#15 on the list is especially troubling to me, since it seems like something that could be trapped, yet gcc compiled without comment, even with the vaunted -Wall switch. I couldn't figure out a way to get gcc to complain, nor did my extensive 10 minute search find anything that would trap this sort of error. I even tried valgrind with it, and it too didn't seem to think anything was wrong (but that was only after spending another 10 minutes).
(On an aside, there is a switch in gcc called "-fbounds-check", but that only works for Fortran and Java. Go figure).