1
00:00:00,100 --> 00:00:03,060
There are some important considerations I want to make on final state

2
00:00:03,060 --> 00:00:06,410
machines. And more in general, on model based testing. The first one

3
00:00:06,410 --> 00:00:10,600
is about applicability. Testing based on final state machines is a very

4
00:00:10,600 --> 00:00:13,890
general approach, that we can apply in a number of contexts. And in

5
00:00:13,890 --> 00:00:16,510
particular, if you are working with UML, you have state machines for

6
00:00:16,510 --> 00:00:19,720
free. Because state charts are nothing else but a special kind of

7
00:00:19,720 --> 00:00:22,210
state machine. So you can apply the technique that we just saw

8
00:00:22,210 --> 00:00:26,010
directly on state charts, and try to cover their states and their transitions.

9
00:00:26,010 --> 00:00:29,980
Another important point is that abstraction is key. You have to find the

10
00:00:29,980 --> 00:00:33,280
right level of abstraction. The bigger the system, the more you have to

11
00:00:33,280 --> 00:00:36,130
abstract if you want to represent it with a model, and in particular, with

12
00:00:36,130 --> 00:00:38,710
the final state machine. So it's like having a slider, and you have

13
00:00:38,710 --> 00:00:41,840
to decide where you want to move on that slider. The more you represent,

14
00:00:41,840 --> 00:00:44,700
the more complex your system is going to be and the more thorough your

15
00:00:44,700 --> 00:00:48,110
testing is going to be but also more expensive. The less you represent the

16
00:00:48,110 --> 00:00:51,330
less expensive testing is going to be, but also testing might not be as

17
00:00:51,330 --> 00:00:53,500
thorough as it would be otherwise. So you have to find

18
00:00:53,500 --> 00:00:56,150
the right balance between abstracting the weight too much and abstracting

19
00:00:56,150 --> 00:00:59,840
the weight too little. And finally there are many other approaches.

20
00:00:59,840 --> 00:01:02,840
So we just scratched the surface, and we just saw one possible

21
00:01:02,840 --> 00:01:05,760
approach. But for instance, other models that you can use are

22
00:01:05,760 --> 00:01:09,780
decision tables, flow graphs and even historical models. Models that can

23
00:01:09,780 --> 00:01:13,560
guide your testing based on problems that occurred in your system

24
00:01:13,560 --> 00:01:16,500
in the past. And also, in this case, I'm going to put pointers

25
00:01:16,500 --> 00:01:18,460
to additional materials in the class notes.