1 They key point is to use event listener. Instead of tracking any possible action? Two challenges: (1) select events and (2) select data for events. Data is usually random or manual. 2 They are for Java SWT. Barad combine event-flow and data-flow white-box testing.

They generate event sequence and run in symbolically. It is based on bytecode instrumentation. They introduce symbolic widgets: remove visualization and performance (only functionality).

4 They use Emma for code coverage. Try it. So, they have a symbolic version of GUI. For comparioson, they generate events up to given bounds (length of event chain) using conventional GUI techniquies without SE and with random data.

7 Widget: button, field, etc.

Sergey Vartanov, 2007–2020