100% Test Covɘrage

Transcript

1. 100% Test Covɘrage Myth or Reality? Serge Matveenko twitter.com/lig1
2. Myths
3. Myths • Full coverage doesn’t defend logic errors
4. Myths • Full coverage doesn’t defend logic errors • Full coverage is for Unit Tests
5. Myths • Full coverage doesn’t defend logic errors • Full coverage is for Unit Tests • Full coverage is for TDD
6. Myths • Full coverage doesn’t defend logic errors • Full coverage is for Unit Tests • Full coverage is for TDD • Full coverage leads to very long test runs
7. Myths • Full coverage doesn’t defend logic errors • Full coverage is for Unit Tests • Full coverage is for TDD • Full coverage leads to very long test runs • Full coverage is good but I have no time for that
8. First, count it properly
9. First, count it properly • Linters, static checkers, code analysis — cover everything
10. First, count it properly • Linters, static checkers, code analysis — cover everything • Unit tests — cover everything
11. First, count it properly • Linters, static checkers, code analysis — cover everything • Unit tests — cover everything • Integration tests ◦ Delegate API/contract testing to Unit Tests, ensure API specs are implemented ◦ Delegate interoperability testing to Functional Tests, ensure API do their job
12. First, count it properly • Linters, static checkers, code analysis — cover everything • Unit tests — cover everything • Integration tests ◦ Delegate API/contract testing to Unit Tests, ensure API specs are implemented ◦ Delegate interoperability testing to Functional Tests, ensure API do their job • Functional tests — cover everything again
13. Full coverage doesn’t defend logic errors
14. Full coverage doesn’t defend logic errors • True
15. Full coverage doesn’t defend logic errors • True for Unit Tests
16. Full coverage doesn’t defend logic errors • True for Unit Tests • False for Functional Tests
17. Full coverage is for Unit Tests
18. Full coverage is for Unit Tests • True
19. Full coverage is for Unit Tests • True if you like to carry orphaned code for years
20. Full coverage is for TDD
21. • True Full coverage is for TDD
22. • True • But not exclusively Full coverage is for TDD
23. • True • But not exclusively • Full coverage doesn’t depend on strategie Full coverage is for TDD
24. • True • But not exclusively • Full coverage doesn’t depend on strategie • Full coverage applies for any strategie independently Full coverage is for TDD
25. Full coverage leads to very long test runs
26. Full coverage leads to very long test runs • True
27. Full coverage leads to very long test runs • True when you have a lot to test
28. Full coverage leads to very long test runs • True when you have a lot to test • Automate Smoke Testing on development branches
29. Full coverage leads to very long test runs • True when you have a lot to test • Automate Smoke Testing on development branches • “Namespaces are one honking great idea -- let's do more of those!”
30. Full coverage is good but I have no time for that
31. Full coverage is good but I have no time for that • True
32. Full coverage is good but I have no time for that • True if you like dig a hole for yourself
33. Full coverage is good but I have no time for that • True if you like dig a hole for yourself • Make yourself get used to it • You will learn to do it faster • You will like it • You won’t see you life without it☺
34. To TDD or not to TDD?
35. To TDD or not to TDD? • TDD is good
36. To TDD or not to TDD? • TDD is good • TDD is useless
37. To TDD or not to TDD? • TDD is good • TDD is useless • It doesn’t matter
38. Full Coverage Reality • Have full coverage on different test strategies ◦ Static ◦ Unit ◦ Functional • Know your tools ◦ Smoke ◦ Bisect ◦ Mocks ◦ Fixtures ◦ Log Levels • Full Coverage means 100% ◦ 99% is the same is 1%
39. Thank you! Serge Matveenko github.com/lig twitter.com/lig1