home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!ogicse!uwm.edu!cs.utexas.edu!csc.ti.com!tilde.csc.ti.com!mksol!duitz
- From: duitz@mksol.dseg.ti.com (mitchel n duitz)
- Newsgroups: comp.software-eng
- Subject: Re: Who tests?
- Message-ID: <1992Nov13.153637.8643@mksol.dseg.ti.com>
- Date: 13 Nov 92 15:36:37 GMT
- Article-I.D.: mksol.1992Nov13.153637.8643
- References: <1992Nov4.132732.27072@sei.cmu.edu> <1992Nov12.192215.18133@iqsc.COM> <kcampb.721639735@edson>
- Organization: Texas Instruments, Inc
- Lines: 54
-
- In article <kcampb.721639735@edson> kcampb@ee.ualberta.ca (Kevin Campbell) writes:
- >rex@iqsc.COM (Rex Black) writes:
- >
- >>Robert Firth writes:
- >>>So, in conclusion, I don't think more testing, by whomever, is the answer.
- >>>The answer must be to work smarter, not harder; and it's pretty clear
- >>>for the work of Boehm, Mills and others that it is far, far better not
- >>>to put defects into the product in the early days than to make heroic
- >>>efforts to weed them out in the final days.
- >
- >>Here be dragons. I think we need to distinguish between TQM, which is a
- >>methodology for improving processes, and testing, which is a methodology
- >>for finding defects. TQM is about improving a process so that it runs
- >>more efficiently and produces fewer defects. Testing is about finding
- >>defects before the customer does. Robert, if your point is that once
- >>one has a _mature_ TQM program in place throughout the organization one
- >>will expend fewer resources on testing and testing-related coding, then
- >>I'll buy that. However, I believe it would be very foolish for any
- >>company to scrap its testing efforts entirely immediately upon starting
- >>at TQM program. TQM is a long term investment, while testing is a risk-
- >>reduction effort.
- >
- >I like the way Bersoff (et al.) handles this in the book Software
- >Configuration Management An Investment in Product Integrity. In this
- >book he discusses SCM as one of a combination of methods for attaining
- >and maintaining product integity.
- >Product integrity meaning a product:
- > 1) Which fulfills user needs
- > 2) Which can be easily and completely traced through its life cycle
- > 3) which meets specified performance criteria
- > 4) whose cost expectations were met
- > 5) whos delivery expectations were met
- >
- >Bersoff identifies the following product assurance disciplines:
- > 1) Configuration Management
- > 2) Quality assurance
- > 3) Verification and validation
- > 4) Test and evaluation
- >
- >The major point I'd like to emphasis is, as stated by in the above
- >argument, "Product integrity" can only be achieved through
- >combined efforts in all areas. A quality product can only be
- >achieved by ensuring quality at every step in the process.
-
- One area of "Product Integrity" that is continually being ignored
- is reliability. Given the 4 assurance disciplines above does not
- assure quality. Depending on the method of testing it is usually
- inpossible to give a measure of software reliability. Without random
- test cases based upon some sort of usage model, the testing will only
- take care of expected or though out cases. The 4 listed above will give
- some level of confidence in the software, depending upon how serious
- the software engineers too verification and validation.
- duitz@dseg.ti.com
- standard disclaimer - my views and not of TI's.
-