Leaf Petersen

@munificent the type of the arguments are not taken into account when resolving the generic type parameters of the extension. This was a deliberate choice, based on design discussion at...

> However, when in strong mode, String is not assignable to T extends String, even though the reverse is true. This is correct. If `T` extends `String`, then `String` may...

One obvious way to do this is to reserve some seldom used (or even illegal) pattern for ddc use, and mangle any user defined identifiers that collide. So, e.g., we...

Fairly extensive initial draft here: https://dart-lang.github.io/dev_compiler/strong-dart.pdf

Is this still something we want to do, or should this test die?

Don't have strong opinions. No one outside of the library can directly call the constructor, but they could end up indirectly calling it (and so could end up stepping through...

This is addressed (albeit not including the new tear-off syntax). The static typing is addressed under the "Expression typing" section, and the semantic behavior is discussed under the "Generic instantiation...

We're still not quite on the same page. The term "second class" is a term of art for this style of polymorphism in some communities - maybe confusing, but it...

Aargh, markdown fail - if you leave off the back quotes around the generic instantiation syntax, then markdown drops the type arguments (so List <T> renders as List). Fixed inline,...

I don't object to this per se. I don't think it would be terrible for the platforms (which have to do resolution eventually anyway), though probably annoying to have argument...