??? How much effort is required to maintain the test once written. This is tool-dependent. Get your
suppliers to demonstrate the tools first.
??? How much effort is required to run the test.
??? How can the tester can be sure that some action (pressing a screen ???button,??? pulling down a ???pulldown???)
has actually occurred.
??? How well does the tool interface with other tools such as Word and Excel.
In the case of performance test tools there are three moments when you should automate:
??? As soon as you have some idea of the system??™s structure you should consider building a demonstration
model to show if there are any bottlenecks. This may need the developers to build it for
2
Much effort has been expended by test software companies on ensuring that such a cost remains minimal. The
jury remains out on whether they have succeeded in the long term.
Testing Processes and Infrastructure
99
you and for you to create the tests. But note that the cost of this work can be balanced against
the cost of finding out, rather late, that the system is simply too slow.
??? As soon as you have an end-to-end thread available which you can test with. Developer resistance
to this may indicate exactly why you need to do it. An end-to-end thread may be unrealistically
slow since it contains much unoptimized code but at least it will warn developers how much
optimizing remains to be done. And they can remember Michael Jackson??™s two rules for optimizing:
??“ Don??™t optimize.
Pages:
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261