Comment by pizlonator

I don't see a contradiction between the statement that vulns in C++ code are generally on a downward trajectory, and the statement that most vulns are in new code.