Xref: utzoo comp.lang.smalltalk:572 comp.lang.c++:1193 comp.lang.misc:1628
Path: utzoo!attcan!uunet!grand!day
From: day@grand.UUCP (Dave Yost)
Newsgroups: comp.lang.smalltalk,comp.lang.c++,comp.lang.misc
Subject: Re: Request for book review : OOP books
Summary: good book
Message-ID: <412@grand.UUCP>
Date: 27 May 88 17:46:14 GMT
References: <3274@pdn.UUCP>
Reply-To: day@grand.UUCP (Dave Yost)
Organization: Grand Software, Inc., Los Angels, CA 213-650-1089
Lines: 73

In article <3274@pdn.UUCP> bob@pdn.UUCP (Bob Hickle) writes:
>Has anyone out there read these books?  If so, how about posting
>a brief book review.
>
>[one of] The books:
>
>Object-oriented Software Construction by Bertrand Meyer.
>May, 1988, 534 pages, cloth book.
>Price on the flyer from Prentice-Hall is
>$31.45 (a 15% discount).
>
>    [ This book appears to present o-o concepts and design
>      using the Eiffel programming language for its examples.
>      It is a long book.  Is quantity = quality here?  Is
>      this just an advertisement for Eiffel? ]

So far, I've read the first 9 chapters, and have peeked
at some later chapters.  This is a very well-written,
readable book, loaded with historical references.  The
author's presentation of his point of view is well-
motivated and well-organized, and he nearly always
presents and refutes interesting objections to his line
of reasoning.  The book starts with an astute survey
of existing software design practices and problems, and
lays out a set of desirable goals.  He then proceeds to
present his version of the principles of OO design and
shows how they meet many of those goals.

To illustrate and advance the discussion, he presents
his Eiffel language and uses it for the examples.
The book could not have been written any other way
because there is no other existing language that
could demonstrate his view of how software should be
constructed.  If this makes the book an advertisement
for Eiffel, then it is advertisement at the very
loftiest, most educational level.

One of the main thrusts of the book and of the Eiffel
language is the goal of producing error-free software.
The language provides for a very interesting
integration of classes and assertions.  The assertions
serve not only to generate optional runtime checking
code, but as self-documentation of the semantics of
classes and their functions and procedures, and they
mesh with inheritance in a very elegant way.

One message of the book comes through loud and clear:
if you want to write error-free software, you want as
much help from the language as possible; the more the
language enforces sanity the better.  This leads to the
obvious (unstated) conclusion that C (or pascal or
whatever) extended with optional higher-level features
(such as classes) is not as powerful a tool for the job
as a well-designed purely higher-level language.  (Is
it a wonder that C wasn't implemented as extensions to
assembler language?  I guess that was back when they
didn't care if anyone else ever used it.)

One amazing tidbit about the book is that the author is
French (though he's been mostly stateside since '83),
yet his writing in English is superb.  I guess he's
good at languages (he also has a Master's degree in the
Russian language.)  He has an M.S. from Stanford and a
Ph.D. from the University of Nancy in France.

My perspective:
I have programmed in C for nine years, and as time goes
by I have become increasingly impatient with the grunge
level of C.  My quest for a higher level language has
led me to Icon, C++, and Eiffel, which I hope to try
soon.

 --dave yost