Path: utzoo!utgpu!watserv1!watmath!att!tut.cis.ohio-state.edu!snorkelwacker!usc!samsung!uunet!mcsun!ukc!mucs!cliff
From: cliff@cs.man.ac.uk (Cliff Jones)
Newsgroups: comp.specification
Subject: Re: specification for nonfunctional requirments ?
Message-ID: <1174@m1.cs.man.ac.uk>
Date: 12 Apr 90 13:21:40 GMT
References: <33855@shemp.CS.UCLA.EDU>
Sender: news@cs.man.ac.uk
Reply-To: cliff@cs.man.ac.uk (Cliff Jones)
Organization: Department of Computer Science, University of Manchester UK
Lines: 21

In article <33855@shemp.CS.UCLA.EDU> swu@maui.cs.ucla.edu (Chi-Sharn Wu) writes:
>Does anyone know any work on specification for nonfunctional requirements,
>especially on fault tolerance and security ?
>Any pointer will be appreciated.
>
>Chi-Sharn Wu
>swu@cs.ucla.edu


The work that I did on concurrency added rely-/guarantee-conditions to
describe "interference" tollerated and caused by shared-variable
programs. I realized some time ago that *some* forms of
fault-tolerance can be specified by the rely-condition alone. I
recently consulted on another project where this idea can yield a 
precise statement of fault-tollerance. Nothing of this work as such is
published - the best source for the rel/guarantee idea is my paper in
the proceedings of the 1983 IFIP Congress. Zhou Chaochen used a
similar idea in describing the faults which can be tolerated by (HDLC)
protocols.

cliff jones