Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!rutgers!tut.cis.ohio-state.edu!uc!nic.MR.NET!ns!hughes
From: hughes@ns.network.com (Jim Hughes x1676)
Newsgroups: comp.lang.c++
Subject: Re: STDARGS problem on sun4
Message-ID: <1618@ns.network.com>
Date: 4 Sep 89 17:57:03 GMT
References: <4150@cuuxb.ATT.COM>
Sender: hughes@ns.network.com (Jim Hughes x1676)
Reply-To: hughes@ns.UUCP (Jim Hughes x1676)
Organization: Network Systems Corporation
Lines: 40

In article <4150@cuuxb.ATT.COM> att!spock!jcd (Jack Dixon) writes:
>
>I'm in the process of porting a C++ application from a sun3 to a sun4 and
>I've run into a problem with stdargs.  I wrote the following little test
>program to show the problem.  I'm using C++ 2.0.  We're running Sun OS 4.0.
>
>The stdarg.h files for sun3 and sun4 are identical.  Does anyone know the fix
>for this or can anyone at least confirm that this problem exists on their
>sun4s?  Thanks much.
>--
>Jack Dixon,  AT&T

If you try using "form(...)" from <streams.h>, I think you will
find the same results.

If you look at the file /usr/include/varargs.h on your sun4 you
will see:

#if defined(sparc)
# define va_alist __builtin_va_alist

This define causes the C compiler to go through some special
gyrations to allow variable arguments.  Even if you merge this 
include file into /usr/include/CC/varargs.h (at least under C++
1.2) the name "__builtin_va_alist" becomes "_au0___builtin_va_alist"
which will NOT trigger the C compiler to produce the correct code from
the "..c" file.

I hope that I am wrong here.  Can anyone tell me is this is still the
case with 2.0. 

Also, is there anyone from Sun (or anywhere else) willing to
talk about a purchasable, supported, with tasking, object version of
the C++ 2.0 port for Sun3 or Sun4?

thanks

jim

hughes@network.com