Rule suggestion: static-this

[JoshuaKGoldberg]

Rule Suggestion

Is your rule for a general problem or is it specific to your development style?

General problem

What does your suggested rule do?

Ban the use of this in static methods.

Static this usage can be confusing for newcomers.
It can also become imprecise when used with extended classes when a static this of a parent class no longer specifically refers to the parent class.

List several examples where your rule could be used

Instead of:

class Foo {
    private static value = true;
    static getValue() {
        return this.value;
    }
}

Prefer:

class Foo {
    private static value = true;
    static getValue() {
        return Foo.value;
    }
}

See https://github.com/ajafff/tslint-consistent-codestyle/blob/master/docs/no-static-this.md

Additional context

Spinning off of #2040: moving some rules of tslint-consistent-codestyle to tslint core.
Note that as ajafff mentioned, copied code without substantial changes needs to retain the original license.

[mbelsky]

Hey,
I'm working on this rule's implementation. Could you please assign this issue to me?
Also if the 'static-this' name isn't final please let me know what do you think about name it 'no-static-this'?

[JoshuaKGoldberg]

@mbelsky there's no way to assign someone who's not in the Palantir org unfortunately. Posting here should be enough indicator to ask others not to tackle it soon, hopefully!

Re no-static-this vs static-this, is there a reason you'd prefer the name change? Prefixing rule names with no- has a couple downsides:

• It makes them less discoverable (harder to scan through) in lists of rules
• It discourages adding options for the opposite style - in this case, if folks would prefer the this for brevity

[onehorsetown]

This improperly flags this in an anonymous class returned by a static function, something that is typically used to construct higher-order components:

interface Bar {}

class Foo {
  public static higherOrderComponent(): any {
    return class implements Bar {
      public constructor() {
        console.log(this);   // "this" is valid here but tslint is complaining
      }
    };
  }
}

[mbelsky]

Hey,
Thanks for update. I'm going to bed now, so I'll try to fix that tomorrow if no one take it.

[mbelsky]

@JoshuaKGoldberg could you please create a new issue for this false positive case and assign to me?

[onehorsetown]

I opened a new issue: #4579

I also noticed above that Joshua indicated that he could not assign something to you:

#4428 (comment)

[JoshuaKGoldberg]

Oh yes, this issue was resolved by the original PR, and #4579 tracks the followup false positive case. Thanks folks!