Path: utzoo!utgpu!news-server.csri.toronto.edu!mailrus!uwm.edu!zaphod.mps.ohio-state.edu!math.lsa.umich.edu!srvr1!cicada.engin.umich.edu!zarnuk
From: zarnuk@caen.engin.umich.edu (Paul Steven Mccarthy)
Newsgroups: comp.software-eng
Subject: Re: Reuse and Abstraction (was: reu
Keywords: off-the-shelf, correctness
Message-ID: <1990Jun1.023724.14244@caen.engin.umich.edu>
Date: 1 Jun 90 02:37:24 GMT
References: <5122@stpstn.UUCP <19855@duke.cs.duke.edu> <5131@stpstn.UUCP>
Sender: news@caen.engin.umich.edu (CAEN Netnews)
Organization: University of Michigan, Ann Arbor
Lines: 29

>(Brad Cox) writes: 
>Because during any major paradigm shift, the conservatives *radically*
>outnumber the revolutionaries. That is why the market for a better C is so
>much larger than the market for off-the-shelf software components. 

This is a chicken-and-the-egg problem.  

Not to impune the putative quality of Stepstone's products (I have never
used any), but just to add my $.02 worth from practical experience:
Every off-the-shelf package that I have used has burned me in the end.
It is exactly an issue of quality and correctness.  I always read all 
available documentation with careful scrutiny and design my systems 
"by the book" -- only to find thousands of wasted development dollars
and hundreds of unrecoverable hours later that the package does not
match the documentation or contains substantive bugs.  (This does not
even address the countless packages that are inadequately defined by
their associated documentation).

I am in unequivocal agreement with both you concerning the need for 
*better* (more/less formal--I don't care) means of establishing
correctness.  More formal systems should be easier to automate, but
that doesn't mean that adequate manual systems can't be developed.
The key concepts are CLARITY and COMPLETENESS.  

As far as off-the-shelf components go, I would be much more comfortable
buying formally specified and tested components than the hit-or-miss 
choices that I'm currently facing.  

---Paul...