[quake3] Mergine unix_net.c and win_net.c questions

[Ryan C. Gordon]

> I thought about #defining SOCKET_ERROR -1 and stuff like that, but I
> wasn't sure if that would be acceptable. Is it?

Most games I've worked on end up #define'ing and typedef'ing a handful 
of things at the start to unify WinSock and BSD differences, and then 
use mostly the same code after that:

typedef int SOCKET;
typedef struct hostent HOSTENT;
typedef in_addr IN_ADDR;
typedef struct sockaddr SOCKADDR;
typedef struct sockaddr_in SOCKADDR_IN;
typedef struct linger LINGER;
typedef struct timeval TIMEVAL;
typedef TCHAR* LPSTR;
typedef socklen_t SOCKLEN;
#define INVALID_SOCKET		-1
#define SOCKET_ERROR		-1
#define WSAEWOULDBLOCK		EAGAIN
#define WSAEINPROGRESS		EINPROGRESS
#define WSAENOTSOCK		ENOTSOCK
#define WSATRY_AGAIN		TRY_AGAIN
#define WSAHOST_NOT_FOUND	HOST_NOT_FOUND
#define WSANO_DATA		NO_ADDRESS
#define LPSOCKADDR		sockaddr*
#define WSAECONNRESET		ECONNRESET
#define closesocket		close
#define ioctlsocket		ioctl
#define WSAGetLastError()	errno

(etc)

...these can map from BSD sockets to WinSock too, of course, but I 
usually end up moving WinSock code to Unix and not the other way around.

The number of #ifdef's past that block tends to be extremely minimal.

--ryan.