socket length arg fix (SAMBA_2_0 branch)

[Andrew Tridgell]

> The problem, even with your patch, is that there is no
> portable way to discover what prototype a particular
> system is using for the last argument to accept() etc.

actually, I think there is.

You could have configure code that includes the right headers then
declares its own prototype for the function. If the code compiles then
it means that either the OS headers don't prototype that function or
the prototype used matches the OS prototype (this relies on the fact
that most compilers give an error and not just a warning if you have
conflicting prototypes).

so on systems where the OS headers do have a prototype we should be
able to tell. On other systems we just have to go with what we have
now.

Anyone want to implement this?