Path: utzoo!mnetor!uunet!husc6!cmcl2!brl-adm!umd5!cvl!elsie!nih-csl!keith
From: keith@nih-csl.UUCP (keith gorlen)
Newsgroups: comp.lang.c++
Subject: Re: C++ Questions (Objective-C)
Message-ID: <290@nih-csl.UUCP>
Date: 27 Dec 87 22:57:51 GMT
References: <2054@cup.portal.com> <2132@cup.portal.com>
Organization: NIH-CSL, Bethesda, MD
Lines: 54
Summary: Some corrections/comments on the OOPS library for C++

In article <2132@cup.portal.com>, Andre_Louis_Marquis@cup.portal.com writes:
> >We are investigating the differences between Objective-C and C++.
> 
> >What class libraries are available for C++?
> Try getting a copy of Keith Gorlen's OOPS Library.  It is perhaps even larger
> than the Objective-C library, and there's no copyright restriction. ...

The OOPS library is about where the Objective-C library was circa
'84-'85.  The current Objective-C library is much larger, and has an
extensive set of classes for interactive graphics.

> OOPS is
> in the "guru-only" stage, but I found it as easy to use and it's free.
> Mr. Gorlen's uucp address is uunet!mimsy!elsie!nih-csl!keith.  Basically,
> you send him a tape and he returns it with a copy of the library and some
> preliminary documentation.

Please do not do this!  I'm trying to get out of the OOPS magtape
distribution business.  A new release of OOPS will be placed on the
C++ tape that USENIX is preparing.  Watch this newsgroup for an
announcement!

In the future, OOPS will be available via anonymous ftp.

> >If I define a common superclass for all objects (Object), as
> >others have mentioned, do I get the flexibility that Objective-C provides?
> >Are there any problems with that strategy?  Since C++ doesn't provide this
> >capability directly, it seems to conflict with C++'s underlying philosophy.
> OOPS defines a common class hierarchy.  You may test object classes and
> phyla.  Virtual methods are inherited and can be overridden.

Objective-C has "Smalltalk-style" messaging, C++ has "Simula-style"
messaging.  C++ messaging is safer and probably faster, Obj-C more
flexible -- and dangerous.

> >Is there any equivalent of [super foo] in C++?  This allows method foo to
> >invoke the inherited method that was overridden.
> You just scope the name -- super_class::foo().

The disadvantage of this is that you must know the name of the base
class, the advantage is that this extends easily to handle multiple
inheritance.

> >Are there any fundamental differences between the messagers, or is that
> >simply an implementation issue?
> There is no dynamic messaging in C++, i.e. no "perform: selector" message, but
> normal messaging is faster.

See above.
-- 
	Keith Gorlen			phone: (301) 496-5363
	Building 12A, Room 2017		uucp: uunet!mimsy!elsie!nih-csl!keith
	National Institutes of Health
	Bethesda, MD 20892