🚀 go-pugleaf

In article <Dq0poC.3Iy@news.hawaii.edu> phinely@Hawaii.Edu (Peter Hinely) writes:

   In article <BLUME.96Apr16103345@zayin.cs.princeton.edu>,
   Matthias Blume <blume@zayin.cs.princeton.edu> wrote:
   >In article <goochb.340.00156E50@rwi.com> goochb@rwi.com (William D. Gooch) writes:
   >
   >   I do think there's a serious lack of adequate instruction on 
   >   the proper use of MI.  Perhaps this can be rectified.
   >
   >The only kind of `adequate' instruction on the `proper use' of MI is:
   >
   >	DON'T!
   >
   >BTW, there are people who think that *inheritance* is a bad idea
   >already.  MI is only taking a bad idea to the extreme...
   >

   There are people that think *computers* are a bad idea.  

True.  But those people usually know very little about computers.  On
the other hand, the people I'm talking about know more about
programming language design than a sizable fraction of the readers of
this newsgroup taken together.

   Perhaps you could elaborate a little more why you/they think
   inheritance/multiple inheritance is a bad idea. 

It turns out that inheritance with overiding hasn't been successfully
modelled yet in any clean semantic framework.  There has been some
progress, but usually people stop just short of adding inheritance,
because it messes everything up.

Inheritance with overriding runs counter to any kind of desirable
modularity -- it basically becomes impossible to establish useful
invariants.

The true benefits from contemporary OO languages come from the fact
that they provide means to express abstraction (albeit considerable
caution is necessary) and some limited form of polymorphism.  (Better)
Abstraction and polymorphism are available from other (non-OO)
languages as well.  It is definitely worth having a look at something
like Standard ML, which has a very powerful type- and module-system,
while being soundly based on a cleanly defined semantic framework.

--
-Matthias