Hi
Im currently writing a server/client system. I use c++ on the server and java on the client.
Therefore I want to write a protocol that handles the byte orders and other communication issues. For example to write a integer onto the socket stream, I do like this:
Code:
void writeInt(int i)
{
write(socket, (void*)((i >> 24) & 0xff), sizeof(char));
write(socket, (void*)((i >> 16) & 0xff), sizeof(char));
write(socket, (void*)((i >> 8) & 0xff), sizeof(char));
write(socket, (void*)(i & 0xff), sizeof(char));
}
This would convert it from little-endian to big-endian. So this works. The problem is if I want to write/read a float or a double. First of all, you can't use the shift operator on a floating-point value. I know on java they have readDouble & writeDouble functions in the DataInput/DataOutput interfaces. I think that the writeDouble function are implemented in a way that they convert the double to a long and then write it over the network. And the readDouble converts it back again.
Maybe I have to do something similar. I know I have to do some bit manipulation, but I don't know what. The problem with java approach, is of course that the size of a long is 4 bytes on the linux box(c++), but on the client(java) it is 8 bytes. So I might have deal with that problem as well.
Anyone who have struggled with something similar?. I will be very thankfull for any help.
Thanks