Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!uunet!seismo!rutgers!sri-spam!ames!ptsfa!ihnp4!inuxc!iuvax!pur-ee!j.cc.purdue.edu!k.cc.purdue.edu!l.cc.purdue.edu!cik
From: cik@l.cc.purdue.edu (Herman Rubin)
Newsgroups: news.admin
Subject: Re: Making binary groups obsolete (a week's worth of replies in one message)
Message-ID: <564@l.cc.purdue.edu>
Date: Fri, 7-Aug-87 08:46:42 EDT
Article-I.D.: l.564
Posted: Fri Aug  7 08:46:42 1987
Date-Received: Sun, 9-Aug-87 06:00:55 EDT
References: <266@brandx.rutgers.edu> <8225@utzoo.UUCP> <272@brandx.rutgers.edu> <314@brandx.rutgers.edu>
Organization: Purdue University Statistics Department
Lines: 52
Summary: If we had an adequate language....

In article <314@brandx.rutgers.edu>, webber@brandx.rutgers.edu (Webber) writes:
> 
> Having been away for a week, a considerable pile of misunderstanding
> has accumulated on the question of the relation between a free public
> domain C compiler and the needlessness of binary groups.

>     4) Having programmed and studied a vast number of languages, I feel
>        that C offers the best compromise between language facilities and
>        execution quality of code generated in a straight-forward manner.

I agree, but read further.

>     6) As far as I am concerned (other governments may have other notions),
>        a programmer cannot offer binary-only and deny responsibility
>        for the consequences of bugs in the program.  Since there is nothing
>        that can be done conveniently with binary except execute it, clearly
>        posting binary is an encouragement of others to execute.  Since the
>        recipient can not study the algorithm and judge for theirself whether
>        or not the program will cause trouble, clearly they must rely on
>        the judgement of the poster.  

This may or may not be true.  I have successfully disassembled binaries
when source was not available.  A poster of source has no more obligations
than a vendor of source only.

>     8) Note: I am not claiming everyone must give the net all of their
>        sources, only that if they want to share a program, this is the
>        appropriate way to do it.  Sharing is not mandatory.  For example,
>        my original note stipulated that the mentioned compiler would only
>        be posted to Usenet under certain conditions.
>     9) Finally, ANSI C is/will be the product of a committee, whereas
>        C originally had the flavour of a language designed by a human.

There are many constructs that C does not support.  A particularly nasty
problem is that names are underlined.  For a short program, I would prefer
to disassemble a binary to reading obfuscated C code.  Note that I often
use C, but I would really prefer a "natural" assembler.  The only semi-
natural assemblers that I know are the ones produced by Cray.  For a
language to be acceptable for replacing binaries, _any_ computer operation
or construct that has ever appeared in hardware or been suggested as a
hardware operation _must_ be expressible in the language.

I doubt that I would ever post a program without providing an explanation of
how the program works.  Of course, what I consider sufficiently obvious that
it needs no explaining is different from what you consider so.  I have
produced program that I would consider non-portable, although the _idea_
is portable.  In fact, I am currently working on something for which at most
the program generating the program is portable.
-- 
Herman Rubin, Dept. of Statistics, Purdue Univ., West Lafayette IN47907
Phone: (317)494-6054
hrubin@l.cc.purdue.edu or pur-ee!stat-l!cik or hrubin@purccvm.bitnet