问题
C99 states integer types like uint32_t, int16_t etc, where it's easy to see the number of bits used. Good to know in for instance embedded programming.
I have not found any similar types for floating point values. Is there a standard? If not, why?
回答1:
I found the answer in Any guaranteed minimum sizes for types in C?
Quoting Jed Smith (with corrected link to C99 standard):
Yes, the values in float.h and limits.h are system dependent. You should never make assumptions about the width of a type, but the standard does lay down some minimums. See §6.2.5 and §5.2.4.2.1 in the C99 standard.
For example, the standard only says that a char should be large enough to hold every character in the execution character set. It doesn't say how wide it is.
For the floating-point case, the standard hints at the order in which the widths of the types are given:
§6.2.5.10
There are three real floating types, designated as float, double, and long double. 32) The set of values of the type float is a subset of the set of values of the type double; the set of values of the type double is a subset of the set of values of the type long double.
They implicitly defined which is wider than the other, but not specifically how wide they are. "Subset" itself is vague, because a long double can have the exact same range of a double and satisfy this clause.
This is pretty typical of how C goes, and a lot is left to each individual environment. You can't assume, you have to ask the compiler.
回答2:
To answer the "why" part of your question, the name uint32_t, for example, tells you everything you need to know about the type: it's unsigned and exactly 32 bits wide, which implies a range of exactly 0 to 4294967295. (The C standard requires that the uintN_t types have no padding bits or trap representations, and that the signed intN_t types use two's-complement.)
On the other hand, specifying the size in bits of a floating-point type doesn't tell you nearly as much. A hypothetical float64_t would be exactly 64 bits, but the bits that make up a floating-point object are divided into the sign bit, exponent, and significand (mantissa). Furthermore, the meaning of those bits can vary. In most representations, the exponent denotes a power of 2, but it can also be a power of 16. And there are numerous special representations: subnormals, denormals, NaNs, infinities, negative zero.
There are more variations in floating-point representations than can easily be encoded in a type name.
The IEEE floating-point standard (more properly IEC 60559) defines all this -- but if your implementation supports IEC 60559, then the C standard specifies the meanings of float and double anyway (though it allows some more flexibility for long double).
回答3:
While the vast majority of C compilers include a 32-bit integer type, there are many where the only 32-bit type is int and on others the only 32-bit types is long. Having a standard name int32_t mean "one of the signed 32-bit type" avoids the need for the programmer to know which built-in type is 32 bits, except on compilers which use aggressive type-based aliasing.
By contrast, the vast majority of compilers that can use an IEEE single-precision value call it float, and those that can use an IEEE double-precision value call it double. It would not be inconceivable that a compiler might use the names float and double to identify something other than IEEE types and yet include the IEEE types under some other name, but such a scenario would likely be too rare for the Standard to worry about.
来源：https://stackoverflow.com/questions/12332366/are-there-float-and-double-types-with-fixed-sizes-in-c99