4.2.1.Binding of names¶

Names refer to objects. Names are introduced by name binding operations.

The following constructs bind names:

• formal parameters to functions,

• class definitions,

• function definitions,

• assignment expressions,

• targets that are identifiers if occurring inan assignment:

  • for loop header,

  • afteras in awith statement,exceptclause,except* clause, or in the as-pattern in structural pattern matching,

  • in a capture pattern in structural pattern matching

• import statements.

• type statements.

• type parameter lists.

Theimport statement of the formfrom...import* binds allnames defined in the imported module, except those beginning with an underscore.This form may only be used at the module level.

A target occurring in adel statement is also considered bound forthis purpose (though the actual semantics are to unbind the name).

Each assignment or import statement occurs within a block defined by a class orfunction definition or at the module level (the top-level code block).

If a name is bound in a block, it is a local variable of that block, unlessdeclared asnonlocal orglobal. If a name is bound atthe module level, it is a global variable. (The variables of the module codeblock are local and global.) If a variable is used in a code block but notdefined there, it is afree variable.

Each occurrence of a name in the program text refers to thebinding ofthat name established by the following name resolution rules.

4.2.2.Resolution of names¶

Ascope defines the visibility of a name within a block. If a localvariable is defined in a block, its scope includes that block. If thedefinition occurs in a function block, the scope extends to any blocks containedwithin the defining one, unless a contained block introduces a different bindingfor the name.

When a name is used in a code block, it is resolved using the nearest enclosingscope. The set of all such scopes visible to a code block is called the block’senvironment.

When a name is not found at all, aNameError exception is raised.If the current scope is a function scope, and the name refers to a localvariable that has not yet been bound to a value at the point where the name isused, anUnboundLocalError exception is raised.UnboundLocalError is a subclass ofNameError.

If a name binding operation occurs anywhere within a code block, all uses of thename within the block are treated as references to the current block. This canlead to errors when a name is used within a block before it is bound. This ruleis subtle. Python lacks declarations and allows name binding operations tooccur anywhere within a code block. The local variables of a code block can bedetermined by scanning the entire text of the block for name binding operations.Seethe FAQ entry on UnboundLocalErrorfor examples.

If theglobal statement occurs within a block, all uses of the namesspecified in the statement refer to the bindings of those names in the top-levelnamespace. Names are resolved in the top-level namespace by searching theglobal namespace, i.e. the namespace of the module containing the code block,and the builtins namespace, the namespace of the modulebuiltins. Theglobal namespace is searched first. If the names are not found there, thebuiltins namespace is searched next. If the names are also not found in thebuiltins namespace, new variables are created in the global namespace.The global statement must precede all uses of the listed names.

Theglobal statement has the same scope as a name binding operationin the same block. If the nearest enclosing scope for a free variable containsa global statement, the free variable is treated as a global.

Thenonlocal statement causes corresponding names to referto previously bound variables in the nearest enclosing function scope.SyntaxError is raised at compile time if the given name does notexist in any enclosing function scope.Type parameterscannot be rebound with thenonlocal statement.

The namespace for a module is automatically created the first time a module isimported. The main module for a script is always called__main__.

Class definition blocks and arguments toexec() andeval() arespecial in the context of name resolution.A class definition is an executable statement that may use and define names.These references follow the normal rules for name resolution with an exceptionthat unbound local variables are looked up in the global namespace.The namespace of the class definition becomes the attribute dictionary ofthe class. The scope of names defined in a class block is limited to theclass block; it does not extend to the code blocks of methods. This includescomprehensions and generator expressions, but it does not includeannotation scopes,which have access to their enclosing class scopes.This means that the following will fail:

classA:a=42b=list(a+iforiinrange(10))

However, the following will succeed:

classA:typeAlias=NestedclassNested:passprint(A.Alias.__value__)# <type 'A.Nested'>

4.2.3.Annotation scopes¶

Annotations,type parameter listsandtype statementsintroduceannotation scopes, which behave mostly like function scopes,but with some exceptions discussed below.

Annotation scopes are used in the following contexts:

• Function annotations.

• Variable annotations.

• Type parameter lists forgeneric type aliases.

• Type parameter lists forgeneric functions.A generic function’s annotations areexecuted within the annotation scope, but its defaults and decorators are not.

• Type parameter lists forgeneric classes.A generic class’s base classes andkeyword arguments are executed within the annotation scope, but its decorators are not.

• The bounds, constraints, and default values for type parameters(lazily evaluated).

• The value of type aliases (lazily evaluated).

Annotation scopes differ from function scopes in the following ways:

• Annotation scopes have access to their enclosing class namespace.If an annotation scope is immediately within a class scope, or within anotherannotation scope that is immediately within a class scope, the code in theannotation scope can use names defined in the class scope as if it wereexecuted directly within the class body. This contrasts with regularfunctions defined within classes, which cannot access names defined in the class scope.

• Expressions in annotation scopes cannot containyield,yieldfrom,await, or:=expressions. (These expressions are allowed in other scopes contained within theannotation scope.)

• Names defined in annotation scopes cannot be rebound withnonlocalstatements in inner scopes. This includes only type parameters, as no othersyntactic elements that can appear within annotation scopes can introduce new names.

• While annotation scopes have an internal name, that name is not reflected in thequalified name of objects defined within the scope.Instead, the__qualname__of such objects is as if the object were defined in the enclosing scope.

4.2.4.Lazy evaluation¶

Most annotation scopes arelazily evaluated. This includes annotations,the values of type aliases created through thetype statement, andthe bounds, constraints, and default values of typevariables created through thetype parameter syntax.This means that they are not evaluated when the type alias or type variable iscreated, or when the object carrying annotations is created. Instead, theyare only evaluated when necessary, for example when the__value__attribute on a type alias is accessed.

Example:

>>>typeAlias=1/0>>>Alias.__value__Traceback (most recent call last):...ZeroDivisionError:division by zero>>>deffunc[T:1/0]():pass>>>T=func.__type_params__[0]>>>T.__bound__Traceback (most recent call last):...ZeroDivisionError:division by zero

Here the exception is raised only when the__value__ attributeof the type alias or the__bound__ attribute of the type variableis accessed.

This behavior is primarily useful for references to types that have notyet been defined when the type alias or type variable is created. For example,lazy evaluation enables creation of mutually recursive type aliases:

fromtypingimportLiteraltypeSimpleExpr=int|ParenthesizedtypeParenthesized=tuple[Literal["("],Expr,Literal[")"]]typeExpr=SimpleExpr|tuple[SimpleExpr,Literal["+","-"],Expr]

Lazily evaluated values are evaluated inannotation scope,which means that names that appear inside the lazily evaluated value are looked upas if they were used in the immediately enclosing scope.

4.2.5.Builtins and restricted execution¶

The builtins namespace associated with the execution of a code blockis actually found by looking up the name__builtins__ in itsglobal namespace; this should be a dictionary or a module (in thelatter case the module’s dictionary is used). By default, when in the__main__ module,__builtins__ is the built-in modulebuiltins; when in any other module,__builtins__ is analias for the dictionary of thebuiltins module itself.

4.2.6.Interaction with dynamic features¶

Name resolution of free variables occurs at runtime, not at compile time.This means that the following code will print 42:

i=10deff():print(i)i=42f()

Theeval() andexec() functions do not have access to the fullenvironment for resolving names. Names may be resolved in the local and globalnamespaces of the caller. Free variables are not resolved in the nearestenclosing namespace, but in the global namespace.[1] Theexec() andeval() functions have optional arguments to override the global and localnamespace. If only one namespace is specified, it is used for both.

4.4.1.General Computing Model¶

Python’s execution model does not operate in a vacuum. It runs ona host machine and through that host’s runtime environment, includingits operating system (OS), if there is one. When a program runs,the conceptual layers of how it runs on the host look somethinglike this:

host machine

process (global resources)

thread (runs machine code)

Each process represents a program running on the host. Think of eachprocess itself as the data part of its program. Think of the process’threads as the execution part of the program. This distinction willbe important to understand the conceptual Python runtime.

The process, as the data part, is the execution context in which theprogram runs. It mostly consists of the set of resources assigned tothe program by the host, including memory, signals, file handles,sockets, and environment variables.

Processes are isolated and independent from one another. (The sameis true for hosts.) The host manages the process’ access to itsassigned resources, in addition to coordinating between processes.

Each thread represents the actual execution of the program’s machinecode, running relative to the resources assigned to the program’sprocess. It’s strictly up to the host how and when that executiontakes place.

From the point of view of Python, a program always starts with exactlyone thread. However, the program may grow to run in multiplesimultaneous threads. Not all hosts support multiple threads perprocess, but most do. Unlike processes, threads in a process are notisolated and independent from one another. Specifically, all threadsin a process share all of the process’ resources.

The fundamental point of threads is that each one doesrunindependently, at the same time as the others. That may be onlyconceptually at the same time (“concurrently”) or physically(“in parallel”). Either way, the threads effectively runat a non-synchronized rate.

4.4.2.Python Runtime Model¶

The same conceptual layers apply to each Python program, with someextra data layers specific to Python:

host machine

process (global resources)

Python global runtime (state)

Python interpreter (state)

thread (runs Python bytecode and “C-API”)

Python threadstate

At the conceptual level: when a Python program starts, it looks exactlylike that diagram, with one of each. The runtime may grow to includemultiple interpreters, and each interpreter may grow to includemultiple thread states.

As a whole, the Python runtime consists of the global runtime state,interpreters, and thread states. The runtime ensures all that statestays consistent over its lifetime, particularly when used withmultiple host threads.

The global runtime, at the conceptual level, is just a set ofinterpreters. While those interpreters are otherwise isolated andindependent from one another, they may share some data or otherresources. The runtime is responsible for managing these globalresources safely. The actual nature and management of these resourcesis implementation-specific. Ultimately, the external utility of theglobal runtime is limited to managing interpreters.

In contrast, an “interpreter” is conceptually what we would normallythink of as the (full-featured) “Python runtime”. When machine codeexecuting in a host thread interacts with the Python runtime, it callsinto Python in the context of a specific interpreter.

Each interpreter completely encapsulates all of the non-process-global,non-thread-specific state needed for the Python runtime to work.Notably, the interpreter’s state persists between uses. It includesfundamental data likesys.modules. The runtime ensuresmultiple threads using the same interpreter will safelyshare it between them.

A Python implementation may support using multiple interpreters at thesame time in the same process. They are independent and isolated fromone another. For example, each interpreter has its ownsys.modules.

For thread-specific runtime state, each interpreter has a set of threadstates, which it manages, in the same way the global runtime containsa set of interpreters. It can have thread states for as many hostthreads as it needs. It may even have multiple thread states forthe same host thread, though that isn’t as common.

Each thread state, conceptually, has all the thread-specific runtimedata an interpreter needs to operate in one host thread. The threadstate includes the current raised exception and the thread’s Pythoncall stack. It may include other thread-specific resources.

Each thread state, over its lifetime, is always tied to exactly oneinterpreter and exactly one host thread. It will only ever be used inthat thread and with that interpreter.

Multiple thread states may be tied to the same host thread, whether fordifferent interpreters or even the same interpreter. However, for anygiven host thread, only one of the thread states tied to it can be usedby the thread at a time.

Thread states are isolated and independent from one another and don’tshare any data, except for possibly sharing an interpreter and objectsor other resources belonging to that interpreter.

Once a program is running, new Python threads can be created using thethreading module (on platforms and Python implementations thatsupport threads). Additional processes can be created using theos,subprocess, andmultiprocessing modules.Interpreters can be created and used with theinterpreters module. Coroutines (async) canbe run usingasyncio in each interpreter, typically onlyin a single thread (often the main thread).

Footnotes

This limitation occurs because the code that is executed by these operationsis not available at the time the module is compiled.