Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.1 6/24/83; site fortune.UUCP
Path: utzoo!linus!philabs!seismo!hao!hplabs!hpda!fortune!crane
From: crane@fortune.UUCP (John Crane)
Newsgroups: net.lang
Subject: Self-modifying code
Message-ID: <2157@fortune.UUCP>
Date: Tue, 3-Jan-84 14:58:52 EST
Article-I.D.: fortune.2157
Posted: Tue Jan  3 14:58:52 1984
Date-Received: Wed, 4-Jan-84 05:11:48 EST
Organization: Fortune Systems, Redwood City, CA
Lines: 29

I can think of a very good reason for writing self-modifying code or
for a program to generate code which it later executes.

I once wrote a report generator similar to RPG. I wrote it in IBM
assembler because I needed the ability to generate code, mnodify it,
and execute it in real time.  

I look for this ability in a language and consider its presence an
asset and its absence a serious drawback.

I agree that the ALTER statement in COBOL is an abomination as is the
rest of the language as presently used. COBOL could be a useful
business programming language if it had some of the control structures
that C has. 

A lot is made of the ability of C and Ada to have
separately-compiled modules which can be linked together as well as
functions which could be called with a list of parameters. The features
were there in COBOL and PL/I, but programmers  didn't understand their usefulness
and IBM made them so difficult to use that everybody was afraid to try
them.

In my opinion, COBOL's verbosity is not a drawback. It forces
the programmer to document his code. I think every line of a C or
assembler program should be documented even if you as a programmer
think the code is so obvious anybody could understand it. I came to this
conclusion  after years of trying to decipher other people's spaghetti. As a
result, my C programs look a lot like COBOL.