Variable inheritance

Alex Martelli aleaxit at yahoo.com
Tue May 22 23:00:14 CEST 2001


"Roman Suzi" <rnd at onego.ru> wrote in message
news:mailman.990551555.16454.python-list at python.org...
    ...
> MI, is, IMHO, "goto" of OOdesign.
> But who says we can't use it? Just don't tell you've done
> OOD design right.

I will never give you the power to constrain what I can
say about my OO designs.  I consider myself very good
at that, and with a deep understanding of OO from its
bases to its pragmatic issues in all stages of software
development, depending on methodologies, languages,
and other frameworks in use.  And I get paid very good
sums of money because other people are convinced of
the same things about me.

Bertrand Meyer (a person with whom I've had many
differences, but still an outstanding figure in the field)
is widely acknowledged as a guru of OO software
development.  His book by that title is a long-seller
and, I believe, in its third editions.  And he has ALWAYS
campaigned for multiple inheritance -- his language,
Eiffel, pioneered it back when the only others having it
were LISP variants/dialects.

Go tell Dr Meyer he can't say he's done OO design
right, PLEASE... I'd LOVE to see the sparks FLY!


> Python is very flexible and allows everything. I am not happy
> I started MI debate here. Not that I feel defeated, but it
> shows that there is no common understanding of good OO habits.
> OOP remains art. And there is no right thing in an art,

Yeah, right.  Mr Suzi, *YOU* started with an assertion to
the contrary -- that MI is always bad design.  You have
reiterated this at the start of this post by pretending to
tell a LOT of guys widely respected as OO authorities that
they can't say they get their design right.  Forget me, though
I hope to end up editing the OO chapter of the Python
Cookbook as is currently the plan... Meyer; Gamma, Helm,
Johnson, Vlissides; Coplien; Martin; Lakos; Booch, Rumbaugh,
Jacobson... *THESE* are the guys of whom you've said their
OO designs are bad, because their books/languages/tools/
patterns/methodologies/etc all support and may even DEPEND
on multiple-inheritance.  And now they don't even "understand
good OO habits"?!  This is beginning to border on 'crazy'...


Alex







More information about the Python-list mailing list