Comments+on+Morse+Exercise

I'm putting some comments on submissions here, as they apply to more than one student/project. We CAN learn from others' mistakes.

The notion of code a little, test a little... then run more tests is hard for people to adopt.

I REALLY want to see test runs on hand-checkable samples. E.g. run your algorithm on samples such example strings as "e", "t", "h", "et", "e t", "no punctuation", "(no)???? punctuation!..." and see that results are not surprising.

I do expect a write-up. It can include screen shots, but it should also include observations or at least a story of what you did.

NB: The very problem statement for Binary Tree Exercise suggests several preliminary steps. Don't be bashful about just doing those in sequence, and pasting your results into a text file.

The problem statement specifically said to make it data-driven. I've seen a submission with 3 huge switch statements. I understand the argument that behind the curtains, the compiled code may have data structures which implement the switch, but I want YOU to understand the data structures.

Sample Code Comparison Output includes Rik's results for the Gettysburg Address. If your results for Morse and Lex are very different, you might want to investigate why. One project did not count any spaces. A series of simple tests would disclose that serious bug.

Parameterize: the assignment says to run on at least 2 texts, at least 3 codes. That should require about ONE procedure which takes as arguments the string to be estimated and the code to use as estimator. One project had THREE copies of the Gettysburg address, each built in to procedure which applied one code.

One pretty good project has three DIFFERENT structs which are shaped just the same.... instead of 3 instances of the same struct. It's as if they purposely wanted to prevent generality.

Not all projects state the relative efficiency. That's the "moral of the story" result, the reason for doing the measurement.