The byteoder API consists of two layers. One layer is responsible for swapping the byteorder of a type and the high-level API can be used to convert from an exernal byteorder to the host byteorder.
The low-level API consists of a generic swab() function, which is able to swap the byteorder of a type by bytewise copying. This works for all bitwise copyable types. Though, for many types, byteorder sapping can be implemented with better performance. Therefore overloads of swab() are offered for plain data types such as uint16_t that use shift operations.
The swab() function can be overloaded for custom types as well. For example the following struct can have an overloaded swab():
struct value
{
value(char a, uint16_t b, char c)
: _a(a), _b(b), _c(c)
char _a;
uint16_t _b;
char _c;
};
value swab(const value& val)
{
return value(val._a, swab(val.b), val._c);
}
The low-level swab() algorithm is used by the high level functions for external to host byteorder conversions, for example beToHost(). On a little endian system this function needs to swap the byteorder, on a big endian system this funtion should return the value unchanged. The macros PTV_LE and PRV_BE define whether the code should be compiled for a little endian or big endian machine and adjust the behaviour of the high-level conversion functions. If a swab() algorithm is overloaded for a custom type it can be directly used with the high level functions:
value beVal = getBeValue();
value hostVal = beToHost(beVal);
