It's also much easier to fix when you know where the overrun occurs compared to getting an abort on a random access later.