Implement Powerful Generics

[joshradin]

Minimum Implementation

Generic objects can be defined using the for <T, ...> clause on any top level declaration. This creates a generic type declaration. Generic types aren't actually translated to code directly. For compound types, any time a initialization process is called for a generic type, an implementation is generated that metamorphosizes the original declaration for the new types. For functions, this occurs anytime the function is called.

An instance can be declared for a generic type with invariance, contravariance, and covariance using the follow syntax: Type<T> for invariance Type<? : T> for covariance Type<T : ?> for contravariance

Generic declarations can have bounds, like so: for<T : Object>. This is also the default bound.

Extended Implementation

Variable Length parameters

Enable multiple generic implementations per a single type based on number of arguments. For example for <T> class type and for <T, R> class type will be two separate implementations of the same type. In addition, generics can have variadic argument in the last position using the following form: for <T1, T2, ..., Tn, (HD:TL) (: bound)?>. The HD declaration is the first type parameter in the list, the TL part represents everything else. Both arguments can be empty. If the HD or TL is placed after a comma, but it is empty, the comma is ignored in terms of parameter count. The following is can example of a use case for this:

for <R, (HD:TL)> class fn {
    private fn<fn<R : ?, ? : TL>, HD> next;

    public R enact(HD arg1, args...) {
        fn<R : ?, ? : TL> func = next->enact(args1);
        return func(args);
    }
}

for <R> class fn {
    public R enact();
}

[joshradin]

Current implementation parses part of the Generic representation