CS 204: Software Design

Due 5:00PM Monday, 8 June 2009

This is not a takehome exam, so you may speak with each other about it. However, please do the work without a partner.

A miscellany

Processing XML files

A lot of the data on computers these days is stored in XML form. XML is many things, but at its simplest level, it is a system of "tags" (<tagname>...</tagname>) that may be nested around data (<a>data in a <b>data in b in a</b>more data in a</a>) and may include "attributes" (like the href attribute in <a href="http://carleton.edu">Carleton College</a>). If you search your computer for "*.xml" files, you'll find tons of them. Your iTunes library, for example, or left-over RSS feed files or various inscrutable system files.

Python comes standard with a nice XML parsing library. Here is a sample program that just counts the opening tags in an XML file.

For this assignment, I want you to extend my sample program (or rewrite it completely, if you prefer) to perform the following tasks on XML files.

• Report how many times each distinct tag occurs in the file, sorted either alphabetically or by count.
• If the input file fails to parse, report the failure with as much helpful detail as you can. (Line numbers, nature of the failure, etc.)
• Report the deepest nesting of tags within the file.
• For each tag, list the attributes that appear with the tag (not including their values, which could be extremely numerous).
• Come up with two more handy features for an XML file utility program, and implement them.

Profiling

Profilers are programs that help you determine which portions of you programs are taking how long to run. If you write a program that doesn't run fast enough, and you've done everything you can think of to squeeze out better performance, it's time to pull out the profiler and see if it can point you to trouble spots.

Python comes standard with a profiler module called cProfile.

For this part of your assignment, I want you to try to improve the performance of some Python code by using cProfile. One possible challenge would be to improve the performance of John Zelle's graphics.py library that we use in CS 111. For example, the graphics lab from my current CS 111 section included an animation of a bouncing Circle, a polygon drawer, etc. These all depend on Zelle's library. Could they be made more efficient by modifying the library?

Another possible target would be your own RSS filter from earlier this term. The network access may be the slowest part of the program, but you could use cProfile to profile your filtration code to see if there are ways to speed it up.

Whatever program you choose to profile and improve, here's what I want you to hand in:

• A readme briefly explaining what program you're profiling.
• The "before" source code of the program you're profiling.
• The "after" source code.
• An analysis of your work, including profiling data from the original code, a discussion of what you changed and why, and profiling data from the (hopefully) sped-up code. This analysis can go in the readme file or a separate file, as you wish.

Hungarian notation

Six months after graduating in 1998, one of our CS majors who went off to become a programmer wrote to me raving about a paper that I needed to read and then start teaching to my students. The paper, Hungarian Notation by Charles Simonyi of Microsoft, describes Simonyi's naming system for identifiers (variables and constants, in particular).

Read Simonyi's paper, and then answer the following questions.

• Why is the system called Hungarian?
• Summarize the benefits of the Hungarian naming system.
• Summarize the disadvantages of the Hungarian naming system.
• Using the Hungarian notation style, write a static Java method that takes a string and an integer as parameters, and returns a string consisting of the last N characters of the original string, where N is the integer passed in as a parameter. (If the original string has no more than N characters, return the original string. If N < 0, return the empty string.) The point of this exercise is to use the Hungarian notation (changing, for example, the name of N to something more in line with Simonyi's approach). So if you can make the function body essentially trivial by using Java String methods, go right ahead.
• Do you think we should encourage our students to use the Hungarian system? Why or why not?