Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!cs.utexas.edu!uunet!mcsun!ukc!reading!cf-cm!andrew
From: andrew@computing-maths.cardiff.ac.uk (Andrew Jones)
Newsgroups: comp.lang.smalltalk
Subject: Re: super super ?
Message-ID: <1123@cf-cm.UUCP>
Date: 30 Nov 89 13:34:17 GMT
References: <2160@cs-spool.calgary.UUCP>
Reply-To: andrew@computing-maths.cardiff.ac.uk (Andrew Jones)
Organization: University of Wales College of Cardiff, Cardiff, WALES, UK.
Lines: 48

In article <2160@cs-spool.calgary.UUCP> sharp@ksi.cpsc.ucalgary.ca (Maurice Sharp) writes:
>
>     Does anyone know the solution to the following problem.  Assume
>that there are 3 classes each of which implements a message called
>foo.  Call the classes Class1, Class2 and Class3, where each is a
>subclass of the other ...
>
>The message 'foo' is defined in all three classes ... THE PROBLEM:
>
>     I want Class3 to call the foo of Class1 NOT Class2.  Is there a
>way to do this ?
>

Ralph Johnson states:

>Strictly speaking, if Class2 overrides a method defined in its superclass,
>Class1, then there is no way that its subclass, Class3, can access the
>definition in Class1.  However, this is not a big problem in practice.

And provides the standard work-around using the example of basicAt: and at:
for collections. This is the way I have always circumvented the problem.

However, if you have a Smalltalk-80 system containing the extension for
multiple inheritance (you will find the class MetaclassForMultipleInheritance
in category Kernel-Classes) then you might like to try defining the
following method in Class3:

foo
     ^self Class1.foo

The compiler will complain that Class1.foo is a new message. Tell it to
proceed at this point. This method will then kind-of use the method of
Class1 for foo.

The problem is, multiple inheritance in Smalltalk-80 is not bug-free,
so you might suddenly find that things go wrong! One certain problem
is that if Class1>>foo sends a message to super, then Class3>>foo will
send the message to Class2, NOT Object.

So if you're braver than I am, you're welcome to try it! Hope this helps.

Andrew Jones,
University of Wales College of Cardiff
Department of Computing Maths.
Maths. Institute
Senghennydd Road
Cardiff CF2 4AG
UK