Last time, we learned about the dangers of uninitialized floating point variables but left with a puzzle: Why wasn't this caught during internal testing?
I dropped a hint when I described how s work: You have to ask the processor to raise an exception when it encounters a signaling NaN, and the program disabled that exception. Why was an ...