Line matching is imprecise for method bodies with duplicated lines

[pauloborba]

The current implementation could lead to errors in method bodies that contain repeated lines as in

x=x+1;
y=m();
x=x+1;

We should carefully assess (how frequent are such errors? can we discard such scenarios from the analysis?) and document that, as it might impact the study results. We should do the same for any imprecision that might impact the study results.

I wonder if we could do better by, for example, recording the line contents and, for lines that appear more than once, the order in which they appear. For the example above, we would record (x=x+1,1) for the first line, (y=m(),0) for the second line, and (x=x+1,2) for the third line. And maybe this extra information could be used to eliminate or reduce imprecision.

Modified line lists should not be Set.