Java Reference
In-Depth Information
it to 0 and then setting it to 0 again at the end of the outer repetend— is just not
good programming practice, as we will see in a moment.
The student's programming strategy
When asked about the absence of comments, the student replied that he had
not gotten around to those yet.
My way of programming is to write the program first, and later to
fill in statement-comments and loop invariants and things.
He said this in spite of all our discussions of good programming practices and in
spite of all our examples of top-down design. And his practices led to an error
that took him a great deal of wasted time to find and to a bad design. In fact, pro-
viding comments after the fact is a waste of time.
When pressed for the invariant of the outer loop and for a statement-com-
ment for the repetend of the outer loop, after some time and with help, the stu-
dent said that the invariant was:
invariant: pairs (i, 0..C-1) with 0≤i<r have been processed
And along with this came the statement-comment for the repetend:
Process pairs (r, 0), …, (r, C - 1)
so that the segment could be interpreted as:
// Process pairs (r, c) , for 0≤r<R , 0≤r<C
int r= 0; int c= 0;
// inv: pairs (0..r-1, 0..C-1) have been processed
while (r != R) {
Process pairs (r, 0) , …, (r, C - 1)
r= r + 1; c= 0;
// Process pairs (r,c) , for 0≤r<R , 0≤r<C
int r = 0;
// invariant: pairs (0..r-1, 0..C-1) have been processed
while (r != R) {
// Process pairs (r, 0) , , (r, C - 1)
int c= 0;
// invariant: pairs (r, 0) , …, (r , c-1) have been processed
while (c != C) {
Process pair (r , c); c= c + 1;
r= r + 1;
Figure 7.8:
A well-designed segment to process pairs
Search WWH ::

Custom Search