\rSec0[except]{Exception handling}%

\indextext{exception handling|(}

\gramSec[gram.except]{Exception handling}

\indextext{exception object|see{exception handling, exception object}}%

\indextext{object!exception|see{exception handling, exception object}}

\rSec1[except.pre]{Preamble}

Exception handling provides a way of transferring control and information

from a point in the execution of a thread to an exception handler

associated with a point previously passed by the execution.

A handler will be invoked only by throwing an exception

in code executed in the handler's try block

or in functions called from the handler's try block.

\indextext{\idxcode{try}}%

\begin{bnf}

\nontermdef{try-block}\br

\keyword{try} compound-statement handler-seq

\end{bnf}

\indextext{\idxcode{try}}%

\begin{bnf}

\nontermdef{function-try-block}\br

\keyword{try} \opt{ctor-initializer} compound-statement handler-seq

\end{bnf}

\begin{bnf}

\nontermdef{handler-seq}\br

handler \opt{handler-seq}

\end{bnf}

\indextext{\idxcode{catch}}%

\begin{bnf}

\nontermdef{handler}\br

\keyword{catch} \terminal{(} exception-declaration \terminal{)} compound-statement

\end{bnf}

\begin{bnf}

\nontermdef{exception-declaration}\br

\opt{attribute-specifier-seq} type-specifier-seq declarator\br

\opt{attribute-specifier-seq} type-specifier-seq \opt{abstract-declarator}\br

\terminal{...}

\end{bnf}

The optional \grammarterm{attribute-specifier-seq} in an \grammarterm{exception-declaration}

appertains to the parameter of the catch clause\iref{except.handle}.

\indextext{exception handling!try block}%

\indextext{exception handling!handler}%

\indextext{try block|see{exception handling, try block}}%

\indextext{handler|see{exception handling, handler}}%

A \grammarterm{try-block} is a \grammarterm{statement}\iref{stmt.pre}.

\begin{note}

Within this Clause

``try block'' is taken to mean both \grammarterm{try-block} and

\grammarterm{function-try-block}.

\end{note}

\indextext{exception handling!\idxcode{goto}}%

\indextext{exception handling!\idxcode{switch}}%

\indextext{\idxcode{goto}!and try block}%

\indextext{\idxcode{switch}!and try block}%

\indextext{\idxcode{goto}!and handler}%

\indextext{\idxcode{switch}!and handler}%

The \grammarterm{compound-statement} of a try block or of a handler is a

control-flow-limited statement\iref{stmt.label}.

\begin{example}

\begin{codeblock}

void f() {

goto l1; // error

goto l2; // error

try {

goto l1; // OK

goto l2; // error

l1: ;

} catch (...) {

l2: ;

goto l1; // error

goto l2; // OK

}

}

\end{codeblock}

\end{example}

\indextext{\idxcode{goto}!and try block}%

\indextext{\idxcode{switch}!and try block}%

\indextext{\idxcode{return}!and try block}%

\indextext{\idxcode{continue}!and try block}%

\indextext{\idxcode{goto}!and handler}%

\indextext{\idxcode{switch}!and handler}%

\indextext{\idxcode{return}!and handler}%

\indextext{\idxcode{continue}!and handler}%

A

\keyword{goto},

\keyword{break},

\keyword{return},

or

\keyword{continue}

statement can be used to transfer control out of

a try block or handler.

When this happens, each variable declared in the try block

will be destroyed in the context that

directly contains its declaration.

\begin{example}

\begin{codeblock}

lab: try {

T1 t1;

try {

T2 t2;

if (@\grammarterm{condition}@)

goto lab;

} catch(...) { @\tcode{/* handler 2 */}@ }

} catch(...) { @\tcode{/* handler 1 */}@ }

\end{codeblock}

Here, executing

\tcode{goto lab;}

will destroy first

\tcode{t2},

then

\tcode{t1},

assuming the

\grammarterm{condition}

does not declare a variable.

Any exception thrown while destroying

\tcode{t2}

will result in executing

\tcode{handler 2};

any exception thrown while destroying

\tcode{t1}

will result in executing

\tcode{handler 1}.

\end{example}

\indextext{function try block|see{exception handling, function try block}}%

\indextext{exception handling!function try block}%

A

\grammarterm{function-try-block}

associates a

\grammarterm{handler-seq}

with the

\grammarterm{ctor-initializer},

if present, and the

\grammarterm{compound-statement}.

An exception

thrown during the execution of the

\grammarterm{compound-statement}

or, for constructors and destructors, during the initialization or

destruction, respectively, of the class's subobjects,

transfers control to a handler in a

\grammarterm{function-try-block}

in the same way as an exception thrown during the execution of a

\grammarterm{try-block}

transfers control to other handlers.

\begin{example}

\begin{codeblock}

int f(int);

class C {

int i;

double d;

public:

C(int, double);

};

C::C(int ii, double id)

try : i(f(ii)), d(id) {

// constructor statements

} catch (...) {

// handles exceptions thrown from the ctor-initializer and from the constructor statements

}

\end{codeblock}

\end{example}

In this Clause, ``before'' and ``after'' refer to the

``sequenced before'' relation\iref{intro.execution}.

\rSec1[except.throw]{Throwing an exception}%

\indextext{exception handling!throwing}%

\indextext{throwing|see{exception handling, throwing}}

Throwing an exception transfers control to a handler.

\begin{note}

An exception can be thrown from one of the following contexts:

\grammarterm{throw-expression}{s}\iref{expr.throw},

allocation functions\iref{basic.stc.dynamic.allocation},

\keyword{dynamic_cast}\iref{expr.dynamic.cast},

\keyword{typeid}\iref{expr.typeid},

\grammarterm{new-expression}{s}\iref{expr.new}, and standard library

functions\iref{structure.specifications}.

\end{note}

An object is passed and the type of that object determines which handlers

can catch it.

\begin{example}

\begin{codeblock}

throw "Help!";

\end{codeblock}

can be caught by a

\grammarterm{handler}

of

\keyword{const}

\tcode{\keyword{char}*}

type:

\begin{codeblock}

try {

// ...

} catch(const char* p) {

// handle character string exceptions here

}

\end{codeblock}

and

\begin{codeblock}

class Overflow {

public:

Overflow(char,double,double);

};

void f(double x) {

throw Overflow('+',x,3.45e107);

}

\end{codeblock}

can be caught by a handler for exceptions of type

\tcode{Overflow}:

\begin{codeblock}

try {

f(1.2);

} catch(Overflow& oo) {

// handle exceptions of type \tcode{Overflow} here

}

\end{codeblock}

\end{example}

\indextext{exception handling!throwing}%

\indextext{exception handling!handler}%

\indextext{exception handling!nearest handler}%

When an exception is thrown, control is transferred to the nearest handler with

a matching type\iref{except.handle}; ``nearest'' means the handler

for which the

\grammarterm{compound-statement} or

\grammarterm{ctor-initializer}

following the

\keyword{try}

keyword was most recently entered by the thread of control and not yet exited.

Throwing an exception

initializes a temporary object,

called the

\defnx{exception object}{exception handling!exception object}.

If the type of the exception object would be

an incomplete type\iref{basic.types.general},

an abstract class type\iref{class.abstract},

or a pointer to an incomplete type other than

\cv{}~\keyword{void}\iref{basic.compound}

the program is ill-formed.

\indextext{exception handling!memory}%

\indextext{exception handling!rethrowing}%

\indextext{exception handling!exception object}%

The memory for the exception object is

allocated in an unspecified way, except as noted in~\ref{basic.stc.dynamic.allocation}.

If a handler exits by rethrowing, control is passed to another handler for

the same exception object.

The points of potential destruction for the exception object are:

\begin{itemize}

when an active handler for the exception exits by

any means other than

rethrowing,

immediately after the destruction of the object (if any)

declared in the \grammarterm{exception-declaration} in the handler;

when an object of type \tcode{std::exception_ptr}\iref{propagation}

that refers to the exception object is destroyed,

before the destructor of \tcode{std::exception_ptr} returns.

\end{itemize}

Among all points of potential destruction for the exception object,

there is an unspecified last one

where the exception object is destroyed.

All other points happen before that last one\iref{intro.races}.

\begin{note}

No other thread synchronization is implied in exception handling.

\end{note}

The implementation may then

deallocate the memory for the exception object; any such deallocation

is done in an unspecified way.

\begin{note}

A thrown exception does not

propagate to other threads unless caught, stored, and rethrown using

appropriate library functions; see~\ref{propagation} and~\ref{futures}.

\end{note}

\indextext{exception handling!exception object!constructor}%

\indextext{exception handling!exception object!destructor}%

When the thrown object is a class object, the constructor selected for

the copy-initialization as well as the constructor selected for

a copy-initialization considering the thrown object as an lvalue

shall be non-deleted and accessible, even if the copy/move operation is

elided\iref{class.copy.elision}.

The destructor is potentially invoked\iref{class.dtor}.

\indextext{exception handling!rethrow}%

\indextext{rethrow|see{exception handling, rethrow}}%

An exception is considered caught when a handler for that exception

becomes active\iref{except.handle}.

\begin{note}

An exception can have active handlers and still be considered uncaught if

it is rethrown.

\end{note}

\indextext{exception handling!terminate called@\tcode{terminate} called}%

\indextext{\idxcode{terminate}!called}%

If the exception handling mechanism

handling an uncaught exception\iref{except.uncaught}

directly invokes a function that exits via an

exception, the function \tcode{std::terminate} is invoked\iref{except.terminate}.

\begin{example}

\begin{codeblock}

struct C {

C() { }

C(const C&) {

if (std::uncaught_exceptions()) {

throw 0; // throw during copy to handler's \grammarterm{exception-declaration} object\iref{except.handle}

}

}

};

int main() {

try {

throw C(); // calls \tcode{std::terminate} if construction of the handler's

// \grammarterm{exception-declaration} object is not elided\iref{class.copy.elision}

} catch(C) { }

}

\end{codeblock}

\end{example}

\begin{note}

If a destructor directly invoked by stack unwinding exits via an exception,

\tcode{std::terminate} is invoked.

\end{note}

\rSec1[except.ctor]{Constructors and destructors}%

\indextext{exception handling!constructors and destructors}%

\indextext{constructor!exception handling|see{exception handling, constructors and destructors}}%

\indextext{destructor!exception handling|see{exception handling, constructors and destructors}}

\indextext{unwinding!stack}%

As control passes from the point where an exception is thrown

to a handler,

objects are destroyed by a process,

specified in this subclause, called \defn{stack unwinding}.

Each object with automatic storage duration is destroyed if it has been

constructed, but not yet destroyed,

since the try block was entered.

If an exception is thrown during the destruction of temporaries or

local variables for a \keyword{return} statement\iref{stmt.return},

the destructor for the returned object (if any) is also invoked.

The objects are destroyed in the reverse order of the completion

of their construction.

\begin{example}

\begin{codeblock}

struct A { };

struct Y { ~Y() noexcept(false) { throw 0; } };

A f() {

try {

A a;

Y y;

A b;

return {}; // \#1

} catch (...) {

}

return {}; // \#2

}

\end{codeblock}

At \#1, the returned object of type \tcode{A} is constructed.

Then, the local variable \tcode{b} is destroyed\iref{stmt.jump}.

Next, the local variable \tcode{y} is destroyed,

causing stack unwinding,

resulting in the destruction of the returned object,

followed by the destruction of the local variable \tcode{a}.

Finally, the returned object is constructed again at \#2.

\end{example}

If the initialization of an object

other than by delegating constructor

is terminated by an exception,

the destructor is invoked for

each of the object's subobjects

that were known to be initialized by the object's initialization and

whose initialization has completed\iref{dcl.init}.

\begin{note}

If such an object has a reference member

that extends the lifetime of a temporary object,

this ends the lifetime of the reference member,

so the lifetime of the temporary object is effectively not extended.

\end{note}

\indextext{subobject!initialized, known to be}%

A subobject is \defn{known to be initialized}

if it is not an anonymous union member and

its initialization is specified

\begin{itemize}

\item in \ref{class.base.init} for initialization by constructor,

\item in \ref{class.copy.ctor} for initialization by defaulted copy/move constructor,

\item in \ref{class.inhctor.init} for initialization by inherited constructor,

\item in \ref{dcl.init.aggr} for aggregate initialization,

\item in \ref{expr.prim.lambda.capture} for the initialization of

the closure object when evaluating a \grammarterm{lambda-expression},

\item in \ref{dcl.init.general} for

default-initialization, value-initialization, or direct-initialization

of an array.

\end{itemize}

\begin{note}

This includes virtual base class subobjects

if the initialization

is for a complete object, and

can include variant members

that were nominated explicitly by

a \grammarterm{mem-initializer} or \grammarterm{designated-initializer-clause} or

that have a default member initializer.

\end{note}

If the destructor of an object is terminated by an exception,

each destructor invocation

that would be performed after executing the body of the destructor\iref{class.dtor} and

that has not yet begun execution

is performed.

\begin{note}

This includes virtual base class subobjects if

the destructor was invoked for a complete object.

\end{note}

The subobjects are destroyed in the reverse order of the completion of

their construction. Such destruction is sequenced before entering a

handler of the \grammarterm{function-try-block} of the constructor or destructor,

if any.

If the \grammarterm{compound-statement}

of the \grammarterm{function-body}

of a delegating constructor

for an object exits via

an exception, the object's destructor is invoked.

Such destruction is sequenced before entering a handler of the

\grammarterm{function-try-block} of a delegating constructor for that object, if any.

\begin{note}

If the object was allocated by a \grammarterm{new-expression}\iref{expr.new},

the matching deallocation function\iref{basic.stc.dynamic.deallocation},

if any, is called to free the storage occupied by the object.

\end{note}

\rSec1[except.handle]{Handling an exception}

\indextext{exception handling!handler|(}%

The

\grammarterm{exception-declaration}

in a

\grammarterm{handler}

describes the type(s) of exceptions that can cause

that

\grammarterm{handler}

to be entered.

\indextext{exception handling!handler!incomplete type in}%

\indextext{exception handling!handler!rvalue reference in}%

\indextext{exception handling!handler!array in}%

\indextext{exception handling!handler!pointer to function in}%

The

\grammarterm{exception-declaration}

shall not denote an incomplete type, an abstract class type, or an rvalue reference type.

The

\grammarterm{exception-declaration}

shall not denote a pointer or reference to an

incomplete type, other than ``pointer to \cv{}~\keyword{void}''.

A handler of type

\indextext{array!handler of type}%

``array of \tcode{T}'' or

\indextext{function!handler of type}%

function type \tcode{T}

is adjusted to be of type

``pointer to \tcode{T}''.

\indextext{exception handling!handler!match|(}%

A

\grammarterm{handler}

is a match for

an exception object

of type

\tcode{E}

if

\begin{itemize}

The \grammarterm{handler} is of type \cv{}~\tcode{T} or

\cv{}~\tcode{T\&} and

\tcode{E} and \tcode{T}

are the same type (ignoring the top-level \grammarterm{cv-qualifier}{s}), or

the \grammarterm{handler} is of type \cv{}~\tcode{T} or

\cv{}~\tcode{T\&} and

\tcode{T} is an unambiguous public base class of \tcode{E}, or

the \grammarterm{handler} is of type \cv{}~\tcode{T} or \tcode{const T\&}

where \tcode{T} is a pointer or pointer-to-member type and

\tcode{E} is a pointer or pointer-to-member type

that can be converted to \tcode{T} by one or more of

\begin{itemize}

a standard pointer conversion\iref{conv.ptr} not involving conversions

to pointers to private or protected or ambiguous classes

a function pointer conversion\iref{conv.fctptr}

a qualification conversion\iref{conv.qual}, or

\end{itemize}

the \grammarterm{handler} is of type \cv{}~\tcode{T} or \tcode{const T\&} where \tcode{T} is a pointer or pointer-to-member type and \tcode{E} is \tcode{std::nullptr_t}.

\end{itemize}

\begin{note}

A

\grammarterm{throw-expression}

whose operand is an integer literal with value zero does not match a handler of

pointer or pointer-to-member type.

A handler of reference to array or function type

is never a match for any exception object\iref{expr.throw}.

\end{note}

\begin{example}

\begin{codeblock}

class Matherr { @\commentellip@ virtual void vf(); };

class Overflow: public Matherr { @\commentellip@ };

class Underflow: public Matherr { @\commentellip@ };

class Zerodivide: public Matherr { @\commentellip@ };

void f() {

try {

g();

} catch (Overflow oo) {

// ...

} catch (Matherr mm) {

// ...

}

}

\end{codeblock}

Here, the

\tcode{Overflow}

handler will catch exceptions of type

\tcode{Overflow}

and the

\tcode{Matherr}

handler will catch exceptions of type

\tcode{Matherr}

and of all types publicly derived from

\tcode{Matherr}

including exceptions of type

\tcode{Underflow}

and

\tcode{Zerodivide}.

\end{example}

The handlers for a try block are tried in order of appearance.

\begin{note}

This makes it possible to write handlers that can never be

executed, for example by placing a handler for a final derived class after

a handler for a corresponding unambiguous public base class.

\end{note}

A

\tcode{...}

in a handler's

\grammarterm{exception-declaration}

functions similarly to

\tcode{...}

in a function parameter declaration;

it specifies a match for any exception.

If present, a

\tcode{...}

handler shall be the last handler for its try block.

If no match is found among the handlers for a try block,

the search for a matching

handler continues in a dynamically surrounding try block

of the same thread.

A handler is considered \defnx{active}{exception handling!handler!active} when

initialization is complete for the parameter (if any) of the catch clause.

\begin{note}

The stack will have been unwound at that point.

\end{note}

Also, an implicit handler is considered active when

the function \tcode{std::terminate}

is entered due to a throw. A handler is no longer considered active when the

catch clause exits.

\indextext{currently handled exception|see{exception handling, currently handled exception}}%

The exception with the most recently activated handler that is

still active is called the

\defnx{currently handled exception}{exception handling!currently handled exception}.

If no matching handler is found,

the function \tcode{std::terminate} is invoked;

whether or not the stack is unwound before this invocation of

\tcode{std::terminate}

is \impldef{stack unwinding before invocation of

\tcode{std::terminate}}\iref{except.terminate}.

Referring to any non-static member or base class of an object

in the handler for a

\grammarterm{function-try-block}

of a constructor or destructor for that object results in undefined behavior.

Exceptions thrown in destructors of objects with static storage duration or in

constructors of objects associated with non-block variables with static storage duration are not caught by a

\grammarterm{function-try-block}

on

the \tcode{main} function\iref{basic.start.main}.

Exceptions thrown in destructors of objects with thread storage duration or in constructors of objects associated with non-block variables with thread storage duration are not caught by a

\grammarterm{function-try-block}

on the initial function of the thread.

If a \keyword{return} statement\iref{stmt.return} appears in a handler of the

\grammarterm{function-try-block}

of a

constructor, the program is ill-formed.

The currently handled exception

is rethrown if control reaches the end of a handler of the

\grammarterm{function-try-block}

of a constructor or destructor.

Otherwise, flowing off the end of

the \grammarterm{compound-statement}

of a \grammarterm{handler}

of a \grammarterm{function-try-block}

is equivalent to flowing off the end of

the \grammarterm{compound-statement}

of that function (see \ref{stmt.return}).

The variable declared by the \grammarterm{exception-declaration}, of type

\cv{}~\tcode{T} or \cv{}~\tcode{T\&}, is initialized from the exception object,

of type \tcode{E}, as follows:

\begin{itemize}

if \tcode{T} is a base class of \tcode{E},

the variable is copy-initialized\iref{dcl.init}

from an lvalue of type \tcode{T} designating the corresponding base class subobject

of the exception object;

\item otherwise, the variable is copy-initialized\iref{dcl.init}

from an lvalue of type \tcode{E} designating the exception object.

\end{itemize}

The lifetime of the variable ends

when the handler exits, after the

destruction of any objects with automatic storage duration initialized

within the handler.

When the handler declares an object,

any changes to that object will not affect the exception object.

When the handler declares a reference to an object,

any changes to the referenced object are changes to the

exception object and will have effect should that object be rethrown.%

\indextext{exception handling!handler!match|)}%

\indextext{exception handling!handler|)}

\rSec1[except.spec]{Exception specifications}%

\indextext{exception specification|(}

The predicate indicating whether a function cannot exit via an exception

is called the \defn{exception specification} of the function.

If the predicate is false,

the function has a

\indextext{exception specification!potentially-throwing}%

\defnx{potentially-throwing exception specification}%

{potentially-throwing!exception specification},

otherwise it has a

\indextext{exception specification!non-throwing}%

\defn{non-throwing exception specification}.

The exception specification is either defined implicitly,

or defined explicitly

by using a \grammarterm{noexcept-specifier}

as a suffix of a function declarator\iref{dcl.fct}.

\begin{bnf}

\nontermdef{noexcept-specifier}\br

\keyword{noexcept} \terminal{(} constant-expression \terminal{)}\br

\keyword{noexcept}\br

\end{bnf}

\indextext{exception specification!noexcept!constant expression and}%

In a \grammarterm{noexcept-specifier}, the \grammarterm{constant-expression},

if supplied, shall be a contextually converted constant expression

of type \keyword{bool}\iref{expr.const};

that constant expression is the exception specification of

the function type in which the \grammarterm{noexcept-specifier} appears.

A \tcode{(} token that follows \keyword{noexcept} is part of the

\grammarterm{noexcept-specifier} and does not commence an

initializer\iref{dcl.init}.

The \grammarterm{noexcept-specifier} \keyword{noexcept}

without a \grammarterm{constant-expression}

is

equivalent to the \grammarterm{noexcept-specifier}

\tcode{\keyword{noexcept}(\keyword{true})}.

\begin{example}

\begin{codeblock}

void f() noexcept(sizeof(char[2])); // error: narrowing conversion of value 2 to type \keyword{bool}

void g() noexcept(sizeof(char)); // OK, conversion of value 1 to type \keyword{bool} is non-narrowing

\end{codeblock}

\end{example}

If a declaration of a function

does not have a \grammarterm{noexcept-specifier},

the declaration has a potentially throwing exception specification

unless it is a destructor or a deallocation function

or is defaulted on its first declaration,

in which cases the exception specification

is as specified below

and no other declaration for that function

shall have a \grammarterm{noexcept-specifier}.

In an explicit instantiation\iref{temp.explicit}

a \grammarterm{noexcept-specifier} may be specified,

but is not required.

If a \grammarterm{noexcept-specifier} is specified

in an explicit instantiation,

the exception specification shall be the same as

the exception specification of all other declarations of that function.

A diagnostic is required only if the

exception specifications are not the same

within a single translation unit.

\indextext{exception specification!virtual function and}%

If a virtual function has a

non-throwing exception specification,

all declarations, including the definition, of any function

that overrides that virtual function in any derived class

shall have a non-throwing

exception specification,

unless the overriding function is defined as deleted.

\begin{example}

\begin{codeblock}

struct B {

virtual void f() noexcept;

virtual void g();

virtual void h() noexcept = delete;

};

struct D: B {

void f(); // error

void g() noexcept; // OK

void h() = delete; // OK

};

\end{codeblock}

The declaration of

\tcode{D::f}

is ill-formed because it

has a potentially-throwing exception specification,

whereas

\tcode{B::f}

has a non-throwing exception specification.

\end{example}

\indextext{exception handling!terminate called@\tcode{terminate} called}%

\indextext{\idxcode{terminate}!called}%

Whenever an exception is thrown

and the search for a handler\iref{except.handle}

encounters the outermost block of a function with a

non-throwing exception specification,

the function \tcode{std::terminate} is invoked\iref{except.terminate}.

\begin{note}

An implementation is not permitted to reject an expression merely because, when

executed, it throws or might

throw an exception from a function with a non-throwing exception specification.

\end{note}

\begin{example}

\begin{codeblock}

extern void f(); // potentially-throwing

void g() noexcept {

f(); // valid, even if \tcode{f} throws

throw 42; // valid, effectively a call to \tcode{std::terminate}

}

\end{codeblock}

The call to

\tcode{f}

is well-formed despite the possibility for it to throw an exception.

\end{example}

An expression $E$ is

\defnx{potentially-throwing}{potentially-throwing!expression} if

\begin{itemize}

$E$ is a function call\iref{expr.call}

whose \grammarterm{postfix-expression}

has a function type,

or a pointer-to-function type,

with a potentially-throwing exception specification,

or

$E$ implicitly invokes a function

(such as an overloaded operator,

an allocation function in a \grammarterm{new-expression},

a constructor for a function argument,

or a destructor if $E$ is a full-expression\iref{intro.execution})

that has a potentially-throwing exception specification,

or

$E$ is a \grammarterm{throw-expression}\iref{expr.throw},

or

$E$ is a \keyword{dynamic_cast} expression that casts to a reference type and

requires a runtime check\iref{expr.dynamic.cast},

or

$E$ is a \keyword{typeid} expression applied to a

(possibly parenthesized) built-in unary \tcode{*} operator

applied to a pointer to a

polymorphic class type\iref{expr.typeid},

or

any of the immediate subexpressions\iref{intro.execution}

of $E$ is potentially-throwing.

\end{itemize}

An implicitly-declared constructor for a class \tcode{X},

or a constructor without a \grammarterm{noexcept-specifier}

that is defaulted on its first declaration,

has a potentially-throwing exception specification

if and only if

any of the following constructs is potentially-throwing:

\begin{itemize}

the invocation of a constructor selected by overload resolution

in the implicit definition of the constructor

for class \tcode{X}

to initialize a potentially constructed subobject, or

a subexpression of such an initialization,

such as a default argument expression, or,

for a default constructor, a default member initializer.

\end{itemize}

\begin{note}

Even though destructors for fully-constructed subobjects

are invoked when an exception is thrown

during the execution of a constructor\iref{except.ctor},

their exception specifications do not contribute

to the exception specification of the constructor,

because an exception thrown from such a destructor

would call the function \tcode{std::terminate}

rather than escape the constructor\iref{except.throw,except.terminate}.

\end{note}

The exception specification for an implicitly-declared destructor,

or a destructor without a \grammarterm{noexcept-specifier},

is potentially-throwing if and only if

any of the destructors

for any of its potentially constructed subobjects

has a potentially-throwing exception specification or

the destructor is virtual and the destructor of any virtual base class

has a potentially-throwing exception specification.

The exception specification for an implicitly-declared assignment operator,

or an assignment-operator without a \grammarterm{noexcept-specifier}

that is defaulted on its first declaration,

is potentially-throwing if and only if

the invocation of any assignment operator

in the implicit definition is potentially-throwing.

A deallocation function\iref{basic.stc.dynamic.deallocation}

with no explicit \grammarterm{noexcept-specifier}

has a non-throwing exception specification.

The exception specification for a comparison operator function\iref{over.binary}

without a \grammarterm{noexcept-specifier}

that is defaulted on its first declaration

is potentially-throwing if and only if

any expression

in the implicit definition is potentially-throwing.

\begin{example}

\begin{codeblock}

struct A {

A(int = (A(5), 0)) noexcept;

A(const A&) noexcept;

A(A&&) noexcept;

~A();

};

struct B {

B() noexcept;

B(const B&) = default; // implicit exception specification is \tcode{\keyword{noexcept}(\keyword{true})}

B(B&&, int = (throw 42, 0)) noexcept;

~B() noexcept(false);

};

int n = 7;

struct D : public A, public B {

int * p = new int[n];

// \tcode{D::D()} potentially-throwing, as the \keyword{new} operator may throw \tcode{bad_alloc} or \tcode{bad_array_new_length}

// \tcode{D::D(const D\&)} non-throwing

// \tcode{D::D(D\&\&)} potentially-throwing, as the default argument for \tcode{B}'s constructor may throw

// \tcode{D::\~D()} potentially-throwing

};

\end{codeblock}

Furthermore, if

\tcode{A::\~{}A()}

were virtual,

the program would be ill-formed since a function that overrides a virtual

function from a base class

shall not have a potentially-throwing exception specification

if the base class function has a non-throwing exception specification.

\end{example}

An exception specification is considered to be \defnx{needed}{needed!exception specification} when:

\begin{itemize}

\item in an expression, the function is selected by

overload resolution\iref{over.match,over.over};

\item the function is odr-used\iref{term.odr.use} or, if it appears in an

unevaluated operand, would be odr-used if the expression were

potentially-evaluated;

\item the exception specification is compared to that of another

declaration (e.g., an explicit specialization or an overriding virtual

function);

\item the function is defined; or

\item the exception specification is needed for a defaulted

function that calls the function.

\begin{note}

A defaulted declaration does not require the

exception specification of a base member function to be evaluated

until the implicit exception specification of the derived

function is needed, but an explicit \grammarterm{noexcept-specifier} needs

the implicit exception specification to compare against.

\end{note}

\end{itemize}

The exception specification of a defaulted

function is evaluated as described above only when needed; similarly, the

\grammarterm{noexcept-specifier} of a specialization of a function

template or member function of a class template is instantiated only when

needed.

\indextext{exception specification|)}