C++中NaN异常值捕捉方法

问题

debug代码发现某个值偶发变为-NaN需要定位具体位置。

尝试

基本可以判断是除0引起，想当然尝试了val < -10000的条件断点，无法捕捉。

解决

查阅文献

NaN is designed to propagate through all calculations, infecting them like a virus, so if somewhere in your deep, complex calculations you hit upon a NaN, you don't bubble out a seemingly sensible answer.

因此根据NaN的特性，一句简单的断言即可命中：

assert(val == val);

参考

floating point - Why is NaN not equal to NaN? - Stack Overflow