1
00:00:00,340 --> 00:00:02,290
So let's talk more about testing, as we said a

2
00:00:02,290 --> 00:00:06,500
little earlier in the lesson, testing means executing the program on

3
00:00:06,500 --> 00:00:09,400
a sample of the input domain, that is of all

4
00:00:09,400 --> 00:00:12,210
the possible input data and I really want to stress that

5
00:00:12,210 --> 00:00:16,190
this sample is tiny sample of the input domain. There

6
00:00:16,190 --> 00:00:19,154
are two important aspects of testing that I'm want to mention here,

7
00:00:19,154 --> 00:00:22,360
there first one is that testing is a dynamic technique. And

8
00:00:22,360 --> 00:00:25,370
what that means is that the program must be executed in

9
00:00:25,370 --> 00:00:28,130
order to perform testing. The second important point is that

10
00:00:28,130 --> 00:00:32,040
testing is an optimistic approximation. And what does it mean

11
00:00:32,040 --> 00:00:35,590
to be optimistic? Well, it means that the program under

12
00:00:35,590 --> 00:00:38,820
test is exercised with a very small subset of all the

13
00:00:38,820 --> 00:00:41,260
possible inputs as we just said. And this is done

14
00:00:41,260 --> 00:00:45,150
under the assumption that the behavior with any other input

15
00:00:45,150 --> 00:00:47,770
is consistent with the behavior shown for the selected subset

16
00:00:47,770 --> 00:00:51,140
of input data, that is why it is an optimistic approach.

17
00:00:51,140 --> 00:00:54,620
Another concept that I want to mention explicitly, is the

18
00:00:54,620 --> 00:00:57,930
concept of successful test. And I'm going to do that,

19
00:00:57,930 --> 00:01:01,260
using another quote. This one from Goodenough and Gerhart

20
00:01:01,260 --> 00:01:03,850
in their paper Towards a Theory of Test Data Selection,

21
00:01:03,850 --> 00:01:06,420
and what the quote says is that a test

22
00:01:06,420 --> 00:01:10,000
is successful if the program fails. And this might sound

23
00:01:10,000 --> 00:01:13,650
counterintuitive, but the point here is that testing cannot

24
00:01:13,650 --> 00:01:16,490
prove the absence of errors, but only reveal their presence.

25
00:01:16,490 --> 00:01:21,550
If a set of tests does not produce any failure, we are either in the extremely

26
00:01:21,550 --> 00:01:24,050
unlikely case of a correct program, or in

27
00:01:24,050 --> 00:01:26,650
the very likely situation of a bad set of

28
00:01:26,650 --> 00:01:30,932
tests that are not able to reveal failures of the program. And that is why we

29
00:01:30,932 --> 00:01:32,730
say that the test is successful if you

30
00:01:32,730 --> 00:01:35,110
can show that there are problems in the program.