Add rule no-void

[acarballo-cs]

Avoid useless void operator

[LinusU]

Sounds good to me, I don't really see when one would need it ��

[feross]

This sounds good to me. Only one repo was failing and it was an easy fix: feross/chrome-net@8cc2520

Let's release this in standard 14.

[feross]

LGTM

[mightyiam]

I was just recently contemplating whether there are useful usages of void. Do you feel like the ESLint rule doc lists all them? I never used it, even once. So I'd probably not miss it. Yet, I'm still wondering. How about non-leaky arrow functions?

[feross]

void is used so rarely and there are better alternatives, so it's probably easier to just ban it so most folks won't even have to learn what it is.

If I have to make an argument for void, I agree that non-leaky arrow functions is a reasonable one:

button.onclick = () => void doSomething()

However, you can do this with fewer characters like this:

button.onclick = () => { doSomething() }

And this avoids the need to understand a new concept. So it's clearer, IMO.

Of course, you can always add an eslint ignore comment if you really need to use void for some reason and nothing else works.

[mightyiam]

No argument here.

[Jessidhia]

I find void to be a more elegant solution to the "semicolon prefix" workaround to the ASI problem.

void (() => { /* some iife */ })()
// vs
;(() => { /* another iife */ })()

but this is kind of a style choice.

[nolanlawson]

I use void 0 as a shorthand for undefined, but this is probably confusing and something I should stop doing anyway. Especially given that undefined can't be redefined anymore. 👍

[feross]

Especially given that undefined can't be redefined anymore. 👍

Yep! So happy about this change!