This page contains an archived post to the Design Forum (formerly called the Flexible Java Forum) made prior to February 25, 2002. If you wish to participate in discussions, please visit the new Artima Forums.

Message:

Classification issues

Posted by Alex Bunardzic on 22 Oct 1998, 10:19 AM

I am a biologist by education, and a software developer by choice.
The concept of defining inheritance by classifying the components
that comprise the problem domain has always been very familiar to
me, but I can appreciate that it can be a stumbling block for someone
who's not formally trained in that area.

In this respect, I agree with the direction Bill's article is pointing
to: if you're muddled about the correct classification, use the damn
composition! Nothing good is ever going to come out of a muddled
and short-circuited inheritance schema. So, definitely, don't go
for inheritance in order to achieve some special effect (like,
polymorphism). Things like late binding and polymorphism should
emerge only as an afterthought, should be the result, not the motive
for classifying things under the inheritance umbrella.

However, I also agree with the criticism offered to Bill: just because
the bulk of people are not yet properly trained in OO design, doesn't
mean that those design principles are invalid. True, composition has
the merit of being based on common-sense, while inheritance requires
transending the common-sense reasoning; but that doesn't make inheritance
in any way inferior.

I still think that developers should be encouraged to master the
proper design methodologies. They should be taught to resist the
temptation to take the path of least resistance. In saying this,
must confess that I am a double-tongued liar, because I cheat in
my OO design all the time. I take all the shortcuts (as God has
intended) like there is no tomorrow. But at least I think I know
the value and the merits of proper classification scheme.

Alex