Path: utzoo!utgpu!jarvis.csri.toronto.edu!clyde.concordia.ca!uunet!cs.utexas.edu!wuarchive!kuhub.cc.ukans.edu!markv
From: MARKV@kuhub.cc.ukans.edu (MARK GOODERUM - UNIV. OF KANSAS ACS - MARKV@UKANVAX)
Newsgroups: comp.lang.c++
Subject: Re: How common are C++ Compilers?
Message-ID: <20176@kuhub.cc.ukans.edu>
Date: 20 Dec 89 16:35:12 GMT
References: <3058@ssc-vax.UUCP>
Organization: University of Kansas Academic Computing Services
Lines: 77

>   Well, that depends upon what you'd define 'many different systems' to be.
>   I am working on a graphical user interface for a data management program
>   that has to be ported to:
> 
>   SUN
>   Apollo
>   IBM PC
>   Intergraph
>   MicroVax
>   IBM RT
>   HP3000
>   Macintosh
>   Cimlinc    ( what ?? )
> 
>   To me, that qualifies as 'many different systems'.  I would love to use
>   C++, but I cannot, because I cannot get C++ compilers for all of the
>   above systems.  So, I am stuck writing in plain old C.
> 
>>> If you want to port to a system that only has a C compiler, are there
>>> any chances whatsoever of getting the C code you get from the AT&T C++
>>> compiler to run on it?
> 
>>The AT&T C++ preprocessor 'cfront' used in many systems outputs real honest
>>to goodness C code.  You could in theory compile your code on a C++ system
>>and take the C code back to your little machine.  The problem with this
>>approach is the Cplusplus library that contains many functions reference
>>by the C code cfront produces.  You COULD liscense the stuff from AT&T
>>and port it, but that would be major would and lots of $$.
> 
>   Yeah, this idea occurred to me a while back, but it is unworkable
>   due to the fact that you must link in libraries that have many 
>   built in functions defined.  You need not only the C code, but
>   the C code for the libraries.

I checked a little more, I've been told that initial liscence of cfront
from AT&T is $20,000 plus royalties on sales, although that includes source
to cfront, cpp (the c++ preprocessor), and the C++ libraries.  So it's 
possible but not viable...
 
>>Honestly, I cant think of any system (of sufficient size to support a 
>>'decent' development environmen) that doesnt have C++ available except
>>if your doing embedded systems development, in which case there are 
>>C++ cross development systems coming out for most processor families.
>  
>   Well, how about a Cimlinc?  That sucker was the real clincher in
>   holding me back from using C++.  Of course, no one reading this
>   will even know what a Cimlinc is.  (Cimlincs are discontinued
>   Unix machines which support a 'decent' (word used very usely)
>   development environment, but it will *never* have a C++ compiler.
>   ;-(
Youve got me there, but what kind of Unix does it run?  Have you checked
into porting g++?  (GNU C++)  You need gcc and the binutils.  Sources
for all of these are available at prep.ai.mit.edu for anonymous FTP.

Gcc (and g++) generate real code, the code gereration is as portable as 
can reasonably be expected.  Basically you need to supply 2? machine 
specific configuration files, including things like executable format,
instructions, addressing mode.  There is already support for about a dozen
or more machines, on a variety of processors.  If the Cimlinc uses some
kind of common processor, then the port is even easier.

BTW, speaking of g++/gcc... Does anyone know of any work being done to
port it to RISC (MIPS chips) based Ultrix?  Weve got some DECSystems here
now and we dont' want to reinvent the wheel.
  
> ~~~~~~~~  David Geary, Boeing Aerospace, Seattle  ~~~~~~~~
> ~  Seattle:  America's most attractive city ...          ~
> ~            to the *jetstream*.                         ~
-- 
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Mark Gooderum			Only...		\   Merry Christmas !!!
Academic Computing Services	       ///	  \___________________________
University of Kansas		     ///  /|         __    _
Bix:	  gooderum	      \ \  ///  /__| |\/| | | _   /_\  makes it
Bitnet:   MARKV@UKANVAX		\/\/  /    | |  | | |__| /   \ possible...
Internet: mark@kuhub.cc.ukans.edu
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~