|
|
|---|
|
Type safety |
A linear type system is a particular form of type system used in a programming language. Linear type systems allow references but not aliases. To enforce this, a reference goes out of scope after appearing on the right-hand side of an assignment, thus ensuring that only one reference to any object exists at once. Note that passing a reference as an argument to a function is a form of assignment, as the function parameter will be assigned the value inside the function, and therefore such use of a reference also causes it to go out of scope. Linear typing is related to uniqueness typing but is generally more restrictive.
Affine types are a weaker version of linear types; an affine resource can only be used once, while a linear one must be used once.
A linear type system is similar to C++'s auto_ptr class, which behaves like a pointer but is invalidated by being set to null after use in an assignment. However, the linearity constraint can be checked at compile time, whereas auto_ptr can only raise exceptions at run-time if it is misused.
This example uses C++-like notation, but shows a language with a linear type system (which therefore is not C++).
Dog* d = new Dog( name="Fido" ); // creates a reference to a new object
Dog* p = d; // uses d to create a reference denoted by p to the object referred to by d
// this forces d out of scope, thus d cannot be legally used after the assignment to p
print p->getName(); // output "Fido"
print d->getName(); // COMPILE-TIME ERROR: d was forced out of scope
// by its use above and is therefore
// no longer a valid variable
The following programming languages support linear (or affine) types: