krees@axion.bt.co.uk (Kearton Rees) (06/19/91)
Does any one have (or know of) information on how Z and VDM handle non-functional requirements? (By non-functional requirements I mean attributes like ease of use, performance, ease of maintenance, availability, security, reliability etc.) Kearton #---------------------------------------------------------------# Kearton Rees krees@axion.bt.co.uk Software Technology Division, British Telecom Research Labs., Martlesham Heath, Ipswich, Suffolk, IP5 7RE United Kingdom. #---------------------------------------------------------------#
wallace@shodha.enet.dec.com (Richard Wallace) (06/20/91)
In article <1991Jun19.143628@axion.bt.co.uk> Kearton Rees (krees@axion.bt.co.uk) writes: >Does any one have (or know of) information on how >Z and VDM handle non-functional requirements? I really don't know how you would resolve the oxymoron of "non-functional requirements." Could it be that you are looking for subjective intangables such as beauty, goodness, etc.? A system, and specifically any system that has software in it, has to have each requirement to be a functional one by definition. If it isn't a function of the system, then it isn't a requirement on the system. >(By non-functional requirements I mean attributes like ease of use, >performance, ease of maintenance, availability, security, reliability etc.) Can't "ease of use" be futher broken down into specifications for: * number of operations per function * logical flow of commands * handling of errors Can't "availabiltiy" be sepcified by a specification of time for the system to be operational (auditing software does this already so I don't see why this can't be specified in Z). "Reliability" is done in the same vein with checksums or some other validation criteria. I know that security CAN be stated as a functional requirement (in Z) as I've done it for Digital for our customer (the U.S. Air Force). "Maintenance" is a bit more intangible than the rest, but it can be formally specified by using a Delta schema to specify the operations for mean time to repair. This schema would be loosely bound with the system of interest. I would maintain that if it can be quantified, it can be specified. This topic could become philosophica! 8-) Richard +-------------------------------------------------------------------+ Richard Wallace wallace%cookie@decpa.dec.com Digital Equipment Corporation 1175 Chapel Hills Dr. Colorado Springs, CO 80920 U.S.A. +-------------------------------------------------------------------+