In computer science, a dynamic programming language is a class of high-level programming languages, which at runtime execute many common programming behaviours that static programming languages perform during compilation. These behaviors could include an extension of the program, by adding new code, by extending objects and definitions, or by modifying the type system. Although similar behaviors can be emulated in nearly any language, with varying degrees of difficulty, complexity and performance costs, dynamic languages provide direct tools to make use of them. Many of these features were first implemented as native features in the Lisp programming language.
Most dynamic languages are also dynamically typed, but not all are. Dynamic languages are frequently (but not always) referred to as scripting languages, although that term in its narrowest sense refers to languages specific to a given run-time environment.
Some dynamic languages offer an eval function. This function takes a string or abstract syntax tree containing code in the language and executes it. If this code stands for an expression, the resulting value is returned. Erik Meijer and Peter Drayton distinguish the runtime code generation offered by eval from the dynamic loading offered by shared libraries, and warn that in many cases eval is used merely to implement higher-order functions (by passing functions as strings) or deserialization.
A type or object system can typically be modified during runtime in a dynamic language. This can mean generating new objects from a runtime definition or based on mixins of existing types or objects. This can also refer to changing the inheritance or type tree, and thus altering the way that existing types behave (especially with respect to the invocation of methods).
As a lot of dynamic languages come with a dynamic type system, runtime inference of types based on values for internal interpretation marks a common task. As value types may change throughout interpretation, it is regularly used upon performing atomic operations.
Static programming languages (possibly indirectly) require developers to define the size of utilized memory before compilation (unless working around with pointer logic). Consistent with object runtime alteration, dynamic languages implicitly need to (re-)allocate memory based on program individual operations.
Reflection is common in many dynamic languages, and typically involves analysis of the types and metadata of generic or polymorphic data. It can, however, also include full evaluation and modification of a program's code as data, such as the features that Lisp provides in analyzing S-expressions.
A limited number of dynamic programming languages provide features which combine code introspection (the ability to examine classes, functions, and keywords to know what they are, what they do and what they know) and eval in a feature called macros. Most programmers today who are aware of the term macro have encountered them in C or C++, where they are a static feature which is built in a small subset of the language, and are capable only of string substitutions on the text of the program. In dynamic languages, however, they provide access to the inner workings of the compiler, and full access to the interpreter, virtual machine, or runtime, allowing the definition of language-like constructs which can optimize code or modify the syntax or grammar of the language.
The example shows how a function can be modified at runtime from computed source code
; the source code is stored as data in a variable CL-USER > (defparameter *best-guess-formula* '(lambda (x) (* x x 2.5))) *BEST-GUESS-FORMULA* ; a function is created from the code and compiled at runtime, the function is available under the name best-guess CL-USER > (compile 'best-guess *best-guess-formula*) #<Function 15 40600152F4> ; the function can be called CL-USER > (best-guess 10.3) 265.225 ; the source code might be improved at runtime CL-USER > (setf *best-guess-formula* `(lambda (x) ,(list 'sqrt (third *best-guess-formula*)))) (LAMBDA (X) (SQRT (* X X 2.5))) ; a new version of the function is being compiled CL-USER > (compile 'best-guess *best-guess-formula*) #<Function 16 406000085C> ; the next call will call the new function, a feature of late binding CL-USER > (best-guess 10.3) 16.28573
This example shows how an existing instance can be changed to include a new slot when its class changes and that an existing method can be replaced with a new version.
; a person class. The person has a name. CL-USER > (defclass person () ((name :initarg :name))) #<STANDARD-CLASS PERSON 4020081FB3> ; a custom printing method for the objects of class person CL-USER > (defmethod print-object ((p person) stream) (print-unreadable-object (p stream :type t) (format stream "~a" (slot-value p 'name)))) #<STANDARD-METHOD PRINT-OBJECT NIL (PERSON T) 4020066E5B> ; one example person instance CL-USER > (setf *person-1* (make-instance 'person :name "Eva Luator")) #<PERSON Eva Luator> ; the class person gets a second slot. It then has the slots name and age. CL-USER > (defclass person () ((name :initarg :name) (age :initarg :age :initform :unknown))) #<STANDARD-CLASS PERSON 4220333E23> ; updating the method to print the object CL-USER > (defmethod print-object ((p person) stream) (print-unreadable-object (p stream :type t) (format stream "~a age: ~" (slot-value p 'name) (slot-value p 'age)))) #<STANDARD-METHOD PRINT-OBJECT NIL (PERSON T) 402022ADE3> ; the existing object has now changed, it has an additional slot and a new print method CL-USER > *person-1* #<PERSON Eva Luator age: UNKNOWN> ; we can set the new age slot of instance CL-USER > (setf (slot-value *person-1* 'age) 25) 25 ; the object has been updated CL-USER > *person-1* #<PERSON Eva Luator age: 25>
In the next example, the class person gets a new superclass. The print method gets redefined such that it assembles several methods into the effective method. The effective method gets assembled based on the class of the argument and the at runtime available and applicable methods.
; the class person CL-USER > (defclass person () ((name :initarg :name))) #<STANDARD-CLASS PERSON 4220333E23> ; a person just prints its name CL-USER > (defmethod print-object ((p person) stream) (print-unreadable-object (p stream :type t) (format stream "~a" (slot-value p 'name)))) #<STANDARD-METHOD PRINT-OBJECT NIL (PERSON T) 40200605AB> ; a person instance CL-USER > (defparameter *person-1* (make-instance 'person :name "Eva Luator")) *PERSON-1* ; displaying a person instance CL-USER > *person-1* #<PERSON Eva Luator> ; now redefining the print method to be extensible ; the around method creates the context for the print method and it calls the next method CL-USER > (defmethod print-object :around ((p person) stream) (print-unreadable-object (p stream :type t) (call-next-method))) #<STANDARD-METHOD PRINT-OBJECT (:AROUND) (PERSON T) 4020263743> ; the primary method prints the name CL-USER > (defmethod print-object ((p person) stream) (format stream "~a" (slot-value p 'name))) #<STANDARD-METHOD PRINT-OBJECT NIL (PERSON T) 40202646BB> ; a new class id-mixin provides an id CL-USER > (defclass id-mixin () ((id :initarg :id))) #<STANDARD-CLASS ID-MIXIN 422034A7AB> ; the print method just prints the value of the id slot CL-USER > (defmethod print-object :after ((object id-mixin) stream) (format stream " ID: ~a" (slot-value object 'id))) #<STANDARD-METHOD PRINT-OBJECT (:AFTER) (ID-MIXIN T) 4020278E33> ; now we redefine the class person to include the mixin id-mixin CL-USER 241 > (defclass person (id-mixin) ((name :initarg :name))) #<STANDARD-CLASS PERSON 4220333E23> ; the existing instance *person-1* now has a new slot and we set it to 42 CL-USER 242 > (setf (slot-value *person-1* 'id) 42) 42 ; displaying the object again. The print-object function now has an effective method, which calls three methods: an around method, the primary method and the after method. CL-USER 243 > *person-1* #<PERSON Eva Luator ID: 42>
(Many use the term "scripting languages".)