Working Isn't Good Enough

Lecture Notes for CS 190
Winter 2018
John Ousterhout

• Reading: Chapter 3 of book

• Programs evolve continuously:
  • Can't design the whole system at once
  • Can't get the design right the first time
  • Requirements change

• Tactical Programming
  • Goal is to get the next feature or bug fix working
  • A few shortcuts and kludges are OK, if it gets things done quickly
  • Results in bad design, high complexity
  • Tactical tornadoes
• Working isn't good enough
• Strategic Programming
  • Primary goal is to produce a great design
    • How easy is it to evolve this code?
  • Continual small investments

• How much to invest?
  • As much as you can afford
  • 10-20% overhead?
  • Investments pay for themselves relatively quickly (6-12 months?)
• For this class: zero tolerance for complexity