Path: utzoo!utgpu!news-server.csri.toronto.edu!mailrus!uunet!microsoft!stevebr
From: stevebr@microsoft.UUCP (Steve BRANDLI)
Newsgroups: comp.software-eng
Subject: Re: Cleanroom software engineering
Message-ID: <55058@microsoft.UUCP>
Date: 5 Jun 90 17:43:44 GMT
References: <1990Jun4.112334.20637@lth.se>
Reply-To: stevebr@microsoft.UUCP (Steve BRANDLI)
Organization: Microsoft Corp., Redmond WA
Lines: 37

In article <1990Jun4.112334.20637@lth.se> rn@tts.lth.se (Richard Niklasson) writes:
>
>I have heard some rumours about "Cleanroom software engineering"
>(Harlan D Mills ?). Anyone out there having any reference to books,
>articles, tools.... or any comments on the methodology?
>
>It should be a methodology for program verification (if I got it
>right) and it has been used by NASA (spaceshuttle).

The article you refer to is:

Harlan D. Mills, Michael Dyer, and Richard C. Linger, "Cleanroom Software
Engineering", IEEE Software Sept. 1987 Pgs. 19-24.

Another:

P. Allen Currit, Michael Dyer, and Harlan D. Mills, "Certifying the Reliability
of Software", IEEE Transactions on Software Engineering, Vol. SE-12, No. 1
(Jan. 1986)

It's been awhile since I read these, but I remember them making the case that
defects in software should be tracked statistically, similarly to how defects
in chip manufacture is tracked.  Also, focusing software development on defect
prevention rather than defect protection leads to better quality.  Formal
testing then fills the role of creating feedback of the software development
process rather than role of insuring the quality of a given piece of software.

These ideas have been partly implemented at Microsoft, with a policy we call
zero-defects.  We have found the time spent to insure good quality before the
code is considered "feature complete" gives us overall better quality and also
gives us the bonus of more timely schedule feedback. 

Interesting stuff!

/Steve

I don't speak for Microsoft.