Gilad: Me too, it's like a get out of jail free card.
Kasper: That means the type behaves like dynamic in all contexts? Even for is checks and checked mode checks?
Gilad: If you assign something to it, it goes through normal rules. But if you assign it to something, works like dynamic. It could go both ways.
Lars: The real solution would be to do it dynamically. Like you have noSuchMethod, you could have isSubtype on ClassMirror that returns a boolean. You could implement that to do what you want.
Gilad: That's complicated to implement.
Lars: Let's talk about the first solution with Ivan and Kasper.
Kasper: That means it will be a subtype of all types. If a class claims it implements dynamic and List, should you get warnings if you implement list methods incorrectly? Are we going to say it's illegal to implement list too?
Gilad: It's legal, but stupid. It won't change anything at all.
Lars: My perspective is, given that this is a corner case, I'm fine with it not having any effect if you also say you implement List.
Gilad: We could say implementing dynamic just affects subtyping, but you still get the same warnings internally in class.
Lars: I'm fine with that too.
Gilad: OK, i'll send around a proposal.
optional arguments and forwarding
Lars: No one is screaming for [the forwarding operator], so we'll just ignore it.
Gilad: Do we use "?" on its own?
Lars: It's being used a lot, so we keep it. As long as you aren't forwarding it's fine. Maybe we need to look at the reflective interface for calling methods. If we implement this proposed solution, it probably won't be used other than in a few places.
Kasper: Where are the uses of "?"? In the DOM mostly?
Lars: There's a lot in the DOM but in other places to.
Kasper: It's not a question of forwarding, but of being forwarded to.
Lars: I'm fine either way. I find it harder to read using null. It's hard to remove things because people are already screaming about breaking lib changes.
Lars: Gilad, are you up on what's going on there?
Gilad: I've seen the patches, but it's not in spec. I haven't had time to talk to Peter yet, but I'm tracking it.
Kasper: We haven't figured out if we want same solution on VM.
Gilad: I assumed that. What does that mean?
Lars: As a minimum, we need the same code to run there.
Kasper: Yes, load() should do something minimal.
Gilad: It would be nice if it actually could defer it.
Lars: I'm fine with the VM preloading it.
Gilad: In that case, it doesn't belong in spec because it has no particular behavior. It's a dart2js deployment detail.
Lars: Yes, easiest solution is to keep out of spec.
As always, view the changelog for the full list of changes, and to get started with the Editor, see our tutorial.