Subject: Re: ways of funding
From: "Stephen J. Turnbull" <turnbull@sk.tsukuba.ac.jp>
Date: Mon, 8 Nov 1999 13:47:49 +0900 (JST)

>>>>> "rn" == Russell Nelson <nelson@crynwr.com> writes:

    rn> Stephen J. Turnbull writes:
    >> I say "somewhat unrealistic" because reuse of software
    >> components is a hard management problem.

    rn> Not really.  It's a hard payment problem -- transaction costs
    rn> kill you.  If you eschew payments, and

DOS/Windows graphics and games libraries.

    rn> you have a good software characterization system

That's only part of the problem I have in mind, and itself is not
necessarily as easy as setting up a hierarchy of software types.

More important, CPAN does not supply proof of correctness with any of
its modules.

    rn> I'll take a leap of faith, and further assert that scripting
    rn> languages get more reuse than do compiled languages.

Not in my mission-critical apps, they don't.

I'm sure they do get reused more often; this is a matter of dynamic
type-checking or lack thereof.  If you don't care if the script
sometimes fails-safe when you aren't around to watch it, this is
great.  But this is "good-enough" software, not the highly-reliable
stuff that we claim is the result of the open source process.

For high-reliability purposes, this is 1/10-reuse.  (Brooks I think it
was claimed that in the specify-design-code-integrate-test process,
coding was 1/10.)  But you don't know what the specification or design
is until you look at the code, the code typically needs a full review
and some hacking either of the rest of your code or the imported
module's interface, and you have to integrate and test the whole
module.  It's better for software that doesn't need to be reliable,
but I imagine that proprietary vendors do the same.

Of course the _best_ modules in CPAN far exceed the standards I'm
talking about.  But you don't even know which ones they are, do you?

-- 
University of Tsukuba                Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
Institute of Policy and Planning Sciences       Tel/fax: +81 (298) 53-5091
__________________________________________________________________________
__________________________________________________________________________
What are those two straight lines for?  "Free software rules."