Rules of Thumb

• From physics: Pushing against a wall is exerting energy, not doing work.
  • When a developer complains that he has worked so hard, but the task is just not progressing - keep the above rule in mind. He needs to be guided to the proper selection of a test and its test data.
  • For test direction, have him write down the problem that he wants to solve. You'll probably have to help him do this, especially in the beginning. If he could do this, the problem would almost be solved already (slight exaggeration). Next, write down how the test data will decide one way or the other about the problem.
  • For speed, always create a specific test case. Don't run with all the data! It takes too long and the results you care about are hidden within a larger mass of output. The time taken to create the test case will pay for itself in the speed in which the

program results can be examined.

• If you cut up a pound of cheese into small bites, everyone will starve.
  • There is a inverse relation between size of modules and number of interfaces within the system. If you make the modules too small, each programmer will spend too much time on interfacing and too little time on solving meaningful problems.
• Design for requirements, not exceptions.
  • The overwhelming amount of work in your system will be to satisfy conditions that meet the requirements specified. In the high-level design, the issues in the designer's mind must be how to create a rock-solid solution. Exceptions can always be handled by adding code to a properly functioning program.
• The mind can only hold 5 to 10 things at a time.
  • The complexity of any level of a module should