The big thing that isn't talked about is the fact that the problem with forcing a vendor-specific definition on the undefined behaviour of signed integer overfl

The big thing that isn't talked about is the fact that the problem with forcing a vendor-specific definition on the undefined behaviour of signed integer overflow is that significant optimization opportunities are lost. The danger of signed integer overflow in C++ is not that your less-than-zero comparison is not working in a reverse for-loop (although, to be fair, that's a danger encountered by newbies). The big problem is that the compiler will optimize away huge chunks of code when it detects a possible signed integer overflow resulting in weird and unexpected behaviour at runtime. It's not really an exploitable security issue so much as a massive cost issue from stretched out development time or increased support costs (depending on how well your code gets tested before release).