[quectophoton]

C standard: "Undefined behavior means such a situation can't happen."

Me: "If it can't happen then it would be fine to just crash on those situations, right? Because such a crash would never be reached. Can we get that?"

C compilers: "No. Would you want to crash on signed integer overflow, for example?"

Me: "Yes? Would be safer than the current situation at least."

C compilers: "What, no, that would make your programs imperceptibly slower. Would you even like that?"

Me: "Yes, I'll be able to live with that."

C compilers: "Well, the answer is still no."

[nullc]

It would have taken you less time to look up -fwrapv / -ftrapv / -fsanitize=signed-integer-overflow + -fsanitize-undefined-trap-on-error than write out that misleading dialog. :)

[quectophoton]

Fair.

Last time I checked ubsan was a few years ago for a small program, and I've read stuff about it not being suitable for anything outside development (more like a valgrind complement than a release option) because of how it messed with environment variables and linked libraries or something, so I might not have the full picture.

[GoblinSlayer]

C is more likely to define wrapping arithmetic. Wrapping is defined as undefined for compatibility with MIPS processors that support checked arithmetic.