Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!usc!apple!snorkelwacker.mit.edu!bloom-beacon!eru!hagbard!sunic!mcsun!cernvax!chx400!chx400!ugun2b!cui!oscar
From: oscar@cui.unige.ch (Oscar Nierstrasz)
Newsgroups: comp.object
Subject: Re: Concurrency vs Inheritence
Message-ID: <5345@cui.unige.ch>
Date: 5 Apr 91 08:15:33 GMT
References: <908@antares.Concordia.CA>
Organization: University of Geneva, Switzerland
Lines: 88

bcheung@omicron.cs.concordia.ca (CHEUNG yik-chi) writes:

>I just begin working on concurrency vs inheritence.
>Is there some people also working on the same thing.
>If yes, may we open this issue for infomation change? and ...etc.
>You can email to me, or post it on this group.

The basic reference is:

%A D.G. Kafura
%A K.H. Lee
%T Inheritance in Actor Based Concurrent Object-Oriented Languages
%J Proceedings of the Third European Conference on Object-oriented Programming
%C Nottingham
%I Cambridge University Press
%D July 10-14, 1989
%P 131-145

The essence of the problem is that (1) synchronization code embedded
in superclasses is hard to reuse by subclasses that must cooperate
with the inherited code, and (2) superclass code has limited reusability
since it has no way of knowing what additional synchronisation policies
the subclasses wish to introduce.  The problem arises because there is no
abstraction of synchronisation available at the interface level visible
to inheriting subclasses.

Kafura and Lee proposed "behaviour abstraction" (abstract states corresponding
to services available at different times -- e.g., full, ampty and partially
full abstract states for a bounded buffer during which different combinations
of put and get operations are available).

A related proposal by Tomlinson and Singh made these abstract states
first class values as "enabled sets" (i.e., indicating which operations
are "enabled").  Neither proposal solves the problem of local delays
due to message contents (e.g., get(N) elements from a buffer may be
delayed if there are less than N elements available).

%A C. Tomlinson
%A V. Singh
%T Inheritance and Synchronization with Enabled Sets
%J ACM SIGPLAN Notices, Proceedings OOPSLA '89
%V 24
%N 10
%P 103-112
%D Oct 1989

Very promising are recent attempts to "unbundle" inheritance to
provide finer control over the way new classes are defined from old.

%A G. Bracha
%A Wm. Cook
%T Mixin-based Inheritance
%J ACM SIGPLAN Notices, Proceedings OOPSLA/ECOOP '90
%V 25
%N 10
%P 303-311
%D Oct 1990

%A A.V. Hense
%T Denotational Semantics of an Object Oriented Programming Language
with Explicit Wrappers
%R Technical report A11/90
%I FB 14, Universit\:at des Saarlandes
%D Nov. 5, 1990
%O submitted for publication

So far no one has really used unbundled inheritance to deal with
concurrent objects, but I think this is the way to go.  We give some
hints in this direction in a paper in last year's OOPSLA.

%A O.M. Nierstrasz
%A M. Papathomas
%T Viewing Objects as Patterns of Communicating Agents
%J ACM SIGPLAN Notices, Proceedings OOPSLA/ECOOP '90
%V 25
%N 10
%P 38-43
%D Oct 1990

Oscar Nierstrasz

-------------------------------------------------------------------
Dr. O.M. Nierstrasz                 | Tel:    41 (22) 787.65.80
University of Geneva                | Fax:            735.39.05
Centre Universitaire d'Informatique | Home:           733.95.68
12 rue du Lac, CH-1207 Geneva       | E-mail: oscar@cui.unige.ch
SWITZERLAND                         |         oscar@cgeuge51.bitnet
-------------------------------------------------------------------