Type support
Basic types
Fundamental types
Fixed width integer types (C++11)
Numeric limits
C numeric limits interface
Runtime type information
Type traits
Type categories
Type properties
(C++11)(deprecated in C++17)
Type trait constants
Supported operations
Relationships and property queries
Type modifications
Type transformations
(C++11)(deprecated in C++17)(C++17)
static const bool traps
(until C++11)
static constexpr bool traps
(since C++11)

The value of std::numeric_limits<T>::traps is true for all arithmetic types T that have at least one value that, if used as an argument to an arithmetic operation, will generate a trap.

[edit] Standard specializations

T value of std::numeric_limits<T>::traps
/* non-specialized */ false
bool false
char usually true
signed char usually true
unsigned char usually true
wchar_t usually true
char16_t usually true
char32_t usually true
short usually true
unsigned short usually true
int usually true
unsigned int usually true
long usually true
unsigned long usually true
long long usually true
unsigned long long usually true
float usually false
double usually false
long double usually false

[edit] Notes

On most platforms integer division by zero always traps, and std::numeric_limits<T>::traps is true for all integer types that support the value 0. The exception is the type bool: even though division by false traps due to integral promotion from bool to int, it is the zero-valued int that traps. Zero is not a value of type bool.

On most platforms, floating-point exceptions may be turned on and off at run time (e.g. feenableexcept() on Linux or _controlfp on Windows), in which case the value of std::numeric_limits<T>::traps for floating-point types reflects the state of floating-point trapping facility at the time of program startup, which is false on most modern systems. An exception would be a DEC Alpha program, where it is true if compiled without -ieee.

[edit] See also

Floating-point environment
identifies floating-point types that detect tinyness before rounding
(public static member constant) [edit]
identifies the floating-point types that detect loss of precision as denormalization loss rather than inexact result
(public static member constant) [edit]