1
00:00:07,132 --> 00:00:09,570
- We've had lots of experiences where

2
00:00:09,570 --> 00:00:12,078
teams have had bad automation practices,

3
00:00:12,078 --> 00:00:14,610
so we'll share a couple of ours right now.

4
00:00:14,610 --> 00:00:17,320
In one team that I was
at, the business people

5
00:00:17,320 --> 00:00:21,438
themselves are not product owners or SMEs,

6
00:00:21,438 --> 00:00:24,078
or Subject Matter Experts,
they were the ones

7
00:00:24,078 --> 00:00:26,152
doing the automation, they were getting

8
00:00:26,152 --> 00:00:27,953
pretty good at it, they liked it,

9
00:00:27,953 --> 00:00:30,101
they were using a Fit style tool.

10
00:00:30,101 --> 00:00:31,514
We had done a homegrown one,

11
00:00:31,514 --> 00:00:34,327
but it was very much like Fit.

12
00:00:34,327 --> 00:00:36,749
The problem was exposed when that test

13
00:00:36,749 --> 00:00:39,463
started running longer, and longer,

14
00:00:39,463 --> 00:00:42,125
it started taking the whole night,

15
00:00:42,125 --> 00:00:44,093
running it every night, so they started

16
00:00:44,093 --> 00:00:46,663
running them in parallel,
broke them up into suites.

17
00:00:46,663 --> 00:00:48,164
That helped for a little bit,

18
00:00:48,164 --> 00:00:50,010
but then it started getting worse.

19
00:00:50,010 --> 00:00:52,170
So I got the team to start looking

20
00:00:52,170 --> 00:00:55,000
at the tests in a different way,

21
00:00:55,000 --> 00:00:57,463
to start looking at why they were

22
00:00:57,463 --> 00:01:00,404
taking so long, started reviewing them.

23
00:01:00,404 --> 00:01:02,811
We found out there were
a lot of duplications

24
00:01:02,811 --> 00:01:05,888
because the people who were automating

25
00:01:05,888 --> 00:01:08,384
don't have test automation experience.

26
00:01:08,384 --> 00:01:10,868
So they would just write up a new one.

27
00:01:10,868 --> 00:01:14,600
They also had a lot of
details that didn't belong.

28
00:01:14,600 --> 00:01:16,592
So when we were trying to debug,

29
00:01:16,592 --> 00:01:20,492
we didn't know what was
actually causing the failure.

30
00:01:20,492 --> 00:01:24,848
So we invested some time,
we actually had somebody,

31
00:01:24,848 --> 00:01:27,741
one of the business users
was a little tech savvy,

32
00:01:27,741 --> 00:01:31,066
so she took it on, and
she put in a month's worth

33
00:01:31,066 --> 00:01:33,629
of work, so it was a man
month's worth of work.

34
00:01:33,629 --> 00:01:35,709
She started cleaning up the tests,

35
00:01:35,709 --> 00:01:37,773
eliminating duplication, taking out

36
00:01:37,773 --> 00:01:42,452
some of those incidental
details as I call them,

37
00:01:42,452 --> 00:01:46,880
we cut our test's run
down into half the time.

38
00:01:46,880 --> 00:01:49,339
So that was a big win.

39
00:01:49,339 --> 00:01:52,928
But the bigger win was
the ease of maintenance.

40
00:01:52,928 --> 00:01:55,306
We were able to now find
out, and look at a test,

41
00:01:55,306 --> 00:01:57,440
and say, "I know why it failed,"

42
00:01:57,440 --> 00:01:59,586
and we could correct it easy.

43
00:01:59,586 --> 00:02:01,677
What about you, Lisa?

44
00:02:01,677 --> 00:02:03,776
- Well, with your story,
it's just kind of like

45
00:02:03,776 --> 00:02:06,682
with Legacy code, you can rescue tests

46
00:02:06,682 --> 00:02:08,372
that aren't necessarily the best

47
00:02:08,372 --> 00:02:10,099
and get some value out of them.

48
00:02:10,099 --> 00:02:13,460
You don't always have
to just toss them out.

49
00:02:13,460 --> 00:02:15,754
In my case, I was working on a team

50
00:02:15,754 --> 00:02:18,189
with no automated tests at all,

51
00:02:18,189 --> 00:02:22,256
and as I said, the developers
were taking some time

52
00:02:22,256 --> 00:02:23,864
to learn test driven involvement,

53
00:02:23,864 --> 00:02:27,021
took them several months to
really get traction on that.

54
00:02:27,021 --> 00:02:29,194
So we didn't, we were all doing the manual

55
00:02:29,194 --> 00:02:31,716
regression testing
scripts, every iteration,

56
00:02:31,716 --> 00:02:33,072
and that was painful.

57
00:02:33,072 --> 00:02:37,117
So, I have decent coding skills,
I was a programmer before,

58
00:02:37,117 --> 00:02:41,162
and so I wrote some smoke
tests through the UI,

59
00:02:41,162 --> 00:02:44,317
using a pretty simple tool to do that,

60
00:02:44,317 --> 00:02:47,450
and I did a pretty good
job of designing them,

61
00:02:47,450 --> 00:02:50,496
except that I feel victim
to chaining the test

62
00:02:50,496 --> 00:02:51,696
'cause it was very difficult

63
00:02:51,696 --> 00:02:53,964
to set up the data for the test.

64
00:02:53,964 --> 00:02:55,213
And I was like, "Well, as long as I've got

65
00:02:55,213 --> 00:02:57,242
"this test user doing this, let's have

66
00:02:57,242 --> 00:02:59,007
"them do all these other things

67
00:02:59,007 --> 00:03:00,184
" 'cause the data's all set up."

68
00:03:00,184 --> 00:03:03,568
So we had chain tests that
were dependent on each other.

69
00:03:03,568 --> 00:03:07,408
And that worked for a
while, but as we changed

70
00:03:07,408 --> 00:03:10,446
the application, those tests became harder

71
00:03:10,446 --> 00:03:12,330
and harder to debug when they failed

72
00:03:12,330 --> 00:03:15,655
because, step number 40 failed, and it was

73
00:03:15,655 --> 00:03:18,665
because of something that
happened up in step number five.

74
00:03:18,665 --> 00:03:21,450
So we actually had to invest time,

75
00:03:21,450 --> 00:03:24,448
we actually had to stop
for two or three weeks,

76
00:03:24,448 --> 00:03:26,538
and not deliver any new features

77
00:03:26,538 --> 00:03:27,725
because you can't really deliver

78
00:03:27,725 --> 00:03:30,208
new features with, you
can't change their test

79
00:03:30,208 --> 00:03:32,269
while you're delivering new features.

80
00:03:32,269 --> 00:03:34,192
So we re-factored those tests so that each

81
00:03:34,192 --> 00:03:36,330
test script had one, clear purpose,

82
00:03:36,330 --> 00:03:39,461
and when they failed, we
knew what had happened.

83
00:03:39,461 --> 00:03:42,246
And we got help from the
developers to do that.

84
00:03:42,246 --> 00:03:45,354
So, it's hard to ask for helps sometimes

85
00:03:45,354 --> 00:03:49,144
and admit that you don't
necessarily know the answers.

86
00:03:49,144 --> 00:03:52,133
- Yes, as you can see,
we all make mistakes.

87
00:03:52,133 --> 00:03:54,221
Don't get hung up about it.

88
00:03:54,221 --> 00:03:57,638
Automation is hard to do well, but learn.

89
00:03:58,829 --> 00:04:01,360
My example with those business users,

90
00:04:01,360 --> 00:04:03,856
once we cleaned them up, I took the time

91
00:04:03,856 --> 00:04:07,661
to teach them what to do so
we didn't run into that again.

92
00:04:07,661 --> 00:04:10,650
You learned not to chain your tests.

93
00:04:10,650 --> 00:04:13,337
Automation really is hard to do well.

94
00:04:13,337 --> 00:04:14,935
But there are some simple ways

95
00:04:14,935 --> 00:04:19,071
you can get benefits, and
an easier learning curve.

96
00:04:19,071 --> 00:04:21,053
So we'll finish off this lesson

97
00:04:21,053 --> 00:04:23,911
on automation with some quick wins.