tallstack

A small library which allows you to define tail recursive functions in JavaScript that will not cause stack overflow.

It is published for download on npm.

Contents

1. How do I use it?
2. How does it work?
3. What are the drawbacks?
4. License
5. Contributing

How do I use it?

The library only exports 3 functions, call, callWithContext, and recursive. Let's look at an example function which computes the N factorial:

const { recursive, call } = require('tallstack');
 
const multiply = (x, y) => (x * y);
 
const factorial = recursive((N) => {
    if (N === 0) return 1;
    return call(multiply, N, call(factorial, N - 1));
});

Each recursive function you want to define should be wrapped in the currying function, recursive, which returns a new function with the same call signature.

Each call to the recursive function within its body must be done by using call, supplying the function you wish to execute as the first argument and all the parameters for that function as subsequent arguments.

It is important to note that even if you are calling an auxiliary function which uses a recursive case as an argument, you still need to use call (e.g. multiply in the example above).

You can also use callWithContext to set the this argument in whichever function you are calling. Extending the N factorial example above:

const { recursive, call, callWithContext } = require('tallstack');
 
const object = { multiply: (x, y) => (x * y) };
function multiply(x, y) {
    return this.multiply(x, y);
}
 
const factorial = recursive((N) => {
    if (N === 0) return 1;
    return callWithContext(object, multiply, N, call(factorial, N - 1));
});

The function recurse also takes an optional 2nd parameter and binds that object to the recursive function's this object. Extending the N factorial example again:

const { recursive, call, callWithContext } = require('tallstack');
 
const obj = { multiply: (x, y) => (x * y) };
 
function factorialWithThis(N) {
    if (N === 0) return 1;
    return call(this.multiply, N, callWithContext(this, factorialWithThis, N - 1));
}
 
const factorial = recursive(factorialWithThis, obj);

How does it work?

The library works by implementing the callstack in JavaScript and then evaluating the recursion iteratively. The iterative evaluation guarantees that the stack is not bound by the size of the call stack in the interpreter. This allows you to evaluate recursions that are thousands or possibly millions of calls deep.

What are the drawbacks?

So far there are two big drawbacks on the library, and I am going to be actively looking for ways to improve the library (PRs are welcome, see Contributing).

1. You cannot set local variables with call:

For example you could not do this:

const factorial = recursive((N) => {
    if (N === 0) return 1;
    const M = call(factorial, N - 1);
    return M * N; // would return NaN
});

with the current implementation I do not see a way to do this. Though it is possible to use the evaluated result later as an argument to a first argument of call. For example:

const factorial = recursive((N) => {
    if (N === 0) return 1;
    return call((M) => (M * N), call(factorial, N - 1)); // this is ok!
});

usually the let keyword in functional languages is syntactic sugar for this type of evaluation strategy.

2. Performance:

So far this library does not evaluate functions as quickly as native JavaScript recursion. To some extent this is to be expected, due to the added complexity on top of normal function evaluation. Right now the performance seems to degrade as your recursion tree gets wider (i.e. each call has many recursive calls). I have done some work to speed it up but it is not quite as fast as it can get to native, but I am sure there is more that can be done and PRs are welcome.

License

This software is issued under the Apache 2.0 license. It is free for use as is without warranty for personal or recreational use. You can read more by looking at LICENSE.

Contributing

I every intention of keeping this software open source and free for use. If someone can use this code to make a different library that does what my library does better then that would be great. If people want to help make this library better, that's also great!

PRs are welcome but must be reviewed by me. Any changes to the code must pass existing tests and any additional functionality should also be tested as well. PRs with new test cases are also welcome.

For more information on community guidlines and expectations, see CONTRIBUTING.md.