home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!icd.ab.com!iccgcc.decnet.ab.com!kambic
- From: kambic@iccgcc.decnet.ab.com (Bonus, Iniquus, Celer - Delegitus Duo)
- Newsgroups: comp.software-eng
- Subject: Re: Test Completion Criteria
- Message-ID: <1992Nov20.160312.9326@iccgcc.decnet.ab.com>
- Date: 20 Nov 92 16:03:12 EST
- References: <1efq9dINN5ps@griffin.orpington.sgp.slb.com> <MARTINC.92Nov20132339@grover.cs.unc.edu>
- Lines: 22
-
- In article <MARTINC.92Nov20132339@grover.cs.unc.edu>, martinc@grover.cs.unc.edu (Charles R. Martin) writes:
- > <1992Nov20.102710.9319@iccgcc.decnet.ab.com>
- > In article <1992Nov20.102710.9319@iccgcc.decnet.ab.com> kambic@iccgcc.decnet.ab.com (Bonus, Iniquus, Celer - Delegitus Duo) writes:
- > Testing is done when: You tested what you agreed to test.
- >
- > Actually, I think this is the canonical answer, which I propose we name
- > Kambic's Law. The amount and kind of testing ought to be part of the
- > specification process (along with specific tests); testing is therefore
- > done when the specification of the tests is fulfilled.
-
- Gulp! Actually this is something testers know intrinsically but which is very
- seldom agreed to by other parts of the development organization. It is
- crucial and very important (as well as satisfying Satchel Paige's rules for
- life for software test engineers) for the testers to establish the contract for
- testing *before* the testing starts. Then the testing is done when the
- contract is satisfied. This is not to say there will not be more testing.
- This removed what I call the delta-function effect on testing near the end of
- a product development.
-
- George Kambic
- sd
-
-