为什么是 Round(Fix(x))? [英] Why Round(Fix(x))?

问题描述

在 Microsoft.VisualBasic 的 Reflector 摘录中，它引用了 Microsoft.VisualBasic.Conversion.Fix 在 3 个地方不包括 Fix(Object) 重载.

In the Reflector extract for Microsoft.VisualBasic it references Microsoft.VisualBasic.Conversion.Fix in 3 places not including the Fix(Object) overload.

在每种情况下都适用 Math.Round 到结果.(特别是在 DateAdd 和 DateDiff 中；Choose 中的第三次使用确实减去 1，并冗余地转换为 Round 之前再次 >Double.)

In each case it applies Math.Round to the result. (Especially in DateAdd and DateDiff; the third use in Choose does subtract 1, and redundantly cast to Double again before applying Round.)

什么时候可以 Math.Round(Conversion.Fix(x)) <>双x的Conversion.Fix(x)?

When can Math.Round(Conversion.Fix(x)) <> Conversion.Fix(x) for Double x?

(我会自己检查参考源，但我找不到可以下载的下载em> 包括 Microsoft.VisualBasic.) 参考源现在可用 在线.

推荐答案

然后将所有这些引用转换为 Integer 或 Long:

All of these references are then cast to Integer or Long:

CInt 和 CLng 在它们对应的 IL conv.ovf.i4/8 之前显式调用 Math.Round从 Single 和 Double 转换时转换.

CInt and CLng explicitly call Math.Round before their corresponding IL conv.ovf.i4/8 cast when casting from Single and Double.

这会强制执行 Banker's Rounding，这是 VB.NET 已知的怪癖".

This enforces the Banker's Rounding that is a VB.NET known "quirk".

conv.ovf.i4 单独向零截断，这恰好与 Fix 相同(对于适合 Integer 或 conv.ovf.i8 的长).

conv.ovf.i4 alone truncates towards zero, which happens to be the same as Fix (for the numbers that fit in an Integer, or a Long for conv.ovf.i8).

_{(当我最初输入问题时，我以为我已经检查过这不是原因.)}

有一段时间，Microsoft 提供了参考源，因此我能够确认它们都已附上在 CInt 或 CLng 中，Math.Round 由编译器添加.

For a while, Microsoft made the Reference Source available, so I was able to confirm they were all enclosed in CInt or CLng and the Math.Round is added by the compiler.

目前在上面链接中可用的 VB.NET源代码"现在只提供没有代码的参考程序集":-(

The VB.NET "source" currently available at the above link now only provides the "reference assembly" which has no code :-(

这篇关于为什么是 Round(Fix(x))?的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！