Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.1 6/24/83; site mit-eddie.UUCP
Path: utzoo!linus!decvax!harpo!floyd!clyde!ihnp4!mit-eddie!gs
From: gs@mit-eddie.UUCP (Gordon Strong)
Newsgroups: net.lang
Subject: Re: CLU mutations
Message-ID: <1004@mit-eddie.UUCP>
Date: Tue, 6-Dec-83 09:27:49 EST
Article-I.D.: mit-eddi.1004
Posted: Tue Dec  6 09:27:49 1983
Date-Received: Fri, 9-Dec-83 01:46:53 EST
References: <1084@uvacs.UUCP>
Organization: MIT, Cambridge, MA
Lines: 20

You've basically got it right on CLU mutations.  For those of you
who haven't used CLU, there are mutable and immutable data types
available for use.  For example, an array is mutable (and they can
grow dynamically).  An immutable array is called a sequence.  Other
types are records and variants (mutable) corresponding to structs
and oneofs (immutable).  You can always define your own data types
in addition to the ones provided.  Creating your own data types is
winning because you can use them to make abstractions.  Modular
programming using abstractions is a very good way of writing large
programs.  For example,  you might want to have a stack abstraction.
Typical operations might be create, push, pop, and empty. You don't 
have to know anything about how it is implemented, just the interface.
This enables you to say things like: stack$create(), stack$push(s,n),
etc.  If anyone wants to see more of an example on CLU usage, I will
post the implementation of the stack abstraction (it's real short).

-- 
Gordon Strong
decvax!genrad!mit-eddie!gs
GS@MIT-XX