Teaching
670 S '05
 
Lectures
Lecture 1
Lecture 2
Lecture 3
Lecture 4
Lecture 5
Lecture 6
Lecture 7
Lecture 8
Lecture 9
Lecture 10
Lecture 11
Lecture 12
Lecture 13

Lecture 3: Uses Cases Again; Working in Pairs

Use Cases Revisited

Study this simple story, tease out the concepts, sketch a graph, write use cases:

A Silly Game: 21

The players meet at a table and roll a dice. Every player rolls the dice once per round. They add up their scores. A player can skip a turn or declare that he is done. A player is eliminated if/when his score surpasses 21. The game is over when all players are eliminated or done.

Scoring: The winners are those players with the maximum score.

Your customer would like to sell this game on-line. He would like to run one automated player per game.


Compare your use cases with the graphs of other people. Map the concepts in your use cases to those of the graph. Are any concepts missing in the graph? Are relations missing? Are your use cases overlooking concepts from the graph?


Working in Pairs

Working in pairs means working together. It does not working separately and combining the results of the separate work into one product.

At any moment, one person is the driver and the other one is the navigator. The terms are borrowed from driving cars; you may also find pilot and co-pilot. Furthermore, partners may switch roles while they work together on a task, though they don't have to.

Some pairs designate drivers for entire tasks. Sometimes they just switch during tasks. It depends on the emphasis and the organization of the surrounding team.

Examples of how you should work together in this class:

  • When you analyze a problem statement and tease out the first list of concepts, your pair work is probably a form of brain storming. In this case, it is probably acceptable if you work on an equal footing, without specific roles. If you have difficulties working together, assign roles.
  • When you write up use cases, designated one person per case as the writer and the editor. For other cases, switch the roles. Separate the writing and the editing phases, but go through at least two cycles of writing and editing per case.
  • When you program, you may wish to dedicate a class, a module, or a sub as owned by one person, which means the person is the driver for this piece of code. Naturally the other person is the navigator. A different but also effective allocation of work is based on use cases.

Perceived Problems of Pair Programming

Pair programming comes with several perceived and true disadvantages. Williams and Kessler (see end) call them "myths". Here are the ones that you need to know about both for this course and for working in industry later:
  1. The most obvious objection is that pair programming reduces the work force in half. Anecdotes from industry and controlled experiments show, however, that this is not the case. Pairs typically complete their tasks faster than solo programmers, and their products exhibit fewer defects than those of solo workers.
  2. The second most frequent objection is that people need time and space of their own to work. This is true, and nobody says that people work with partners all the time. Instead, people continue to think about their work after meetings and come back to the next working session with additional thoughts. The critical point is that people produce artifacts together and correct each other's work frequently in joint working sessions.
  3. A similar object is that the partners have dissimilar backgrounds. In a course context, this just means that one of them teaches the other person and thus benefits from deepening his knowledge. The other person picks up knowledge and skills. In an industrial context, knowledge transfer happens and the company benefits from dispersing knowledge among the members of its team.
  4. An important objection is that it becomes incredibly difficult to evaluate individuals and to pay them according to their contributions. In a course context, this problem can be overcome in many ways. In the industrial world, this is a true problem without an obvious solution.

Sometimes people report that a partner discovers nothing but syntax errors during programming sessions. This is an easily overcome problem with pair programming. Give your driver time to find mistakes on his own. Instead focus on the larger picture. Don't look for the pot holes on the road; look for rail road crossings, forks, and other markers on the road map. Once you figure out this part your collaboration is likely to be smooth.

Teams

Many companies use programming teams. Typically a programming team consists of a chief programmer and his co-pilot; a tools person who knows the chosen language, its libraries, and the chosen IDE really well; the archive person who manages the code archive and the version control system; the programmers who take on specific components; and sometimes other support staff, including testers, technical writers, and customer representatives (literal or virtual).

The chief programmer and his co-pilot tend to design components and their interfaces. The rest of the team implements these components. At some point, the chief programmer and his co-pilot begin the integration work. This last step is often full of surprises and requires deep and broad changes to the product. Don't ever believe that integration is a one-shot activity.

In this course, we will not practice team programming. Your projects are pieces that are usually assigned to solo programmers in a team. Do not distribute the work and integrate it later. Integrating causes such a huge overhead that you are likely to fail. Work in pairs.

Bibliography

  • Williams and Kessler. Pair Programming Illuminated. Addison-Wesley, 2003.

  • last updated on Tue Jun 9 22:03:19 EDT 2009generated with PLT Scheme