How are co- and contra-variance used in designing business applications?

Question

I know about using co- and contravariance in the standard library (e.g. collections and trait Function) I wonder how co- and contravariance are used in design of "real world" business applications.

Answer 1

The classic example is functions, taking the Scala interface for a function with a single argument:

trait Function1[-T1, +R]

Which is contravariant (the -) for the argument, and covariant (the +) for the return type.

Why?

Imagine you have these classes:

class Timelord { ... }
class Doctor extends Timelord { ... }

class Enemy { ... }
class Dalek extends Enemy { ... }

If you have a method that takes, as a parameter, a Doctor => Enemy function; then it's okay to supply an instance of TimeLord => Enemy. It'll still accept instances of Doctor.

So TimeLord => Enemy is a subclass of Doctor => Enemy because TimeLord is a superclass of Doctor, it's contravariant in that parameter.

Likewise, a function returning a Dalek is valid when you need a function returning some Enemy, because a Dalek is-an Enemy

So Doctor => Dalek is a subclass of Doctor => Enemy because Dalek is a subclass of Enemy, it's covariant in that parameter.