jwcolby
jwcolby at colbyconsulting.com
Thu May 29 11:19:47 CDT 2008
I had a very interesting conversation with my tax guy this morning. He had previously worked at Lowes Hardware in their accounting department. The upshot of the conversation was that Lowes has to interface to a huge number of legacy systems, systems from suppliers, banks, systems from companies they purchased 10 years ago and inherited etc. His comment of interest is that in many cases they did not do testing. Or more correctly they threw data at the live system and looked at what happened, and then used a feedback loop to work around to what the system at the other end would accept. Sounds like poor Roz at her current assignment. I have to tell you that I have experienced this same thing. My disability insurance call center software has to interface to about 10 (so far) completely different insurance companies. Each company has dozens of legacy systems. We get "specs" for how to receive and transmit data to these mainframe systems. I write code to meet the spec, then we send data. We then get feedback from them about how to change the program to meet the "REAL spec" which is never the printed spec. Iterate until the feedback stops and they start accepting what we send without complaint. In one case they had specified how to output claim data that they wanted to enter into their system. In this one (admittedly extreme) case it was in a vertical report format, which we did. Some six months later we discovered that the format was designed to allow some key entry person to key it into their system manually, and the "report order" was the order that the data entry fields appeared on her screen. I got a chuckle out of that one. John W. Colby www.ColbyConsulting.com Gustav Brock wrote: > Hi all > > A local computer paper brought attention to this work related to Daimler-Chrysler but useful in many areas: > > http://www.systematic-testing.com/functional_testing/cte_main.php?cte=1 > > <quote> > This page contains papers on the classification-tree method CTM and the classification-tree editor CTE. The classification-tree method is a testing method for the systematic design of test cases on basis of the specification. The classification-tree editor is a graphical editor supporting the application of the classification-tree method. CTM and CTE are widely used in industrial practice. > </quote> > > The CTE software is free to download and use and the page contains links to a bunch of documentation. > Here's an intro: > > http://www.systematic-testing.com/documents/eurostar2000.pdf > > > > >