NetNews Usenet Archive 1993 #3/NN_1993_3.iso/spool/comp/lang/cplus/19662

Text File | 1993-01-21 | 1.8 KB | 49 lines

Newsgroups: comp.lang.c++
Path: sparky!uunet!fmrco!pandrews!pandrews
From: [e-mail] (Paul Andrews)
Subject: Re: MI vs. SI
Message-ID: <[e-mail]>
Sender: [e-mail]
Reply-To: [e-mail]
Organization: Fidelity Investments Ltd.
References: <[e-mail]>
Date: Thu, 21 Jan 1993 09:39:18 GMT
Lines: 36

In article [e-mail], [e-mail] (Pete Becker) writes:
> In article <[e-mail]> <[e-mail]> writes:
> >The only way I can think of to enforce this is
> >through the use of multiple inheritance.
>
>     Then use multiple inheritance.  It is not a "Pandora's box", except
> to those who jump into hacking code without thinking about what they are doing.
> But that sort of problem isn't limited to multiple inheritance...
>     -- Pete
>
Totally agree. Why oh why oh why (:-)) do folks have such a downer on MI? YES, I
suppose you can do the same stuff with SI, but thats like saying "I can get from
A to B by walking, why drive?" (well kind of). The point is MI is USEFUL. As people
have said before: "Use it for Mixins". There is too much design of classes that
can't be re-used for various reasons (private not protected members, standard not
virtual functions (esp. destructors), monolithic SI class libraries).

If you want to see the best argument yet for MI, take a look at NIH (it uses SI). The
library would have been much more useful written as a set of Mixins. Programmers could
then decide for themselves which parts of the Object class they wanted in their
programs. (P.S. NIH was first written prior to MI).

Anyway, enough ranting for now.

(Needless to say, my opinions are my own)

- Paul


---
---
Paul Andrews

EMail                Telephone
[e-mail]    [phone]