don't need to count realised parameters

[michaelficarra]

This is simpler and doesn't introduce new phrasing. I've also changed the semantics when the argument is missing from throwing to no-op.

[ljharb]

It was intentional to throw an error when the setter gets zero arguments; that shouldn't be valid for any setter. I can certainly use the "is present" verbiage tho.

[michaelficarra]

There are 3 existing built-in setters in the language and none of them throw when the parameter is not present.

1. https://tc39.es/ecma262/#sec-object.prototype.valueof
2. https://tc39.es/ecma262/#sec-iterator.prototype-%symbol.tostringtag%
3. https://tc39.es/ecma262/#sec-iterator.prototype.constructor

[ljharb]

Sure - that doesn't mean that's a good precedent, it just means there's precedent to be sloppy. There's precedent to coerce, too, but we chose to change that because it was a bad decision - similarly, having a function that should only ever be called with precisely 1 argument noop when called with zero seems like an oversight to me.

(it's also notable that one of those is legacy, and the other two are web compatibility hacks - so probably not great examples to look at for precedent anyways)

[ljharb]

Closing, per today's plenary and 377c7e7