Beman Dawes wrote:
Not sure I understand the request. Could you explain a bit?
At the moment, the library lets you define big and/or little endian types directly in your code. Am I also correct that the main motivation of the library is for IO? If that's the case, instead of writing big64_t my_integer; my_stream >> my_integer; I would intuitively prefer setting the endianness to be similar to setting a property of an iostream, int my_integer; my_stream >> big_endian >> my_integer; because in the first case I am expecting a performance penalty on computing with the big64_t type, which I am not expecting for the native integer type. I know there may be no performance penalty at all, it's just what I'm expecting from the semantics. Hopefully this clarifies it a bit, Kind regards, Rutger