Type polymorphism
May 27, 2020
Polymorphism can be defined as a way of representing multiple different types in a single abstraction.
Structural vs nominal type systems
How the type system decides if the types are equivalent (1) or have a subtype-relationship (2)?
- Structural typing
Answer: Structure of types. Properties of types must be identical in order for the types to be identical.
Examples: OCaml, Typescript, Haskell.
Common for dynamically-typed languages.
- Nominal typing
Answer: Explicit declarations. Type declarations have the same name.
Examples: C#, Java, Rust, Swift, F#.
Nominal is a subset of structural.
See link
Types of polymorphism
- Subtyping
Restrict the range of types that can be used in a given abstraction (function or type) through relation between types. Common for OOP languages through interface or subclassing. Requires “substitute” relationship, subtypes/supertype.
Subtypes can be used in places of supertypes.
Examples: C#, Java, F# (wtf)
No support in OCaml due to the module systems and type signatures for modules.
- Parametric polymorphism (generics)
Handle values generically without depending on their type.
Generics in Java or C# and templates in C++.
- Row polymorphism
OCaml objects - polymorphic records.
See this SO answer.
- Ad-hoc polymorphism
No relationship between the types is necessary (as opposed to subtyping). There concrete implementations for the different types.
Examples: Haskell operator overloading and type classes.