[RFA] new tcp_open

[Andrew Cagney]

> Those warnings were always there.  My patch reverted those lines back
> to what was there before.  However, it does appear that changing the char * 
> cast to a void * would eliminate the warning and should probably work 
> everywhere.  Systems that based their sockets on very old BSD have char * 
> instead of void * in their setsockopt prototypes.


Can I suggest either removing the cast or making it ``void *'' with a 
comment to the effect that looser OS's such as solaris 2.5.1 have a 
``char*'' as that argument.

Modern OS's and the Single UNIX spec have ``void *'' as the parameter - 
I checked Red Hat 7.2 and NetBSD 1.5.2 and Solaris 2.7.

http://www.opengroup.org/onlinepubs/007908799/xns/getsockopt.html


> I tested void * on Linux and Windows and neither complain.  I have been
> trying all day on a Solaris box, but it just can't seem to compile
> GDB in less than a couple of days.  How can a big Ultrasparc with
> a load average of 0.08 take > 12 hours to build GDB???


configure appears to be getting really slow again :-/

Andrew