Path: utzoo!attcan!uunet!mcvax!diku!seindal
From: seindal@diku.dk (Rene' Seindal)
Newsgroups: comp.lang.c++
Subject: Re: Making AT&T C++ for Sun [34]
Message-ID: <3824@diku.dk>
Date: 20 May 88 15:36:57 GMT
References: <1231@ektools.UUCP>
Organization: DIKU, U of Copenhagen, DK
Lines: 40


In article <1231@ektools.UUCP> ken@ektools.UUCP (Kenneth J. Lester) writes:
>This probably has been discussed before, but could someone tell 
>me how to make AT&T C++ version 1.2.1 for the Sun 3 family of
>computers (O/S 3.x).  It makes just fine, but when executed
>via CC, cfront runs and never returns.  I've run the bsd.fix
>script, and have tried compiling it with compilers gcc, and Sun 
>cc compilers from Sun O/S 3.2 and 3.5.

>I know I'm doing something wrong but I followed the same make
>procedure on a Vax running Ultrix 2.2 and it works just fine.
>(Note Ultrix 2.2 is a BSD derivative).

You are not really doing anything wrong.  It is actually an error in
stdio.h.  The beginning og the definition of _iobuf in my stdio.h look
like this:

# ifndef FILE
extern	struct	_iobuf {
#if vax || u3b || u3b2 || u3b5 || mc68k || sparc || mc68020 || mc68010
	int	_cnt;
	char	*_ptr;
#else
	char 	*_ptr;
	int 	_cnt;
#endif

You are missing the test for "sparc" and "mc680[12]0" in your file, so the
two fields was interchanged, when you compiled cfront.

It annoys me a bit that the test is on the cpu type, instead of on the
operating system.  I would like the test to say

#if BSD || SunOS || ...

instead, since this is more an OS dependency than a hardware
dependency.  Unfortunately, none of the C compilers I have seen will
make any defines like this. (why not??)

Rene' Seindal, DIKU, U. of Copenhagen, Denmark. (seindal@diku.dk)