Overview

• Preliminaries.
  • Admin.
  • Upcoming Work.
  • Extra Credit.
  • Questions.
• DRY.
• Thinking about SaaS.
• Questions from the readings.
• Rails, revisited.

Preliminaries

Admin

Upcoming Work

• SaaSBook 8 for Friday.
• Saas HW 2 (Sinatra Hangman) for Monday. (It's in "Intro to BDD, TDD, and SaaS" on the MOOC.)
• Reading Journal for Friday (due Thursday at 10pm)
  • Subject: Reading Journal for Class 13 (Your Name)
  • How does Fox and Patterson's presentation of test-driven development compare to your prior experiences with unit testing?
  • Choose the most confusing idea from 8.2 - 8.6. Make a serious attempt to explain this idea in your own words. What questions remain?
  • Choose the most confusing idea from 8.7 - 8.9. Make a serious attempt to explain this idea in your own words. What questions remain?
  • What do you see as the relationship(s) between BDD and TDD? Which do you prefer?
  • What do you see as the most important or enduring concept in this chapter overall?
  • Citation: Questions are based on journal questions from JLND.

Good Things to Do

See http://www.cs.grinnell.edu/~rebelsky/s2d@g/

• Thursday at 11:00 a.m. Convocation on Gates. (Not Gates Tower.)
• Thursday at 4:15 p.m. Thursday Extra.
• Tribute to Ornette Coleman 7:30 p.m. Friday in Herrick. What can be better than a 40 minute Free Jazz piece!

Questions

DRY

• Something we strive for in our code.
• Don't Repeat Yourself
  • That is, Don't Repeat Yourself
• What are ways in which we repeat ourselves as programmers?
  • Write the same exact code twice.
    printf("x has value " + x); vs report(x);
  • Same data represented in multiple places.
  • Identical headers and footers on different Web pages.
  • Similar classes.
  • Similar documentation.
  • Program or write by copy/paste/change. (Repeating similar code)
• There are times it's okay to be repetitious. For example, if all of your functions have a local variable named x, DON'T MAKE IT GLOBAL!
• For identical code (and chunks of code): Make it a separate function.

• What unnecessary repetition appears in the following code?

  for (int i = 0; i < length(lst); i++) 
    {
      printf(lst.get(i));
    } // for
  
  
  for (int i = 0; i < length(lst); i++) 
    {
      printf(i + " of " + length(lst) + ": " + lst.get(i));
    } // for
  
  
  int len = length(lst);
  for (int i = 0; i < len; i++) 
    {
      printf(i + " of " + len + ": " + lst.get(i));
    } // for
  

  • print statement. We avoided repeating overselves by putting it in a loop. Much better printf(lst.get(0)); printf(lst.get(1)); printf(lst.get(2)); ...
  • Is it good to ask to repeatedly ask for the length?
    • Yes: The length might change.
    • No: The body doesn't change the length; we've potentially turned an O(n) algorithm into an O(n^2) algorithm.

• Ways to avoid unnecessary repetition.

  • Write a function to encapsulate identical code.
  • Use a loop!

  • Use a variable to name the result of a repeated function call.

    int min(int[] a, int n) { int result = a[0]; for (int i = 1; i <= n; i++) { if (a[i] < result) result = a[i]; } // for return result; } // min

    int max(int[] a, int n) { int result = a[0]; for (int i = 1; i <= n; i++) { if (a[i] > result) result = a[i]; } // for return result; } // max

  • What was bad about the code above?

    • Lots of extra lines!
    • The first one was buggy. When we fix the bug, we have two things to fix.

  • How do we write it only once? Take a comparison function as a parameter. enum compare = { LARGER, SMALLER } int ultimate(int[] a, int n, compare comp) { int result = a[0]; for (int i = 1; i < n; i++) { switch (compare) { case LARGER: if (a[i] > result) result = a[i]; break; case SMALLER: if (a[i] < result) result = a[i]; break; default: break; } // for return result; } // ultimate int max(int[] a, int n) { return ultimate(a, n, LARGER); } int min(int[] a, int n) { return ultimate(a, n, SMALLER); }

  • Issue: Is it better to repeat the for loop.

    int ultimate(int[] a, int n, comparator comp) { int result = a[0]; for (int i = 1; i <= n; i++) { if (comp(result, a[i]) < 0) result = a[i]; } // for return result; } // min

  • Ruby: Use yield in clever ways.

• Identical headers and footers on different Web pages.
  • Take advantage of your platform's support for that. (Use rails.)
  • Generate your Web pages dynamically using php or shtml or ....
  • Use some tool that supports templating.
  • Write you own tool.

• Note: Make is designed to help you write DRY instructions.

• Similar classes.

  • Use inheritance. Write a common superclass.
• Similar documentation.
  • Write the documentation once and write a short hyperlink to the common documentation.

Thinking about SaaS

Questions from the Readings

An idea I found confusing is the cookies. In my understanding, cookies allow the server to identify returning users. What else am I missing?

How HTTP can be stateless but also use cookies?

I am having trouble understanding the relationship between marshalling and serialization. Serialization is when you turn an object into bits for storage. How does marshalling differ?

What's the relationship between HTTP and URIs?

How do websites scale? It seems like they share the load of computations over many virtual computers, but I'm not sure how all the information is stored and computed cohesively.

What's the relationship between the three-tier architecture and MVC? They seem very similar.

Rails, revisited