1
00:00:00,190 --> 00:00:02,800
Now let me conclude the lesson by summarizing a few important

2
00:00:02,800 --> 00:00:06,840
aspects of white-box testing. The first important aspect is that white-box

3
00:00:06,840 --> 00:00:09,770
testing works on a formal model. The code itself are models

4
00:00:09,770 --> 00:00:12,430
derived from the code. So when we do white-box testing, we

5
00:00:12,430 --> 00:00:15,160
don't need to make subjective decision, for example, on the level

6
00:00:15,160 --> 00:00:18,890
of obstruction of our models. Normally, we simply represent what's there.

7
00:00:18,890 --> 00:00:22,010
And so what we will obtain are objective, results and objective

8
00:00:22,010 --> 00:00:25,400
measures. As I also said at the beginning, coverage criteria allows

9
00:00:25,400 --> 00:00:28,790
us to compare different test suites, different sets of tests,

10
00:00:28,790 --> 00:00:31,780
because I can measure the coverage achieved by one test suite

11
00:00:31,780 --> 00:00:34,280
and by the other, and then decide which one to use

12
00:00:34,280 --> 00:00:37,400
based on this measure. And again, remember, these measures aren't perfect,

13
00:00:37,400 --> 00:00:40,700
but they at least give you an objective number, an objective

14
00:00:40,700 --> 00:00:44,470
measure of the likely effectiveness of your tests. So even though

15
00:00:44,470 --> 00:00:48,130
achieving 100% coverage does not mean that you identify all the

16
00:00:48,130 --> 00:00:50,710
problems in the code for sure. If your level of coverage

17
00:00:50,710 --> 00:00:53,700
is 10%, for example, for stemen coverage. That means that

18
00:00:53,700 --> 00:00:57,360
you haven't exercised 90% of your code, and therefore the trouble

19
00:00:57,360 --> 00:01:00,565
is in a piece of software that is inadequately tested

20
00:01:00,565 --> 00:01:03,600
and likely to be of inadequate quality. We also saw that

21
00:01:03,600 --> 00:01:07,370
there are two broad classes of coverage criteria, practical criteria

22
00:01:07,370 --> 00:01:10,430
that we can actually use, and theoretical criteria that are interesting

23
00:01:10,430 --> 00:01:13,410
from a conceptual standpoint, but that they are totally impractical.

24
00:01:13,410 --> 00:01:16,170
They are too expensive to be used on real world software.

25
00:01:16,170 --> 00:01:18,350
And finally, as we also said at the beginning,

26
00:01:18,350 --> 00:01:20,820
one of the great things about white box testing and

27
00:01:20,820 --> 00:01:23,910
coverage criteria, is that they are fully automatable. There are

28
00:01:23,910 --> 00:01:27,380
tools that can take your code, instrument it automatically. And

29
00:01:27,380 --> 00:01:29,050
when you run your test cases, they will tell

30
00:01:29,050 --> 00:01:31,330
you, what is the level of coverage that you achieve

31
00:01:31,330 --> 00:01:34,410
with your test at no cost for you. So there's

32
00:01:34,410 --> 00:01:36,763
really no reason not to measure coverage of your code.