DEV Community

Cover image for Haskell Quicksort in JavaScript

Haskell Quicksort in JavaScript

Caleb Weeks on October 20, 2021

Haskell has a particularly elegant implementation of the quicksort algorithm: qs :: (Ord a) => [a] -> [a] qs [] = [] qs (x:xs) = let sma...
Collapse
 
qm3ster profile image
Mihail Malo

JS and current devices are slow enough.
Even methods like Array.filter should be used sparingly, since they eagerly allocate a new array on every single call.
Unfortunately, neither are iterators and generators optimized enough (yet?) in JS runtimes.

The Haskell definition is indeed elegant, but it just doesn't scale to JS, at least for the time being.

Collapse
 
sethcalebweeks profile image
Caleb Weeks

I agree with your sentiment, and would not recommend actually using this code in production. It definitely won't be quick! JavaScript performance is less of a concern as it used to be. As for Array.filter, the tradeoff has to be made between saving machine performance (both speed and space), and human performance (clarity and maintainability).

Collapse
 
qm3ster profile image
Mihail Malo

I appreciate your readability concerns, but that article looks rather insane to me.
He starts by benchmarking absolutely bare, asm.js level code:

const ITERATIONS = 1000000000;

function main() {
    let myNum = 0;
    for (let i = 0; i <= ITERATIONS; i++) {
        myNum *= i;
        myNum++;
    }
}
Enter fullscreen mode Exit fullscreen mode

Actually, astonished that it didn't optimize to O(1)/constant time/noop in both languages, since the output is unobservable. Was the C compiled without optimization? 🤔

He then proceeds to rewrite

let smallest = Infinity;
const numbersLength = numbers.length; // avoid property look up in loop
for (let i = 0; i < numbersLength; i++) {
    const number = numbers[i]; // avoid repeated index lookups
    if (number > 0 && number < smallest)  smallest = number;
}
Enter fullscreen mode Exit fullscreen mode

again, asm.js spartanity code, but both of the things mentioned in the comments actually optimize well in JIT
into

numbers
  .filter(x => x > 0) // remove negative values
  .sort((a, b) => a-b)[0];
Enter fullscreen mode Exit fullscreen mode

instead of

numbers
  .filter(x => x > 0)
  .reduce((a, x) => x < a ? x : a)
Enter fullscreen mode Exit fullscreen mode

which in the future can be written as follows to avoid the intermediate array:

import {filter, reduce} from 'std:Iterable';
numbers
  |> filter(x => x > 0, %)
  |> reduce((a, x) => x < a ? x : a, %)
Enter fullscreen mode Exit fullscreen mode

introducing a whole ass mutable sorting step instead of a min reduce, fundamentally transforming the algorithm from O(n) to O(n log n), on TOP of the temporary memory allocation and slower constant factor.

Up next, he has delete user.password vs user.password = undefined.
Oh, it takes 1 billion iterations to show a difference between the two? Try benchmarking the whole application that contains this line. Look at memory use and battery power consumption.
How about the fact that the deletion affects the speed of all of your code that handles user-type objects by turning them from template objects into dynamic dictionaries or at best duplicating the monomorphised jitted code in memory, meaning less of it fits into the CPU cache?

Thread Thread
 
sethcalebweeks profile image
Caleb Weeks

Would love to see a better real-world benchmark! If you can find codebase where overuse of Array.filter has resulted in an application that is unusable, I would love to see it!

Thread Thread
 
qm3ster profile image
Mihail Malo

And last but not least, see how competitive WASM is with JS, even when supposedly having to run through a JS glue layer, in JS's historical home of the browser.
And this is not about fancy ergonomic end-user-written business logic, this is about web frameworks, worked on for years to improve performance at any cost. You know they aren't using delete or forEach.
And yet just look at those memory allocation numbers.

How fast JS runtimes are, given the language specification, is a huge achievement and nothing short of a miracle. But when you regard JS as the delivery target, because that's what runs in all browsers, I don't think it's ever right to completely forget about how to write to make the best use of those efforts.

Thread Thread
 
qm3ster profile image
Mihail Malo

I'd love to see that too. I'd tongue-in-cheek say facebook.com, but that's rendered unusable (performance-wise) by quite a bit more than Array methods :)
(They are used there, and not transpiled, though, which I found rather surprising.)

Thread Thread
 
sethcalebweeks profile image
Caleb Weeks

Haha, truthfully, I've been thinking the same thing about Facebook. There are times I can barely get it to fully render. I think their problem (and the problem with React in general) is that literally everything is replicated in the virtual DOM. I would love to see a JavaScript compiler to WebAssembly that turns things like Array.filter into faster solutions. Paired with a UI library that does virtual DOM in web workers, it could give us the best of both worlds (declarative code that compiles to optimized low level constructs).

Thread Thread
 
qm3ster profile image
Mihail Malo

I'm just assmad that Array.filter et al got added to the language specification, with all its awkwardness like creating an array, and passing so so many arguments to the callable passed. In addition to thwarting attempts at .reduce(Math.max) and .forEach(console.log), it causes arity mismatch which once again causes a miniscule deoptimization. Because, honestly, who writes .reduce((a, x, _, _) =>? Not many :v

Virtual DOM seems to be fundamentally too expensive for the performance people want, so a different model like reactive signals that bypasses it entirely is needed. I shitposted about it here yesterday.

Writing the business logic (including bringing in performant libraries that you use in the backend or native application) in the same language as the UI helpers and compiling it to WASM makes sense to me. The speed of WASM DOM modification is sure to increase somewhat in the future, but currently the critical performance downside for me is first-time startup performance. Sure, caching compiled modules is fast and efficient, but is shipping a raw HTML loading/landing/login page while the WASM downloads and compiles really enough?

Perhaps the situation will improve as people make sense of dynamically linking multiple pieces of WASM together, with the granularity and enthusiasm they show for JS bundle code splitting?

Collapse
 
darrylnoakes profile image
Darryl Noakes • Edited

Just a question, I may be missing something.

You said

Then we use the ternary operator to return an empty array if the head is undefined.

Yet your code is:

x !== undefined 
  ? [] 
  : [
    ...
  ]
Enter fullscreen mode Exit fullscreen mode

This will return an empty array when x is not undefined, and do more sorting if it is undefined.

Indeed, I ran the following code, and it logged [].

const qs = ([x, ...xs]) => x !== undefined 
  ? [] 
  : [
    ...qs(xs.filter(s => s <= x)),
    x,
    ...qs(xs.filter(b => b > x))
  ]

let arr = [5, 3, 1, 8, 6, 0, 2, 4, 7, 9];

console.log(qs(arr));
Enter fullscreen mode Exit fullscreen mode

To fix, simply change the !== undefined to === undefined.
You could also swap the clauses in the ternary expression instead.

I tested this and it works.
That will filter out any undefineds, or return an empty array if the first element is undefined.
null gets sorted to the the beginning of the array.

Collapse
 
darrylnoakes profile image
Darryl Noakes • Edited

Update:

I played around with it and came up with this:

let qs = (xs) => (
  xs.length > 0
    ? xs[0] !== undefined
      ? [
          ...qs(xs.slice(1).filter(s => s <= xs[0])),
          xs[0],
          ...qs(xs.slice(1).filter(b => b > xs[0]))
        ]
      : qs(xs.slice(1))
    : []
)
Enter fullscreen mode Exit fullscreen mode

This allows for having undefined at the beginning of the array. Not as elegant, but required due to x being undefined if an empty array is passed in (which is when the quicksort recursion ends and starts "undoing").

I did a few runs in JSBen.ch using my Pi 4 4GB.

  • native: (23981) 100%
  • orig: (15948) 66.5%
  • this: (14596) 60.86%

JSBen.ch test.

JSBench.me was less favorable.
I get this:

  • orig ~6000 - ~9500 ops/s; ~60% - ~85% slower
  • this: ~3000 - ~5000 ops/s; ~75% - ~95% slower
  • native: ~24000 ops/s

JSBench.me test

Note: see updated benchmark code in my later comments.

Collapse
 
sebastianfrey profile image
Sebastian Frey

Or use nested ternary operators:

const qs = ([x, ...xs]) => x === undefined 
  ? xs.length === 0
  ? []
  : qs(xs)
  : [
    ...qs(xs.filter(s => s <= x)),
    x,
    ...qs(xs.filter(b => b > x))
  ]

let arr = [5, 3, 1, 8, 6, 0, 2, 4, 7, 9];

console.log(qs(arr));
Enter fullscreen mode Exit fullscreen mode
Thread Thread
 
darrylnoakes profile image
Darryl Noakes

I tried to figure out something like that, but got stuck.
Much better, thanks!

(Now to rerun the benchmarks...)

Thread Thread
 
darrylnoakes profile image
Darryl Noakes • Edited

New benchmark results:

Run 1:

  • native (23740) 100%
  • orig quicksort (15751) 66.35%
  • undefined-safe quicksort (15673) 66.02%

Run 2:

  • native (24123) 100%
  • orig quicksort (15903) 65.92%
  • undefined-safe quicksort (15776) 65.4%

New JSBen.ch test

P.S. I dunno why I have got so stuck into this benchmarking :).

Collapse
 
sethcalebweeks profile image
Caleb Weeks

Cool! Thanks for adding those benchmarks! I wonder how it compares to this version:

const qs = (arr) => [...arr].sort((a, b) => a - b);
Enter fullscreen mode Exit fullscreen mode
Thread Thread
 
darrylnoakes profile image
Darryl Noakes • Edited

I did it with just .sort() (labeled as native), but I'm not sure how that compares.
It does affect the sort order though: plain .sort() orders it [numbers, nulls, undefineds], while this orders it like the other implementations, with undefineds being put at the end instead of being filtered, i.e. [nulls, numbers, undefineds].

Feel free to write some more benchmarks yourself, if you want :).

I did those on my Pi 4 4GB, which is slowish.
It gets 24k ops/s for native sort while my phone (Galaxy A51 4GB) gets 30k ops/s.

Thread Thread
 
darrylnoakes profile image
Darryl Noakes

Update
I remembered something "quirky" about JavaScript's default sorting algorithm: it coerces values to strings and sorts them lexicographically.
The reason the plain sort worked was because all then numbers had the same number of digits. If they had varying numbers, it would have sorted them like this: [2, 20, 3, 30].

I have made updated JSBen.ch benchmarks using arr.sort((a, b) => a - b) (I can't manage to login to JSBench.me):

Collapse
 
sethcalebweeks profile image
Caleb Weeks

Thanks for the catch! I'll update!

Collapse
 
amn3s1a2018 profile image
Amn3s1a2018

It's a cool refactor example, and its an example for a common error ...
The ternary operator nice replacement for the overload, but the correct way would

surprise, surprise
x !== undefined ? etc.

'cause this way qs filters out all falshy element from the array and drops there tail too.

Collapse
 
sethcalebweeks profile image
Caleb Weeks

Good catch! Without explicitly comparing against undefined, my version would fail for arrays containing zeros. Since this is a sorting algorithm, I'll assume that an array of numbers is being passed in. I'll update to correct it!

Collapse
 
amn3s1a2018 profile image
Amn3s1a2018 • Edited

Sorry, i just replaced your original bug with a bigger one. And maybe i didn't emphasized enough, that those zeros are not just the zeros. Boolean false, empty string and even the document.all (in browser it's an array or object with type of "undefined" – what?) is falsy...

Thread Thread
 
sethcalebweeks profile image
Caleb Weeks

Yeah, that's true. But I guess what should you expect if you try to put a boolean, empty string, or an object in a number sorting algorithm... There is a reason why Array.sort coerces all items in the array to strings.

Collapse
 
tracygjg profile image
Tracy Gilmore • Edited

Beautiful. For all its faults, JavaScript does offer a level of versatility uncommon to most languages. That means we can always learn and apply effective and efficient techniques found in other OO or/and FP languages.

Collapse
 
ashoutinthevoid profile image
Full Name

Cool stuff. Maybe not the most practical, but certainly interesting.

I learned more about Haskell syntax (as, alas, I still haven't 'learned me a Haskell for great good') than anything else, but that's never a bad thing.

Collapse
 
sethcalebweeks profile image
Caleb Weeks

Glad you found it interesting! To be honest, this post is mostly about explaining Haskell in JavaScript terms, and isn't meant to be all that practical.

Collapse
 
akashkava profile image
Akash Kava

Quick sort done inefficiently = Bubble Sort