Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!seismo!mcvax!diku!seindal
From: seindal@diku.UUCP (Rene Seindal)
Newsgroups: comp.lang.c++
Subject: Re: Standard system header files for non-System V environments?
Message-ID: <3228@diku.UUCP>
Date: Fri, 22-May-87 04:24:20 EDT
Article-I.D.: diku.3228
Posted: Fri May 22 04:24:20 1987
Date-Received: Sat, 23-May-87 18:03:23 EDT
References: <736@daimi.UUCP>
Distribution: world
Organization: DIKU, U of Copenhagen, DK
Lines: 31

In article <736@daimi.UUCP> pederch@daimi.UUCP (Peder Chr. N|rgaard) writes:

>But this causes some minor problems for people who work in other environments
>like BSD 4.n and, even worse, MS-DOS. One problem is that it is some work to 
>build up a complete set of C++ header files from the C header files and
>the descriptions of function parameters in the manuals. Another problem
>is that is error-prone. The worst problem is that the result is not unique,
>at least not when you come to some more complicated system interfaces,
>like the SunView system from Sun Microsystems. So if you use your own
>C++ header files you risk end up with a program which is not even source
>level portable to another perfectly identical system.

>All these problems would be solved if someone, preferably the manufacturer
>of the systems (that is, the Berkeley people for BSD, Sun Microsystems
>for SunView etc), would annouce a specific set of C++ header files and
>say: this is THE C++ header files to use. They need not even be very
>fancy; the most important thing is that they are complete and correct.

I have made a complete set of header files for C++ under BSD 4.3.  I have
considered distributing them, but a lot of them contains code from our
standard header files, so how can I get around the various copyrights.

If some of you out there are interested, there must be some way to get them
around (as ed-scripts, for example).  There is no point in redoing the work
I did (it was dead boring), so mail me a note if you want a copy.

Our files hasn't been tested extensively, but they should be both correct
and complete.

Rene' Seindal	(seindal@diku.UUCP)
The Computer Department, Institute of Datalogy, U. of Copenhagen.