Mail Archives: cygwin/1996/11/25/10:53:05
Geoffrey Noer wrote:
>
> Fabricio Chalub taps on the keyboard:
> >
> > The sockets support in gnu-win32 works just fine, but is standing on the
> > thin ice... for example, some functions are missing, some headers points
> > to void and some functions are mapped in a poor way... in
> > <sys/socket.h>, we have this lovely---what should I call---workaround:
> >
> > #define socket cygwin32_socket
> >
> > which of course maps the BSD API to a function that in turn probably
> > calls the Winsock32 API (too many function calls, but I guess that's
> > normal when one maps some system into another.)
> [...]
>
> The way sockets work, we export socket-related functions in the cygwin.dll
> as cygwin32_funcname and then use defines to map these names to the
> ordinary ones.
>
> In normal use, you will get the cygwin32 sockets. However, if you choose
> to use WinSock, you can do so without giving up access to the rest of the
> Cygwin32 API simply by not including the <sys/socket.h> header.
>
> If anyone can think of a better way of doing this, please let me know...
>
But why do you have to define the names as cygwin32_xxx and *then*
redefine as xxx? Why just don't write the functions with their real
names? Are there any conflicts? I mean, I don't think that anyone
would use *both* WinSock and Sockets at the same time. :)
[]s fabricio chalub
-
For help on using this list, send a message to
"gnu-win32-request AT cygnus DOT com" with one line of text: "help".
- Raw text -