(Corrected the link to the paper from Luca Cardelli. Added a followup essay from William Cook.)
|(3 intermediate revisions by one user not shown)|
Revision as of 17:14, 29 April 2013
A value is polymorphic if there is more than one type it can have. Polymorphism is widespread in Haskell and is a key feature of its type system.
1 Parametric polymorphism
Parametric polymorphism refers to when the type of a value contains one or more (unconstrained) type variables, so that the value may adopt any type that results from substituting those variables with concrete types.
In Haskell, this means any type in which a type variable, denoted by a name in a type beginning with a lowercase letter, appears without constraints (i.e. does not appear to the left of a =>). In Java and some similar languages, generics (roughly speaking) fill this role.For example, the function
Since a parametrically polymorphic value does not "know" anything about the unconstrained type variables, it must behave the same regardless of its type. This is a somewhat limiting but extremely useful property known as parametricity.
2 Ad-hoc polymorphism
Ad-hoc polymorphism refers to when a value is able to adopt any one of several types because it, or a value it uses, has been given a separate definition for each of those types. For example, the + operator essentially does something entirely different when applied to floating-point values as compared to when applied to integers – in Python it can even be applied to strings as well. Most languages support at least some ad-hoc polymorphism, but in languages like C it is restricted to only built-in functions and types. Other languages like C++ allow programmers to provide their own overloading, supplying multiple definitions of a single function, to be disambiguated by the types of the arguments. In Haskell, this is achieved via the system of type classes and class instances.
Despite the similarity of the name, Haskell's type classes are quite different from the classes of most object-oriented languages. They have more in common with interfaces, in that they specify a series of methods or values by their type signature, to be implemented by an instance declaration.So, for example, if my type can be compared for equality (most types can, but some, particularly function types, cannot) then I can give an instance declaration of the
Unlike the overloading in some languages, overloading in Haskell is not limited to functions – minBound is an example of an overloaded value, so that when used as a Char it will have value '\NUL' while as an Int it might be -2147483648.
Haskell even allows class instances to be defined for types which are themselves polymorphic (either ad-hoc or parametrically). So for example, an instance can be defined of Eq that says "if a has an equality operation, then [a] has one". Then, of course, a will automatically also have an instance, and so complex compound types can have instances built for them out of the instances of their components.You can recognise the presence of ad-hoc polymorphism by looking for constrained type variables: that is, variables that appear to the left of
3 Other kinds of polymorphism
There are some kinds of polymorphism that Haskell doesn't support, or at least not natively, e.g. inclusion polymorphism and subtyping, common in OO languages, where values of one type can act as values of another type.
4 Further reading
- On Understanding Types, Data Abstraction, and Polymorphism (1985), by Luca Cardelli, Peter Wegner in ACM Computing Surveys.
- On Understanding Data Abstraction, Revisited (2009), by William R. Cook in OOPSLA 2009.
- Type polymorphism at Wikipedia