[zephen]

A study found that, for a particular subset of UB (code that had legal, detectable behavior changes at differing optimization levels), 40% of Debian Wheezy packages exhibited this UB.

https://people.csail.mit.edu/nickolai/papers/wang-stack.pdf

I submit that that's a small fraction of UB, that much of it would exist at any optimization level.

[uecker]

I know, but this still leaves 60% of programs without such UB which is far from "it is practically impossible find a program without UB". Also this this was a study from 2013 and many of those bugs found were fixed. Also GCC got UBSan in 2013 (so after this study).

[pseudohadamard]

That's "UB that was detected in this study". Since gcc will silently break code when it detects UB and you can't tell until you hit that specific case, the 40% is a lower bound. In practice it could be anything up to the full 100%.

[uecker]

In theory. But most C programs do not rely on UB. What is the basis for your claim?

[pseudohadamard]

Uhh... mathematics and logic? Since there's no perfect UB detector, one that detects UB in 40% of programs can only be presenting a lower bound. And I don't know why you think C programs rely on UB, they have it present without the programmer knowing about it.

[uecker]

It follows from mathematics and logic that "larger than 40%" could be 100%, but it does not follow that this is likely or reasonable to assume.