Comments from the markers of Assignment 2
==========================================================================

This is a collection of the comments made by the 3 TAs who marked your
assignment 2.  Pay particular attention to the general remarks that would 
apply equally well to upcoming assignments.


Program:
	It seems that this was the part of the assignment that students
	did best on -- simply making the program run.  

Style:
	Most students have not grasped the idea of organizing a large task
   	into subprograms. Most of them had a single long (3-4 page)
   	Run procedure.

Testing:		
	Testing was universally very poor. The
	majority of students did not annotate anything
	and gave no explanations - just printouts. The
	necessity for a testing strategy cannot be 
	overemphasized.

	Testing was simply the worst part in most assignments. Lack of
   	annotated test runs, a test plan, etc. A comprehensive test plan
   	would have earned most students good marks in testing, even if 
	their program didn't work.

	Some students did their testing more like experiments. They showed
	different performance results rather than showing that each part 
	of the program really works.

Experiments:	  
	This was badly done. Many of the answers
	showed little or no logical coherence. The quality
	of the writing was generally poor. Two contradictory
	statements would often follow each other on the same
	line. 

	A lot of students simply gave tons of
   	experimental runs, and literally translated the results into
   	English. They did not try to understand the patterns in the 
	output, and think what was the underlying reason.

	Some of the students do not really understand the purpose of
	a simulation.  They do not appreciate the interdependence of the
	parameters.  Some made very precise conclusions based on
	very random data.

Part C: Correctness:
	This section was somewhat better. 

	The invariants were a very badly done.
	Almost half of the students did not understand what
   	"data invariant" meant. It was not clear to them that a "data
   	invariant" was just a set of conditions, like "loop invariants".
   	They are not very clear about loop invariants either.

	For question 4, their proofs
	all somehow make sense, but many are not logically valid or
	coherent. They presented some facts, but did not show how these 
	facts help in the proof.  Maybe they think we can figure that 
	out. :)

Quality of Writing:	
	It seems many people did not take this very seriously. 
	Spelling and grammatical mistakes abounded here. 

	More importantly, some students had a very difficult time
	making logical, coherent arguments (for example, regarding
	their experiments).  This is an essential skill in computer
	science.