Benchmark: ./node benchmark/events/eventtarget.js

Before my changes:

events/eventtarget.js listeners=1 n=1000000: 609,398.4958350995
events/eventtarget.js listeners=5 n=1000000: 612,860.7867785434
events/eventtarget.js listeners=10 n=1000000: 605,421.3805186535

After my changes

events/eventtarget.js listeners=1 n=1000000: 697,612.8094972048
events/eventtarget.js listeners=5 n=1000000: 678,713.3416166586
events/eventtarget.js listeners=10 n=1000000: 624,031.739092949

😅 I think I have to make some changes

The latest commit has some decent perf improvements:

events/eventtarget.js listeners=1 n=1000000: 471,435.62430082005
events/eventtarget.js listeners=5 n=1000000: 429,752.2592840155
events/eventtarget.js listeners=10 n=1000000: 489,863.1655030646

Based on @ronag comment I ditched the Listener class and the singular listener optimization and used an array of arrays (each inner array is a listener). I'm going to try replacing the inner array with an object next

I tried using object inside the array and while it makes the code clearer it is about the same speed as the original code.

I think the current improvement is going to be the best improvement for now

@Ethan-Arrowood The numbers printed in the benchmark logs are in operations / second, so when you compare with a previous version, higher is better.

higher is better.

😂 oh whoops! Okay hmm so the array version is not better, and my first attempt was only marginally better 🤔

I'm going to leave this open so others can comment, but I do believe focussing on spec compliance is better right now. Will open a new issue and begin work on that shortly.

To be honest, I really wouldn't expect the array version to be faster than the linked list here, and the linked list is already optimized for the single listener case. There are definitely performance improvements that can be made but I don't think this is it.

Yeah I agree! Do you think I should close this PR? I already have another branch focussing on adding the WPT stuff