Kotlin Download PDF
Table of contents

Kotlin language specification

Version 1.9-rfc+0.1

Marat Akhin

Mikhail Belyaev

Scopes and identifiers

Kotlin program is logically divided into scopes.

A scope is a syntactically-delimited region of code which constitutes a context in which entities and their names can be introduced. Scopes can be nested, with entities introduced in outer scopes possibly available in the inner scopes if linked.

The top level of a Kotlin file is also a scope, containing all the scopes within a file.

All the scopes are divided into two kinds: declaration scopes and statement scopes. These two kinds of scopes differ in how the identifiers may refer to the entities defined in the scopes.

Declaration scopes include:

Statement scopes include:

All declarations in a particular scope introduce new bindings of identifiers in this scope to their respective entities in the program. These entities may be types or values, where values refer to objects, functions or properties (which may also be delegated). Top-level scopes additionally allow to introduce new bindings using import directives from other top-level scopes.

In most situations, it is not allowed to bind several values to the same identifier in the same scope, but it is allowed to bind a value to an identifier already available in the scope through linked scopes or imports.

An exception to this rule are function declarations, which are matched by signatures and allow defining several functions with the same name in the same scope. When calling functions, a process called overload resolution allows for differentiating between such functions. Overload resolution also applies to properties if they are used as functions through invoke-convention, but it does not allow defining several properties with the same name and with the same receivers in the same scope.

Note: platforms may introduce additional restrictions on which identifiers may be declared together in the same or linked scopes.

The main difference between declaration scopes and statement scopes is that names in the statement scope are bound in the order of appearance. It is not allowed to access a value through an identifier in code which (syntactically) precedes the binding itself. On the contrary, in declaration scopes it is fully allowed, although initialization cycles may occur leading to unspecified behaviour.

Note: Kotlin compiler may attempt to detect and report such initialization cycles as compile-time warnings or errors.

It also means that statement scopes nested inside declaration scopes may access values declared afterwards in parent declaration scopes, but any values declared inside a statement scope can be accessed only after their declaration point.


Linked scopes

Scopes A and B in a Kotlin program may be downwards-linked (A ~> B), meaning identifiers from A can be used in B without the need for additional qualification. If scopes A and B are downwards-linked, scopes B and A are considered upwards-linked (B <~ A).

Note: link relation is transitive, unless specified otherwise.

Scopes are downwards-linked (DLD) or upwards-linked (ULD) as follows:

Important: linked scopes do not cover cases when identifiers from supertypes are used in subtypes, as this is covered by the inheritance rules.

Identifiers and paths

Kotlin program operates with different entities, such as classes, interfaces, values, etc. An entity can be referenced using its path: a sequence of identifiers which references this entity in a given scope.

Kotlin supports two kinds of paths.

Besides identifiers which are introduced by the developer (e.g., via declaring classes or introducing variables), there are several predefined identifiers with special semantics.


Labels are special syntactic marks which allow one to reference certain code fragments or elements. Lambda expressions and loop statements are allowed to be labeled, with label identifier associated with the corresponding entity.

Note: in Kotlin version 1.3 and earlier, labels were allowed to be placed on any expression or statement.

Labels can be redeclared, meaning that the same label identifier may be reused with different parts of code (or even on the same expression/loop) several times. Labels are scoped, meaning that they are only available in the scope they were declared in.

Labels are used by certain expressions, such as break, continue and return, to specify exactly what entity the expression corresponds to. Please refer to the corresponding sections for details.

When resolving labels (determining which label an expression refers to), the closest label with the matching identifier is chosen, i.e., a label in an innermost scope syntactically closest to the point of its use.