This outline is also available in PDF.
Held: Thursday, April 15, 2010
Summary:
We consider some basic approaches to the design of classes. We also
visit the current status of the project.
Related Pages:
Notes:
- Reading for Tuesday: Chapter 9.
- Goal for next Thursday: Nine hours of work per person.
- I left out one of the types of problems for the final: There will also be a
write a test
problem.
- As you requested, I've removed the unused parts of the course web. Let me know if that seems better.
Overview:
- Exploring Logs.
- Design Principles.
- Alternatives to Inheritance.
- Small-group Exercise.
- Project Reflection.
- Integration.
Get out your logs and let's talk.
- What were the primary tasks you spent your time on?
- How much time did you spend on each?
- How many classes did you write?
- Simple classes
- Complex classes
- How many lines of code did you write?
- And how should we measure lines of code?
Principle 1: The Open-Closed Principle (OCP)
- What did you write for the
Sharpen Your Pencil
question on p. 381?
- What are reasons you might want to violate this principle (e.g., change the
code in your base class)
Principle 2: The Don't Repeat Yourself (DRY)
- Sam considers this one of the most important aspects of refactoring.
- Sam considers this one of the key reasons we have higher-order procedures.
- However, in the object model, we often use objects to encapsulate the
repetition.
- How does the descripton of DRY on p. 385 differ from the model in XP?
Princple 3: The Single Responsibility Principle (SRP)
- The
do I do this to myself or do others do it to me
question
seems to be the core idea.
- I'm not sure that I completely buy their model.
Principle 4: The Liskov Substitution Principle (LSP)
- Yay! Type theory intrudes on OO design
- Who is Liskov?
- Inheritance isn't the solution to every problem (duh).
- What are some other techniques for sharing behavior
- Delegation: An object contains another object, and delegates the
behavior to that object
- Composition: Delegation plus inheritance (you can vary the behavior
by varying what subclass you include)
- Aggregation: Composition with independence
- Note: I consider the "is there one reference or more than one reference
to this object" a relatively minor matter.
- I believe that it stems from a time in which you had to be careful in
thinking about what to dispose of when you dispose of
an object
- In most modern OO systems, the garbage collector handles the issue
- Exercise: Draw a UML diagram for the five-minute mystery on p. 413.
- A number of questions about the past few readings have asked about the
balance between Head-First
Generalize it All in Advance
and more
XP-ish Let the design emerge from the code
.
- I can only give you my own perspective.
- I have traditionally been a generalize person.
- I believe that generalizations often emerge, so you might as well predict them.
- I find generalizations beautiful.
- I find generalizations intellectually challenging.
- I know that a library of general functions helps me in the long term.
- As time has passed, I've found that I see the value in a more XP-like approach.
- For most of us, the time taken to generalize is not particularly helpful to the task at hand.
- For many of us, the attempt to generalize can lead to anxiety.
- So, I encourage you to generalize when it's obvious and to use more
XP-style programming when generalizations are not obvious.
- But you must refactor mercilessly.
- Three groups of four people. Each group needs to have a member of
each of teams 1, 2, and 3 (or is that 0, 1, and 2)?
- Spend fifteen minutes discussing what you've accomplished.
- Do not let any member/team dominate!
- If you were forced to combine, what would you take from where, and why?
- I've heard some comments that you are sharing code between groups
between the weekly integration meetings.
- My goal had been that you work independently between meetings.
- What would you prefer?
- If you're going to share more regularly, then we should probably
divide the project differently (e.g., a UI group, a Network group,
and a Model group)
- I know that you had questions about the LocalWorld vs. NetworkedWorld
models. We should go over my model for that and your alternatives.
- It's probably time to work out the network protocol.
;
