Identifier
An identifier is an arbitrarily long sequence of digits, underscores, lowercase and uppercase Latin letters, and Unicode characters specified using \u
and \U
escape notation (since C99). A valid identifier must begin with a non-digit character (Latin letter, underscore, or Unicode non-digit character (since C99)). Identifiers are case-sensitive (lowercase and uppercase letters are distinct).
It is implementation-defined if raw (not escaped) Unicode characters are allowed in identifiers: char *\U0001f431 = "cat"; // supported char *🐱 = "cat"; // implementation-defined // (e.g. works with Clang, but not GCC prior to version 10) |
(since C99) |
Identifiers can denote the following types of entities:
- objects
- functions
- tags (struct, union, or enumerations)
- structure or union members
- enumeration constants
- typedef names
- label names
- macro names
- macro parameter names
Every identifier other than macro names or macro parameter names has scope, belongs to a name space, and may have linkage. The same identifier can denote different entities at different points in the program, or may denote different entities at the same point if the entities are in different name spaces.
Reserved identifiers
The following identifiers are reserved and may not be declared in a program (doing so invokes undefined behavior):
- The identifiers that are keywords cannot be used for other purposes. In particular #define or #undef of an identifier that is identical to a keyword is not allowed.
- All external identifiers that begin with an underscore.
- All identifiers that begin with an underscore followed by a capital letter or by another underscore (these reserved identifiers allow the library to use numerous behind-the-scenes non-external macros and functions).
- All external identifiers defined by the standard library (in hosted environment). This means that no user-supplied external names are allowed to match any library names, not even if declaring a function that is identical to a library function.
- Identifiers declared as reserved for the implementation or future use by the standard library (see below).
- Identifiers declared as potentially reserved and provided by the implementation (see below). (since C23)
All other identifiers are available. Identifiers that are not reserved or potentially reserved (since C23) can be used with no fear of unexpected collisions when moving programs from one compiler and library to another.
Note: in C++, identifiers with a double underscore anywhere are reserved everywhere; in C, only the ones that begin with a double underscore are reserved.
Reserved and potentially reserved identifiers in the library
The standard library reserves every identifiers it provides. Reserved identifiers that have external linkage (e.g. name of every standard function) are reserved regardless which header is included. Other reserved identifiers are reserved when any of its associated headers is included.
Potentially reserved identifiers are intended to be used by the implementation and future revision of standard. If a potentially reserved identifier is provided by the implementation, it becomes reserved. Implementations are only allowed to provide external definitions of potentially reserved identifiers that are reserved as function names. Potentially reserved identifiers that are not provided by the implementation are not reserved. They can be declared or defined by the user without undefined behavior. However, such usage is not portable. |
(since C23) |
Following identifiers are reserved or potentially reserved (since C23) for the implementation or future use by the standard library.
- function names, all of which are potentially reserved (since C23)
cerf
,cerfc
,cexp2
,cexpm1
,clog10
,clog1p
,clog2
,clgamma
,ctgamma
,csinpi
,ccospi
,ctanpi
,casinpi
,cacospi
,catanpi
,ccompoundn
,cpown
,cpowr
,crootn
,crsqrt
,cexp10m1
,cexp10
,cexp2m1
,clog10p1
,clog2p1
,clogp1
(since C23) and their -f and -l suffixed variants, in <complex.h> (since C99)- beginning with
is
orto
followed by a lowercase letter, in <ctype.h> and <wctype.h> (since C95) - beginning with
str
orwcs
(since C23) followed by a lowercase letter, in <stdlib.h> and <inttypes.h> (since C23) - beginning with
cr_
, in <math.h> (since C23) - beginning with
wcs
followed by a lowercase letter, in <wchar.h> (since C95) - beginning with
atomic_
followed by a lowercase letter, in <stdatomic.h> (since C11) - beginning with
cnd_
,mtx_
,thrd_
ortss_
followed by a lowercase letter, in <threads.h> (since C11)
- typedef names, all of which are potentially reserved (since C23)
- beginning with
int
oruint
and ending with_t
, in <stdint.h> (since C99) - beginning with
atomic_
ormemory_
followed by a lowercase letter, in <stdatomic.h> (since C11) - beginning with
cnd_
,mtx_
,thrd_
ortss_
followed by a lowercase letter, in <threads.h> (since C11)
- beginning with
- macro names
- beginning with
E
followed by a digit or an uppercase letter, in <errno.h> - beginning with
FE_
followed by an uppercase letter, in <fenv.h> (since C99) - beginning with
DBL_
,DEC32_
,DEC64_
,DEC128_
,DEC_
,FLT_
, orLDBL_
followed by an uppercase letter, in <float.h>; these identifiers are potentially reserved (since C23) - beginning with
INT
orUINT
and ending with_MAX
,_MIN
,_WIDTH
(since C23), or_C
, in <stdint.h>; these identifiers are potentially reserved (since C23) (since C99) - beginning with
PRI
orSCN
followed by lowercase letter or the letterX
, in <inttypes.h>; these identifiers are potentially reserved (since C23) (since C99) - beginning with
LC_
followed by an uppercase letter, in <locale.h> - beginning with
FP_
followed by an uppercase letter, in <math.h> (since C23) - beginning with
MATH_
followed by an uppercase letter, in <math.h>; these identifiers are potentially reserved (since C23) - beginning with
SIG
orSIG_
followed by an uppercase letter, in <signal.h> - beginning with
TIME_
followed by an uppercase letter, in <time.h> (since C11) - beginning with
ATOMIC_
followed by an uppercase letter, in <stdatomic.h>; these identifiers are potentially reserved (since C23) (since C11)
- beginning with
- enumeration constants, all of which are potentially reserved (since C23)
- beginning with
memory_order_
followed by a lowercase letter, in <stdatomic.h> (since C11) - beginning with
cnd_
,mtx_
,thrd_
ortss_
followed by a lowercase letter, in <threads.h> (since C11)
- beginning with
Implementations are recommended to warn when on declaration or definition of potentially reserved identifiers, except when
|
(since C23) |
Translation limits
Even though there is no specific limit on the length of identifiers, early compilers had limits on the number of significant initial characters in identifiers and the linkers imposed stricter limits on the names with external linkage. C requires that at least the following limits are supported by any standard-compliant implementation:
|
(until C99) |
|
(since C99) |
References
- C17 standard (ISO/IEC 9899:2018):
- 5.2.4.1 Translation limits (p: 19-20)
- 6.4.2 Identifiers (p: 43)
- 6.10.8 Predefined macro names (p: 127-129)
- 6.11.9 Predefined macro names (p: 130)
- 7.31 Future library directions (p: 332-333)
- K.3.1.2 Reserved identifiers (p: 425)
- C11 standard (ISO/IEC 9899:2011):
- 5.2.4.1 Translation limits (p: 25-26)
- 6.4.2 Identifiers (p: 59-60)
- 6.10.8 Predefined macro names (p: 175-176)
- 6.11.9 Predefined macro names (p: 179)
- 7.31 Future library directions (p: 455-457)
- K.3.1.2 Reserved identifiers (p: 584)
- C99 standard (ISO/IEC 9899:1999):
- 5.2.4.1 Translation limits (p: 20-21)
- 6.4.2 Identifiers (p: 51-52)
- 6.10.8 Predefined macro names (p: 160-161)
- 6.11.9 Predefined macro names (p: 163)
- 7.26 Future library directions (p: 401-402)
- C89/C90 standard (ISO/IEC 9899:1990):
- 2.2.4.1 Translation limits
- 3.1.2 Identifiers
- 3.8.8 Predefined macro names