Path: utzoo!attcan!uunet!cs.utexas.edu!wuarchive!usc!orion.oac.uci.edu!uci-ics!gateway
From: schmidt@zola.ics.uci.edu (Doug Schmidt)
Newsgroups: comp.lang.c++
Subject: Re: Do class libraries have to be in source form (2 of 4)
Message-ID: <256CEFDE.15973@paris.ics.uci.edu>
Date: 24 Nov 89 07:38:05 GMT
References: <177@taumet.UUCP> <1989Nov22.181203.16204@mentor.com> <1989Nov23.105650.17030@jarvis.csri.toronto.edu>
Reply-To: schmidt@zola.ics.uci.edu (Doug Schmidt)
Distribution: comp
Organization: University of California, Irvine - Dept of ICS
Lines: 55
In-reply-to: sjb@cs.toronto.edu (Stephen Bellantoni)

In article <1989Nov23.105650.17030@jarvis.csri.toronto.edu>, sjb@cs (Stephen Bellantoni) writes:
>If this is true then it may be the death knell for object oriented
>programming. 

[Note: this has certainly been the week to predict the end of OOP...,
was there a fire-sale on crystal balls or something? ;-)]

>In case (2) you are not re-using code, you are modifying it. If this is 
>neccessary for making the best use of object oriented code, then OO has
>failed in its goal of making code re-usable: it merely makes it easier
>to re-use by encouraging a more structured (i.e. easier to modify) style.

I believe this terminology conflates two distinct (but related) forms
of code reuse: Black-Box reuse and White-Box reuse.  Your statements
above pertain mostly to Black-Box reuse, i.e., `reusing a source code
component without modifying its internals or interface in any manner.'
Naturally, if you apply the synecdochical argument that:

                        Black-Box reuse == All reuse

your conclusion above is true by definition.

However, White-Box reuse, i.e., `selectively reusing source code
components by editing portions of their interface or implementation'
also facilitates increased programmer productivity and decrease module
fault-proneness (I'm speaking mostly from personal experience here,
but there are references in the software engineering literature that
address these points more empirically).

I'm sure most good C++ programmers maintain a library of reusable
classes and code templates that are recycled, revised, and refined for
each new project (check out the ./etc and ./gperf subdirectories in
the GNU libg++ release for some examples of mine).  Furthermore,
if/when parameterized types and exception handling become part of the
official C++ definition class library designers and application
programmers will receive increased incentive and support to migrate
the results of their ad hoc, `White-Box reusable' hacks into more
formal and complete `Black-Box reusable' componentry.

I assert that at this stage of OOP's (and C++'s) evolution exploiting
both Black- *and* White-Box reuse is healthy and necessary.  After
all, the main point of software reuse is not to become obsessed with
`purity of essence,' but rather to produce higher quality and more
reliable software systems at lower costs, Questing Quixotically after
the holy-grail of pure Black-Box reuse seems rather counter productive
at this stage of development.

Before dispensing with OOP it is probably useful to understand exactly
what we are dealing with!

Doug
--
----------------------------------------
"The fundamental principle of science, the definition almost, is this:
the sole test of the validity of any idea is experiment." -- R. P. Feynman