为什么C#TryParse不能处理NaN？

Question

在读取一些旧的Fortran数据时,我有一些常用的方法可以使用默认值覆盖进行解析.有时数据NaN (Not a number)应该有数值.我预计TryParse会看到字符串"NaN"并且无法解析.但是,TryParse成功解析并放入NaN数值.这是预期的行为吗？如果这是预期的,我还应该寻找其他任何"陷阱"值吗？

    public static double GetDoubleFromString(string s, double defaultOnFailure = 0)
    {
        //Must be here due to TryParse returning NaN to d and result is set to true below
        if (s.Contains("NaN"))
        {
            log.Warn(string.Format("String contained NaN, returning defaultOnFailure {0} string was {1}", defaultOnFailure, s));
            return defaultOnFailure;
        }

        var d = defaultOnFailure;
        if (!double.TryParse(s.Trim(), out d))
        {
            log.Warn(string.Format("Failed to parse double from string returning defaultOnFailure {0} string was {1}", defaultOnFailure, s));
        }

        return d;
    }

UPDATE

我觉得应该提到,这只发生在double,long和int不返回NaN值.请参阅下面的示例代码,Common ....代码只是格式化的Console.WriteLine或停止执行控制台.请参阅下面的截图输出.

    public static void NanTestMain()
    {
        Common.WriteBlankWithTitle("Test parse NaN");
        string s = "NaN";

        Common.WriteBlankWithTitle("NaN to Int");
        int i;
        var intSuccess = int.TryParse(s, out i);
        Console.WriteLine(string.Format("Int parse of {0} parse = {1}", i, intSuccess));

        Common.WriteBlankWithTitle("NaN to Double");
        double d;
        var doubleSuccess = double.TryParse(s, out d);
        Console.WriteLine(string.Format("Double parse of {0} parse = {1}", d, doubleSuccess));

        Common.WriteBlankWithTitle("NaN to Long");
        long l;
        var longSuccess = long.TryParse(s, out l);
        Console.WriteLine(string.Format("Long parse of {0} parse = {1}", l, longSuccess));

        Common.Pause();
    }

Answer 1

来自MSDN:

该小号参数可以包含NumberFormatInfo.PositiveInfinitySymbol,NumberFormatInfo.NegativeInfinitySymbol,或NumberFormatInfo.NaNSymbol用于通过提供商指示的培养.

有三个"特殊"值需要注意.然而,最后几个字是关键 - 取决于当前的文化,你可能会看到除了以外的东西"NaN"!