Differences between std::is_integer and std::is_integral?

Question

C++11 provides two type trait template classes: std::is_integer and std::is_integral . However, I cannot tell the differences between them.

What type, say T, can make std::is_integer<T>::value true and make std::is_integral<T>::value false?

Answer 1

std::is_integer<T> does not exist.

That being said, std::numeric_limits<T>::is_integer does exist.

I'm not aware of any significant difference between std::numeric_limits<T>::is_integer and std::is_integral<T> . The latter was designed much later and became standard in C++11, whereas the former was introduced in C++98.

Answer 2

There is no type T that has different results for std::is_integral<T>::value and std::numeric_limits<T>::is_integer . To quote the draft Standard :

3.9.1 Fundamental types [basic.fundamental]

7 Types bool, char, char16_t, char32_t, wchar_t, and the signed and unsigned integer types are collectively called integral types. A synonym for integral type is integer type.[...]

18.3.2.4 numeric_limits members [numeric.limits.members]

static constexpr bool is_integer;

17 True if the type is integer.

20.9.4.1 Primary type categories [meta.unary.cat] (table 47)

template <class T> struct is_integral;

T is an integral type (3.9.1)

Answer 3

std::is_integral_v<T> will only return true for built-in integers.

The standard allows std::numeric_limits<T>::is_integer to be specialized and return true for custom integral types like boost::multiprecion::cpp_int .

Answer 4

std::is_integral<T> and std::numeric_limits<T>::is_integer are NOT the same. Eg boost::multiprecision big ints are treated differently.

This is from a corresponding issue:

is_integral returns information about the nature of the type and is true only for "native" integer types, it should never be true for class types. Which is to say is_integer and is_class are mutually exclusive.

numeric_limits on the other hand returns information on the behaviour of a type - whether it models a particular concept if you will - and so should be specialized for UDT's. Note that specializing is_integer for UDTs would break code, since is_integer implies that a lot of other things are also true like trivial move/copy/init etc.

Answer 5

The difference is that std::is_integral<T> will only recognize decimal integers including bool char char16_t char32_t wchar_t short int long long long . While std::numeric_limits<T>::is_integer will recognize all of those aswell as float double . Look at these two pages for more information: is_integer , is_integral