Typedef and Interface

[filt3rek]

Hej,

I don't know if it's intentional but I just realized that Haxe doesn't complain here : https://try.haxe.org/#729BD642

typedef TFoo = {
	a:Int,
	b:Int
}

interface MyInterface {
	public function get():TFoo;
}

class MyClass implements MyInterface {
	public function new() {}

	public function get()	/*: TFoo*/ {  // <-------------------- 
		return {
			a: 0,
			b: 0,
			c: 0
		}
	}
}

class Test {
	static function main() {
		var cl = new MyClass();
		cl.get();
	}
}

If the return type is uncommented it does complain as it should, but without, it should also complain at least that the return type isn't the same as in the interface right ?

[back2dos]

The error you're getting is not related to an interface. This fails too:

typedef TFoo = {
	a:Int,
	b:Int
}

class MyClass {
	public function new() {}

	public function get():TFoo {
		return { a: 0, b: 0, c: 0 }
	}
}

The reason is that if you specify the return type, the compiler uses it for top down inference and you have the same type error as ({ a: 0, b: 0, c: 0 } : TFoo). There are multiple discussions about this, most notably this one: #3192

[filt3rek]

Hej Juraj,

Thx for your answer.
In fact I don't get any error and that why it's confusing me. I would love Haxe to throw me the errror saying that there is an extra field "c" but it doesn't.
And maybe it's not related to an interface but especially in the case of interface, where I'm wainting all to be strict (an interface is here for that I think), I would Haxe throw me an error, but it doesn't.
That's why I permit to write this issue.

[filt3rek]

But there is maybe something with variance because if for example the function doesn't return "a", compiler does throw an error as expected. So feel free to close this, I just wanted to point that

[Simn]

I suppose the question here is if the return type should be typed against the type from the interface, which on first glance sounds like a good idea.