Prev | Current Page 248 | Next

Peter Farrell-Vinay

"Manage Software Testing"


??? Automated tests don??™t ???see??? some quirks a human can see (and only humans can investigate).
??? Some testers have little programming expertise.
??? Some of the features will be rarely-used and could be left for later.
??? Much of the user interface will change.
??? There will be few releases and little repetition of manual tests.
??? The system uses a lot of special-purpose objects which the tool cannot ???see??? or which require
special programming.
??? You cannot (yet) implement any complete end-to-end use of the system by some user (corresponding
to a use case perhaps).
7.4.3 How Do We Maximize Automated Test RoI?
??? Identify the core test actions and automate them first with later variants added as the system
evolves.
??? Make changes to the tests with spreadsheets full of different data.
??? Reduce the tests to a set of individually-callable functions which can be recombined as test sets.
7.4.4 When Do We Automate Tests
This is tool-dependent. Plan and design test automation as soon as requirements specifications are
available. The outline of a test or a function can often be prepared before the feature is available.
Completing the test (which may involve considerable, some or no scripting, depending on the tool)
however requires that the system be available.
In the case of system test tools the key issues are:
??? Whether the tool can ???see??? the objects on the screen which it has to manipulate.


Pages:
236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260