Question 4 on "So you think you know C"

[fluffy-critter]

Question 4 on "So you think you know C"

[akalenuk]

Please don't take the whole thing too seriously. It's just a funny test with funny results.

There is no such thing as standard NaN integers though, and this is a problem in itself. All the range of an integer type is numbers so you have to employ additional mechanisms to register overflows or shift errors. Those are not covered by the C standard at all (although I'm not familiar with C23 to be honest).

[fluffy-critter]

Yeah, I'm not taking it seriously and I also agree that "don't know" is the correct answer.

I was mostly just wondering if you knew of a situation that the UB would actually manifest as a return value other than 1. The explanation given for why it's UB is that the overflow of the int16 is not specified, but there's still no possible bit pattern stored in the int16 that would result in a different result. I suppose there could be something especially zany that an optimizer does though.