Gavin King
Gavin King
> I suspect the bytecode generated for Ceylon clients would indicate the narrowed return type Well, we could make absolutely sure this never happens by making the restriction I proposed...
> There's a whole lot of nuance with that argument I'd rather not get into. Like I said at the start, we should all just stop being so argumentative and...
The principal goal is that when I'm implementing an interface or abstract class, I typically have to write a whole bunch of methods that fill in what are essentially internal...
> I don't believe that the rather large proposed patch accomplishes that I don't understand.
Ahyes, thanks, I had not tested it with default interface methods.
Since there is no consensus on this yet, and because there are still some bugs related to interface members, not to mention missing tests and spec, this is not for...
I don't really see any good reason why not. The only concern is it might lead people to expect that you can use `assert` in other sorts of expressions. But...
@tombentley let me know when you're ready to start working on this one.
Related: #3483
> Specifically, [just this part](https://gist.github.com/lucaswerkmeister/b0c4a7369d87bd0a16034ea0f1ae9796) of the `Ceylon.g` diff is enough to produce the problem. Weird. That bit of the grammar doesn't deal with type expressions *at all*.