The Test Smell Before Christmas

A small presentation about test smells I gave to open the Drupal Community Day 2018 in Turin, Italy.

Transcript

1. T H E
   T E S T S M E L L
   B E F O R E
   C H R I S T M A S
   

   View Slide

2. T E S T S S M E L L
   B U T T H A T O F F E R S G O O D I N S I G H T S O N T H E C O D E
   Code smells
   Behaviour smells
   Project smells
   

   View Slide

3. H A R D T O T E S T
   C O D E
   SYMPTOMS
   • some code is inherently
   difficult to test (GUI,
   threads, remote services)
   • tight coupling
   • private logic
   • lots of dependencies
   

   View Slide

4. H A R D T O T E S T
   C O D E
   IMPACT
   • testing becomes very hard
   and time consuming
   • less tests will be written
   • does not scale
   

   View Slide

5. H A R D T O T E S T
   C O D E
   POSSIBLE SOLUTIONS
   • learn to decouple code
   • separate logic from the
   async mechanism
   • keep test logic
   unconditional
   

   View Slide

6. View Slide

7. O B S C U R E
   T E S T S
   SYMPTOMS
   • no communication of
   intent
   • hard to understand
   • hard to maintain
   • can hardly be used as docs
   

   View Slide

8. O B S C U R E
   T E S T S
   CAUSES
   • too much information (eg:
   Eager test, Irrelevant
   Information)
   • too little information (eg:
   Mystery Guest)
   • lack of attention and care
   

   View Slide

9. O B S C U R E
   T E S T S
   EXAMPLE: EAGER TEST
   • symptom: verifies too much
   functionality in a single test
   • cause: “mocks” a human
   doing a test, chaining lots
   of actions together
   • solution: create multiple,
   single-condition tests for
   better Defect Localisation
   

   View Slide

10. O B S C U R E
    T E S T S
    MYSTERY GUEST
    • symptom: hard to see
    cause-effect relationship
    between input and
    expected outcome
    • cause: tests become harder
    to follow and more fragile
    • solution: setup inline, aided
    by helpers/builders
    

    View Slide

11. View Slide

12. C O N D I T I O N A L
    T E S T L O G I C
    SYMPTOMS
    • presence of control
    structures in tests
    • some test code doesn’t
    get executed
    

    View Slide

13. C O N D I T I O N A L
    T E S T L O G I C
    IMPACT
    • harder to write correct
    tests
    • maintenance overhead
    • decreased confidence in
    tests
    • increased chance of bugs
    in tests
    

    View Slide

14. C O N D I T I O N A L
    T E S T L O G I C
    CAUSES
    • handling failure conditions
    • testing collections
    • testing conditional logic
    • optimising setup/teardown
    logic
    

    View Slide

15. C O N D I T I O N A L
    T E S T L O G I C
    POSSIBLE SOLUTIONS
    • support substitutable deps
    • add guard assertions
    • define test-specific equality
    • enumerate inputs and
    outputs
    

    View Slide

16. C O N D I T I O N A L T E S T L O G I C
    C O D E E X A M P L E
    

    View Slide

17. H A P P Y T E S T I N G !
    

    View Slide

18. T H A N K S !
    @ O M I S S I S
    

    View Slide