I use C++ every day, but boy it is annoying sometimes. It has little symantec flaws that catch you. You can use numeric_limits to avoid hardcoding a "big number". Given that
numeric_limits<int>::max() is 2147483647
numeric_limits<int>::min() is -2147483648
numeric_limits<double>::max() is 1.79769e+308
What do you suppose numeric_limits<double>::min() would be?
I assumed it would be -1.79769e+308, but no. It is 2.22507e-308, the smallestpositive double value. What the hell? If you want to return that, give it a better name.
Sure, I realize that I can use -numeric_limits<double>::max(), but boy that is confusing and makes template code based on this impossible. Asking around the office, everyone seemed to know the value, and they all
found out the hard way by having a bug in their code.
It ranks up there with using abs and fabs with doubles.
本文揭示了C++中一个容易被忽视的陷阱,即numeric_limits<double>::min()返回的是双精度浮点数的最小正数而非负数,这一特性可能导致代码中的错误和混淆。

被折叠的 条评论
为什么被折叠?



