NetNews Usenet Archive 1992 #26/NN_1992_26.iso/spool/comp/std/cplus/1552

Text File | 1992-11-12 | 2.4 KB | 57 lines

Newsgroups: comp.std.c++
Path: sparky!uunet!mcsun!news.funet.fi!network.jyu.fi!sakkinen
From: [e-mail] (Markku Sakkinen)
Subject: Re: c++ implementation (hard)
Message-ID: <[e-mail]>
Keywords: c++
Organization: University of Jyvaskyla, Finland
References: <[e-mail]>
Date: Thu, 12 Nov 1992 14:54:26 GMT
Lines: 45

In article <[e-mail]> [e-mail] (Xristos Kilitzirakis) writes:
> ...
>all right by then.The problem is,that any operator between these
>2 classes , can give any of the 4 classes,depending on many
>things , not only the type of operands.for ex.
>a+b --->c|d|a,
>so I cannot define the return type ,before inspecting the operands.
>
>Any suggestions of implementing (reordering,add a new class,
>  forgetting the hole thing, etc).          ^^^^^^^^^^^^^^^
>
>Well , the implementation is
>a -> m_n matrix
>b-> n-n topelitz matrix (needs only 2*n-1 elems)
>c-> n_n circulant matrix(needs only n elems)
> ...

No wonder that this looked like a potential "standards" problem to you.
However, there is a conventional technique to do this kind of thing;
it is applicable to other statically-typed OOPLs as well.
It's described in Stroustrup's second edition: "13.9 Handle Classes".
Essentially, you define a Matrix class with the correct interface
(this is the handle class), and then any number of classes that
implement the different representations, and whose instances
the Matrix objects create and handle "behind the scenes".
They would probably not be actually derived from Matrix.

The inconvenient thing here is that you must write all member functions
both in the representation classes and in the handle class (which just
passes most calls to the representation object).  Indeed, a delegation
facility was added to an experimental version of C++ together with
the first implementation of multiple inheritance (described in
Stroustrup's paper in the EUUG Spring Conference proceedings, 1987)
to avoid such chores.  However, it was abandoned because it appeared
to be error-prone in use.

----------------------------------------------------------------------
Markku Sakkinen ([e-mail])
       [e-mail] (alternative network address)
Department of Computer Science and Information Systems
University of Jyvaskyla (a's with umlauts)
PL 35
SF-40351 Jyvaskyla (umlauts again)
Finland
----------------------------------------------------------------------