super apply instead of call

[quangv]

Any syntax to get super to do an apply instead of a call with you pass in arguments?

class Child extends Parent
  constructor : ->
    super.apply ['one', 'two']
    # want -> Child.__super__.constructor.apply(this, ['one', 'two']);

repl link

Thanks

[michaelficarra]

super args...

[quangv]

you guys are quick, thank you :)

[michaelficarra]

FWIW, I really don't like the fact that super is an implicit call. It's inconsistent with all other calls in CoffeeScript. If you want to call it paren-free for its side effects, it should be do super. super should just be a special reference to ClassName.__super__.constructor.

[pier-oliviert]

+1 for super reference :)

[jashkenas]

I don't really like the semantics of "naked super" as well ... it is a bit magic -- but exposing the reference is far worse, I think. Within a function super should refer to the parent class's implementation of the current function -- not refer to the class object itself.

[michaelficarra]

That sounds fine to me.

[pier-oliviert]

I prefer an approach like objective-C [super method] which tends to be easier to read.

In coffeescript, super.method(args) sounds more right than super(args).

[jashkenas]

In coffeescript, super.method(args) sounds more right than super(args).

The reason why we don't want to do that is because it violates encapsulation. While it's legit to override a method and defer to the parent's implementation for part of the behavior, it's an anti-pattern to reach behind a different method's back to call a super implementation. You should be calling this.method(args) instead, if method is a different method.

... and just to be clear, I'm not sure that we should change this at this point -- it'd be massively backwards incompatible.

[pier-oliviert]

No worries, it's just a debate. not entitling to anything. And you are right about encapsulation.

[GeoffreyBooth]

Closing as no longer relevant in CS2.