The document discusses reversing the tests pyramid for legacy code by starting with end-to-end tests before introducing unit and integration tests through refactoring. It warns that end-to-end tests are long and costly to maintain, so the goal is to refactor code to introduce unit tests from the outside-in. Maintaining a low technical debt by refactoring during slack times can help pay it back over time. The key is to introduce changes gradually through a reversed tests pyramid rather than rewriting everything from scratch.

1. Reversed Tests Pyramid
Wiktor Żołnowski
@streser
http://www.agileszkolenia.pl
http://www.codespritners.com

2. Agile Coach
Troublemaker
Who am I?
Passionate

3. Can you imagine perfect software?

5. End to End Tests
Functional/Integration Tests
Unit Tests

6. It would be perfect to work with perfect
software every day...

7. But!

8. Reality...

9. Is...

10. Different!

11. Legacy Code

12. How did we get to this point?

13. It's all because of the money...

14. Technical Debt

15. Sometimes Technical Debt could be
considered as something good...

16. Unfortunately the temptation of
easy money is huge...

17. People live in an illusion of continuous,
fast, low cost software development...

18. Forgetting about the Quality...

19. Success in Software Development is something that is
not continuous...

20. Success is a state that you can achieve but also loose
very quickly if you can't respond to changes
fast enough...

21. So, is it possible to deal with legacy code?

22. Reversed Tests Pyramid
End-to-End Tests
Functional/Integration Tests
Unit Tests

23. - What?!

24. You can't do unit testing when there are
no units in your software...

25. You can't do integration tests when
there is nothing to integrate...

26. You need to refactor your code so it
would be testable...

27. Now refactor...
High level tests give you courage to refactor your code...
●

28. But.. There are a few reasons why you shouldn't
reverse tests pyramid
End-To-End tests are too long...

29. End-to-end tests are difficult (costly) to maintain...
If we need end-to-end tests we are probably doing something
wrong with our architecture...

30. So it's all about reversing back our
tests pyramid.

31. Step by step we are introducing
a new architecture...

32. Remember that creating reversed tests
pyramid and reversing it back will take
some time...

33. Remember - WHY are we doing this?

34. Keep your technical debt as low as possible and try to
pay it back every time you can. For example use your
slack time for that!

35. “Leave the world better then you found it”

36. Beware of refactoring just for refactoring!
Few final thoughts...

37. Few final thoughts...
●
●
●
●
●
Beware of refactoring just for refactoring!
Resist temptation to re-write from scratch – history is against you,
such projects usually fail.
Remember to always remove your (duplicated) tests!
Software quality in many cases could be understood as ability to
introduce changes into software!
Keep your technical debt as low as possible and try to pay it back
every time you can. For example use your slack time for that!
Resist temptation to re-write from scratch – history is
against you, such projects usually fail.

38. Few final thoughts...
●
●
●
●
Beware of refactoring just for refactoring!
Resist temptation to re-write from scratch – history is against you,
such projects usually fail.
Remember to always remove your (duplicated) tests!
Software quality always remove be understood as ability to
Remember to in many cases could your (duplicated) tests!
introduce changes into software!
●
Keep your technical debt as low as possible and try to pay it back
every time you can. For example use your slack time for that!

39. False positive or false negative safety
net is even worst than
lack of a safety net...

40. ●
Software quality in many cases could be understood as
ability to introduce changes into software!

41. Wiktor Żołnowski
www.codesprinters.com
Questions?
@streser
http://medium.com/@streser
http://blog.testowka.pl
http://www.agileszkolenia.pl
http://codesprinters.com