Class 31: The Costs of Reliability

Held Wednesday, March 15, 2000

Overview

Today we conclude our discussions of reliability by visiting topics that seem to be of particular interest.

Question 31 for today's class: How do we balance the need for formal specifications required by program proof techniques and the complexity of modern graphical applications?

Question 32 for Monday's class: What feature do you most want to see in our PacMan game? Sketch a strategy we might use to implement that feature.

Notes

• The college's Usenet news service has stopped providing news. Mr. Stone has now identified an alternative Usenet supplier, Supernews (http://www.supernews.com/), which has agreed to a free trial of its Usenet service, starting now and continuing through the coming weekend. To connect, point your news reader at corp.supernet.com (207.126.101.100). Please let Mr. Stone know what you think.
• Some of you have asked how you can move beyond a B in the class. My general rule of thumb is that A indicates exceptional, and doing all the work correctly is not (or should not be) exceptional. You need to show something special to move towards the A range.
  • It would also be nice if you would more carefully check your grammar and spelling. Check out SamR's writing bugaboos for some writing habits that I find particularly painful.
• Did anyone hear the news about the ``software bug'' in the Mars lander?

Contents

Summary

• Can we prove all programs correct?
• How do we balance needs?
• What else can we do to make programs more reliable?

This discussion digressed into issues of AI, which we will return to after break.