinable observable streams as the basic building block. This minimal core is less than 200 SLOC which makes the library transparent – end users can realistically get a full understanding of how the library works.
  • More functional in style. Flyd is more functional than existing FRP libraries. Instead of methods it gives you curried functions with arguments in the order suitable for partial application. This gives more expressive power and modularity.
  • Modularity. The core of the Flyd is powerful and documented. This makes it easy for users of the library to create new FRP abstractions if existing ones do not exist. This in turn makes it viable to capture more patterns than otherwise because they can exist as separate modules. List of existing modules.

Other features

Examples

For other examples check the source code of the modules.

Tutorial

This is not general introduction to functional reactive programming. For that take a look at The introduction to Reactive Programming you've been missing and/or this Elm tutorial if you are comfortable with reading Haskell-like code.

This is not a demonstration of how you would write code with Flyd on a day to day basis. For that take a look at the examples.

This tutorial will however introduce you to the minimal but powerful core that Flyd provides and show you how it can be used to build FRP abstractions.

Creating streams

Flyd gives you streams as the building block for creating reactive dataflows. They serve the same purpose as what other FRP libraries call Signals, Observables, Properties and EventEmitters.

The function flyd.stream creates a representation of a value that changes over time. The resulting stream is a function. At first sight it works a bit like a getter-setter:

// Create a stream with initial value 5.
var number = flyd.stream(5);
// Get the current value of the stream.
console.log(number()); // logs 5
// Update the value of the stream.
console.log(number(7));
// The stream now returns the new value.
console.log(number()); // logs 7

Top level streams, that is streams without dependencies, should typically depend on the external world, like user input or fetched data.

Since streams are just functions you can easily plug them in whenever a function is expected.

var clicks = flyd.stream();
document.getElementById('button').addEventListener('click', clicks);
var messages = flyd.stream();
webSocket.onmessage = messages;

Clicks events will now flow down the clicks stream and WebSockets messages down the messages stream.

Dependent streams

Streams can depend on other streams. Use var combined = flyd.combine(combineFn, [a, b, c, ...]). The combineFn function will be called as (a, b, c, ..., self, changed) => v, where a, b, c, ... is a spread of each dependency, self is a reference to the combine stream itself, and changed is an array of streams that were atomically updated.

Flyd automatically updates the stream whenever a dependency changes. This means that the sum function below will be called whenever x and y changes. You can think of dependent stream as streams that automatically listens to or subscribes to their dependencies.

// Create two streams of numbers
var x = flyd.stream(4);
var y = flyd.stream(6);
// Create a stream that depends on the two previous streams
// and with its value given by the two added together.
var sum = flyd.combine(function(x, y) {
  return x() + y();
}, [x, y]);
// `sum` is automatically recalculated whenever the streams it depends on changes.
x(12);
console.log(sum()); // logs 18
y(8);
console.log(sum()); // logs 20

Naturally, a stream with dependencies can depend on other streams with dependencies.

// Create two streams of numbers
var x = flyd.stream(4);
var y = flyd.stream(6);
var squareX = flyd.combine(function(x) {
  return x() * x();
}, [x]);
var squareXPlusY = flyd.combine(function(y, squareX) {
  return y() + squareX();
}, [y, squareX]);
console.log(squareXPlusY()); // logs 22
x(2);
console.log(squareXPlusY()); // logs 10

The body of a dependent stream is called with the spread of: each dependency, itself, and a list of the dependencies that have changed since its last invocation (due to atomic updates several streams could have changed).

// Create two streams of numbers
var x = flyd.stream(1);
var y = flyd.stream(2);
var sum = flyd.combine(function(x, y, self, changed) {
  // The stream can read from itself
  console.log('Last sum was ' + self());
  // On the initial call no streams has changed and `changed` will be []
  changed.map(function(s) {
    var changedName = (s === y ? 'y' : 'x');
    console.log(changedName + ' changed to ' + s());
  });
  return x() + y();
}, [x, y]);

Note Returning undefined in the combineFn will not trigger an update to the stream. To trigger on undefined, update directly:

flyd.combine((_, self, changed) => { self(undefined); }, [depStream]);

Using callback APIs for asynchronous operations

Instead of returning a value a stream can update itself by calling itself. This is handy when working with APIs that takes callbacks.

var urls = flyd.stream('/something.json');
var responses = flyd.combine(function(urls, self) {
  makeRequest(urls(), self);
}, [urls]);
flyd.combine(function(responses) {
  console.log('Received response!');
  console.log(responses());
}, [responses]);

Note that the stream that logs the responses from the server should only be called after an actual response has been received (otherwise responses() would return undefined). Fortunately a stream's body will not be called before all of its declared streams has received a value (this behaviour can be circumvented with flyd.immediate).

Using promises for asynchronous operations

Flyd has inbuilt support for promises. Similarly to how a promise can never be resolved with a promise, a promise can never flow down a stream. Instead the fulfilled value of the promise will be sent down the stream.

var urls = flyd.stream('/something.json');
var responses = flyd.stream(requestPromise(urls()));
flyd.on(function(responses) {
  console.log('Received response!');
  console.log(responses());
}, responses);

Mapping over a stream

You've now seen most of the basic building block which Flyd provides. Let's see what we can do with them. Let's write a function that takes a function and a stream and returns a new stream with the function applied to every value emitted by the stream. In short, a map function.

var mapStream = function(f, s) {
  return flyd.combine(function(s) {
    return f(s());
  }, [s]);
};

We simply create a new stream dependent on the first stream. We declare the stream as a dependency so that our stream won't return values before the original stream produces its first value.

Flyd includes a similar map function as part of its core.

Scanning a stream

Lets try something else: a scan function for accumulating a stream! It could look like this:

var scanStream = function(f, acc, s) {
  return flyd.combine(function(s) {
    acc = f(acc, s());
    return acc;
  }, [s]);
};

Our scan function takes an accumulator function, an initial value and a stream. Every time the original stream emits a value we pass it to the accumulator function along with the accumulated value.

Flyd includes a scan function as part of its core.

Stream endings

When you create a stream with flyd.stream it will have an end property which is also a stream. That is an end stream:

var s = flyd.stream();
console.log(flyd.isStream(s.end)); // logs `true`

You can end a stream by pushing true into its end stream:

var s = flyd.stream();
s.end(true); // this ends `s`

When you create a dependent stream its end stream will initially depend on all the end streams of its dependencies:

var n1 = flyd.stream();
var n2 = flyd.stream();
var sum = flyd.combine(function(n1, n2) {
  return n1() + n2();
}, [n1, n2]);

sum.end now depends on n1.end and n2.end. This means that whenever one of the sums dependencies end sum will end as well.

You can change what a stream's end stream depends on with flyd.endsOn:

var number = flyd.stream(2);
var killer = flyd.stream();
var square = flyd.endsOn(flyd.merge(number.end, killer), flyd.combine(function(number) {
  return number() * number();
}, [number]));

Now square will end if either number ends or if killer emits a value.

The fact that a stream's ending is itself a stream is a very powerful concept. It means that we can use the full expressiveness of Flyd to control when a stream ends. For an example, take a look at the implementation of takeUntil.

Fin

You're done! To learn more check out the API, the examples and the source of the modules.

API

flyd.stream()

Creates a new top level stream.

Signature

a -> Stream a

Example

var n = flyd.stream(1); // Stream with initial value `1`
var s = flyd.stream(); // Stream with no initial value

flyd.combine(body, dependencies)

Creates a new dependent stream.

Signature

(...Stream * -> Stream b -> b) -> [Stream *] -> Stream b

Example

var n1 = flyd.stream(0);
var n2 = flyd.stream(0);
var max = flyd.combine(function(n1, n2, self, changed) {
  return n1() > n2() ? n1() : n2();
}, [n1, n2]);

###flyd.isStream(stream)

Returns true if the supplied argument is a Flyd stream and false otherwise.

Signature

* -> Boolean

Example

var s = flyd.stream(1);
var n = 1;
flyd.isStream(s); //=> true
flyd.isStream(n); //=> false

###flyd.immediate(stream)

By default the body of a dependent stream is only called when all the streams upon which it depends has a value. immediate can circumvent this behaviour. It immediately invokes the body of a dependent stream.

Signature

Stream a -> Stream a

Example

var s = flyd.stream();
var hasItems = flyd.immediate(flyd.combine(function(s) {
  return s() !== undefined && s().length > 0;
}, [s]);
console.log(hasItems()); // logs `false`. Had `immediate` not been
                         // used `hasItems()` would've returned `undefined`
s([1]);
console.log(hasItems()); // logs `true`.
s([]);
console.log(hasItems()); // logs `false`.

###flyd.endsOn(endStream, s)

Changes which endsStream should trigger the ending of s.

Signature

Stream a -> Stream b -> Stream b

Example

var n = flyd.stream(1);
var killer = flyd.stream();
// `double` ends when `n` ends or when `killer` emits any value
var double = flyd.endsOn(flyd.merge(n.end, killer), flyd.combine(function(n) {
  return 2 * n();
}, [n]);

###flyd.map(fn, s)

Returns a new stream consisting of every value from s passed through fn. I.e. map creates a new stream that listens to s and applies fn to every new value.

Signature

(a -> result) -> Stream a -> Stream result

Example

var numbers = flyd.stream(0);
var squaredNumbers = flyd.map(function(n) { return n*n; }, numbers);

###flyd.on(fn, s)

Similar to map except that the returned stream is empty. Use on for doing side effects in reaction to stream changes. Use the returned stream only if you need to manually end it.

Signature

(a -> result) -> Stream a -> Stream undefined

Example

var numbers = flyd.stream(0);
flyd.on(function(n) { console.log('numbers changed to', n); }, numbers);

###flyd.scan(fn, acc, stream)

Creates a new stream with the results of calling the function on every incoming stream with and accumulator and the incoming value.

Signature

(a -> b -> a) -> a -> Stream b -> Stream a

Example

var numbers = flyd.stream();
var sum = flyd.scan(function(sum, n) { return sum+n; }, 0, numbers);
numbers(2)(3)(5);
sum(); // 10

###flyd.merge(stream1, stream2)

Creates a new stream down which all values from both stream1 and stream2 will be sent.

Signature

Stream a -> Stream a -> Stream a

Example

var btn1Clicks = flyd.stream();
button1Elm.addEventListener(btn1Clicks);
var btn2Clicks = flyd.stream();
button2Elm.addEventListener(btn2Clicks);
var allClicks = flyd.merge(btn1Clicks, btn2Clicks);

flyd.transduce(transducer, stream)

Creates a new stream resulting from applying transducer to stream.

Signature

Transducer -> Stream a -> Stream b

Example

var t = require('transducers.js');

var results = [];
var s1 = flyd.stream();
var tx = t.compose(
  t.map(function(x) { return x * 2; }),
  t.dedupe()
);
var s2 = flyd.transduce(tx, s1);
flyd.combine(function(s2) { results.push(s2()); }, [s2]);
s1(1)(1)(2)(3)(3)(3)(4);
results; // [2, 4, 6, 8]

###flyd.curryN(n, fn)

Returns fn curried to n. Use this function to curry functions exposed by modules for Flyd.

Signature

Integer -> (* -> a) -> (* -> a)

Example

function add(x, y) { return x + y; };
flyd.curryN(2, add);
var add

###stream()

Returns the last value of the stream.

Signature

a

Example

var names = flyd.stream('Turing');
names(); // 'Turing'

###stream(val)

Pushes a value down the stream.

Signature

a -> Stream a

Example

names('Bohr');
names(); // 'Bohr'

###stream.end

A stream that emits true when the stream ends. If true is pushed down the stream the parent stream ends.

###stream.map(f)

Returns a new stream identical to the original except every value will be passed through f.

Note: This function is included in order to support the fantasy land specification.

Signature

Called bound to Stream a: (a -> b) -> Stream b

Example

var numbers = flyd.stream(0);
var squaredNumbers = numbers.map(function(n) { return n*n; });

###stream1.ap(stream2)

stream1 must be a stream of functions.

Returns a new stream which is the result of applying the functions from stream1 to the values in stream2.

Note: This function is included in order to support the fantasy land specification.

Signature

Called bound to Stream (a -> b): a -> Stream b

Example

var add = flyd.curryN(2, function(x, y) { return x + y; });
var numbers1 = flyd.stream();
var numbers2 = flyd.stream();
var addToNumbers1 = flyd.map(add, numbers1);
var added = addToNumbers1.ap(numbers2);

###stream.of(value)

Returns a new stream with value as its initial value. It is identical to calling flyd.stream with one argument.

Signature

Called bound to Stream (a): b -> Stream b

Example

var n = flyd.stream(1);
var m = n.of(1);

Modules

If you're created a module for Flyd open an issue or send a pull request and it will be added to this list.

Misc

The name

The name Flyd was chosen since the author of Flyd is danish and Flyd is a danish word meaning float, afloat or flow. It is furthermore short and not too bad for searching.

For most native English speakers "flyd" is impossible to pronounce like a dane would do it. The "d" is soft like "th" in "the". The "y" is a vocal sound unknown to the English language. If you're curious Google Translates listening feature provides an accurate pronounciation..

Atomic updates

Consider the following example:

var a = flyd.stream(1);
var b = flyd.combine(function(a) { return a() * 2; }, [a]);
var c = flyd.combine(function(a) { return a() + 4; }, [a]);
var d = flyd.combine(function(b, c, self, ch) {
  result.push(b() + c());
}, [b, c]);

The dependency graph looks like this.

    a
  /   \
 b     c
  \   /
    d

Now, when a value flows down a, both b and c will change because they depend on a. If you merely consider streams as being event emitters you'd expect d to be updated twice. Because a triggers b triggers d after which a also triggers c which again triggers d.

But Flyd handles such cases optimally. Since only one value entered the system d will only be updated once with the changed values of b and c.

Flyd guarantees that when a single value enters the system every stream will only be updated once, along with their dependencies in their most recent state.

This avoids superfluous updates of your streams and intermediate states when several streams change at the same time.

Flyd implements atomic updates with a O(n) topological sort where n is number of streams that directly or indirectly depends on the updated stream.

Environment support

Flyd works in all ECMAScript 5 environments. It works in older environments with polyfills for Array.prototype.filter and Array.prototype.map.

Run tests, generate documentation

To run the test, clone this repository and:

npm install
npm test

The npm test command run three tests: a eslint js style checker test, the test of the core library and the test of the modules. If you want to run only the test of the library npm run test.

The API.md file is generated using npm run docs (it assumes it has documentation installed globally: npm i -g documentation)