Revision as of 10:07, 24 November 2010

WARNING: Spoiler!

Try criticising the Frogs design BEFORE reading this.

Criticisms of the Frog Design (2009)

• Riel's heuristics
  • 2.1 Hide data within its class: The AdultFrog class exposes an array of eggs it has laid.
  • 2.9 Keep related data and behavior in one place: Frog and FrogBrain are separate classes.
  • 3.5 Interface should be dependent on model: The Move interface contains a method called display; the model is dependent on the interface
  • 3.9 Avoid verb classes: The Move interface is a verb. It could be called Movable or similar (although that name is still not great for other reasons).
  • 3.10 Agent classes irrelevant: The Biologist class is an agent.
  • 5.14 Avoid becomes: A frog will at various stages be an egg, tadpole and adult frog.
    • Solution: State or strategy pattern
  • 5.1 Inheritance for specialization: A toad is not a frog; it's inheritance for implementation.
  • 5.17 Avoid no-op overrides: The hop and swim methods of Egg do nothing.
  • 5.12 Beware type switches: Tadpole has a type attribute which is used in swim() (5.13 broken as well?)

• The biologist should have the saute method, not the frog itself.
• For a better solution to the XML export ability, see a froggy visitor (last year's class)

See also

• Frogs design