🚀 go-pugleaf

In article <EMERY.92Nov2124532@Dr_No.mitre.org> emery@Dr_No.mitre.org (David Emery) writes:
>One of the important topics in reuse, which the POSIX people have
>started to address, is conformance testing.  The current POSIX
>approach of hand-generating test assertions, which are then
>hand-transformed into a test suite, is a good start.  However, there's
>entirely too much hand-xxx in the process.  

I added an entirely new chapter to the last edition of my book (OOP; An 
Evolutionary Approach; Cox and Novabilsky; Addison Wesley) describing why 
conformance testing is so crucial to moving ahead.

But having gone through the experience of actually doing this for a large
library (Stepstone's ICpak101), I'm now convinced that further investment
in this direction is (overstating drastically for emphasis) a thorough waste
of time.

Software development is expensive. Conformance testing *at least* doubles
the expense (even small/cheap test cases easily approximate the size of the
original code). But the revenue from tested components is no greater than
for untested ones.

Conclusion: nothing much will come of either reuse or testing until we've
solved the revenue collection issues for information age goods, particularly
small-granularity software components.

See my article in the October Dr. Dobbs Journal and June Journal of OOP for
a promising approach the Japanese are exploring.

-- 

    Brad Cox bradcox@sitevax.gmu.edu 703 993 1142 work 703 968 8229 home
    Program on Social and Organizational Learning; George Mason University