Recently I listened to quite an old Miško Hevery presentation about refactoring conditional sentences with the help of polimorphism and inheritance for better testability.

In essence, this refactoring is done because of:

  • conditional sentences reduce readability
  • it is harded to test out code plaqued with conditionals
  • it is eaer to maintain OOP based systems

The presentation uses empirical evidence such as layered conditionals, also known as Pyramid of Doom. Some of the branches (elses) are not meant to be used, testing such behemoths becomes a nuisance and maintenance requires changes in several locations.

Groo’s StackOverflow answer of why conditional sentences should not be used for every occasion, gives good insights as well. It states that modularity increases readability and maintainability by avoiding clustering everything into a singlue unit.

Indeed, conditional sentences are inevitable when primitive types are compared. Miško Hevery summarizes when should conditionals sentences be refactored:

  • if the way object behaves depends on its state
  • if condition has to be checked several times in multiple code locations

In conclusion, this presentation helped me understand when polimorphism should be used and how to tackle complex applications.