Path: utzoo!utgpu!water!watmath!clyde!att!rutgers!mit-eddie!uw-beaver!cornell!rochester!pt.cs.cmu.edu!f.gp.cs.cmu.edu!dld
From: dld@f.gp.cs.cmu.edu (David Detlefs)
Newsgroups: comp.lang.c++
Subject: A C++ revelation/question...
Message-ID: <2767@pt.cs.cmu.edu>
Date: 22 Aug 88 18:39:31 GMT
Sender: netnews@pt.cs.cmu.edu
Organization: Carnegie-Mellon University, CS/RI
Lines: 41


I recently had a misconception that I had held about C++ corrected.  I
am writing here to 1) save other people the embarassment of insisting
that C++ is X, when it is actually Y, as I did, and 2) get some
clarification on the example in The (Stroustrup) Book that caused me
to think Y was the case.

C++ is of course an object-oriented language.  However, some OOL
literature makes a subtle distinction between languages that are
(truly) "object-based" and those that are "type-based."  The
distinction is that in an object-based language, a member function
(method) of a class only has access to the private members of the
object it is invoked on.  If we neglect some hacks used for debugging,
Smalltalk is an example of a language that is object-oriented in this
way.  "Type-based" languages allow a member function of a class access
to the private part of any object of that class (type) that it has
access to.  CLU and OWL are examples of languages that are type-based
in this sense.

Well, my reading of The Book gave me the impression that C++ fell on the
object-based side of this distinction.  Well, lo and behold, I'm wrong
-- it's type based.  My error is obvious in hindsight -- the "tiny"
example on page 175 can be explained in no other way.  However, I'm
still a little confused.  On page 174, in the "complex" example,
and again on page 178 in the "matrix" example, "operator+" is declared
as a friend instead of as a member function.  This is what originally
led me to believe that C++ was object-based (having missed the
importance of the "tiny" example) -- I thought that these operator+'s
were friends so that they could have access to the private parts of
the second argument and the return matrix or complex.  Well, this
works just as well as a non-friend member function (I just compiled
it.)  My question is, then, why were these friends?  Does it have
something to do with operator overloading name resolution?

By the way, don't get me wrong -- I think the type-base approach is
far superior.
-- 
Dave Detlefs			Any correlation between my employer's opinion
Carnegie-Mellon CS		and my own is statistical rather than causal,
dld@cs.cmu.edu			except in those cases where I have helped to
				form my employer's opinion.  (Null disclaimer.)