我有一个项目,在过去的几年里进行了一点点的改变和修改,从一个代码模块到下一个代码模块,很多都是非标准化的。在某些情况下,我使用 Scripting.Dictionary
对象,而在其他情况下,我使用 Collection
对象。对于其中的每一个,有时会按计数进行迭代(即 For i = 1 to Obj.Count
),有时会按 For...Each
进行迭代。
我想对尽可能多的情况应用相同的逻辑,以使 future 的更改更加无缝,但我不确定哪种方法最好。 (我相信某些特定情况是针对一种或另一种方法进行的,但我也很确定某些代码可以使用任何可用的方法。)
我尝试创建一个测试子来帮助我确定哪种方法效果最好,但结果有些不一致。总的来说,循环遍历 Dictionary
中的每个 Item
似乎更快,但在某些情况下,我的测试结果显示循环遍历每个 Item
在 Collection
中速度更快。差异可能取决于任何给定时间系统中发生的所有其他事情。
我想知道是否有人对哪种方法始终是最快的有明确的答案,假设循环中的所有其他内容都是相同的。或者,是否有某种方法可以改进我的测试子,使其在返回结果时更加一致,以便我可以自己回答这个问题?
我想出的测试代码:
Option Explicit
Sub Test_Dictionary_Iteration_Speed()
Dim Coll1 As New Collection, Coll2 As New Collection
Dim Dict1 As New Scripting.Dictionary, Dict2 As New Scripting.Dictionary, Dict3 As New Scripting.Dictionary
Dim i As Integer, j As Integer, l As Integer
Dim StartTime As Single, StopTime As Single
Dim v As Variant
Dim Obj As TestObject 'A custom Class that has only one member variable, MainVal, and no functions/subs
For i = 0 To 32766
Set Obj = New TestObject
Obj.MainVal = i
Dict1.Add CStr(i), Obj
Dict2.Add CStr(i), Obj
Dict3.Add CStr(i), Obj
Coll1.Add Obj, CStr(i)
Coll2.Add Obj, CStr(i)
Next i
StartTime = Timer()
For j = 0 To Dict1.Count - 1
l = CInt(Dict1(CStr(j)).MainVal)
Set Obj = Dict1(CStr(l)) 'Do something useful within the loop
Set Obj = Nothing
Dict1.Remove CStr(l)
Next j
StopTime = Timer()
Debug.Print "Dict1 for x to y: " & StopTime - StartTime
StartTime = Timer()
For Each v In Dict2.Items
l = CInt(v.MainVal)
Set Obj = Dict2(CStr(l))
Set Obj = Nothing
Dict2.Remove CStr(l)
Next v
StopTime = Timer()
Debug.Print "Dict2 for each item: " & StopTime - StartTime
StartTime = Timer()
For Each v In Dict3.Keys
l = CInt(Dict3(v).MainVal)
Set Obj = Dict3(CStr(l))
Set Obj = Nothing
Dict3.Remove CStr(l)
Next v
StopTime = Timer()
Debug.Print "Dict3 for each key: " & StopTime - StartTime
'---------- Division between Dictionary and Collection
StartTime = Timer()
For j = 0 To Coll1.Count - 1
l = CInt(Coll1(CStr(j)).MainVal)
Set Obj = Coll1(CStr(l))
Set Obj = Nothing
Coll1.Remove CStr(l)
Next j
StopTime = Timer()
Debug.Print "Coll1 for x to y: " & StopTime - StartTime
StartTime = Timer()
For Each v In Coll2
l = CInt(v.MainVal)
Set Obj = Coll2(CStr(l))
Set Obj = Nothing
Coll2.Remove CStr(l)
Next v
StopTime = Timer()
Debug.Print "Coll2 for each item: " & StopTime - StartTime
Debug.Print vbNewLine & "-----" & vbNewLine
End Sub
真实的输出示例,表明“最佳”选项并不总是相同:
Dict1 for x to y: 0.2011719
Dict2 for each item: 0.1738281
Dict3 for each key: 0.2167969
Coll1 for x to y: 0.2050781
Coll2 for each item: 0.1386719
Dict1 for x to y: 0.1875
Dict2 for each item: 0.171875
Dict3 for each key: 0.234375
Coll1 for x to y: 0.2050781
Coll2 for each item: 0.1542969
Dict1 for x to y: 0.25
Dict2 for each item: 0.21875
Dict3 for each key: 0.265625
Coll1 for x to y: 0.234375
Coll2 for each item: 0.171875
Dict1 for x to y: 0.265625
Dict2 for each item: 0.203125
Dict3 for each key: 0.296875
Coll1 for x to y: 0.234375
Coll2 for each item: 0.21875
Dict1 for x to y: 0.265625
Dict2 for each item: 0.1875
Dict3 for each key: 0.234375
Coll1 for x to y: 0.203125
Coll2 for each item: 0.15625
Dict1 for x to y: 0.28125
Dict2 for each item: 0.1875
Dict3 for each key: 0.25
Coll1 for x to y: 0.234375
Coll2 for each item: 0.1875
Dict1 for x to y: 0.28125
Dict2 for each item: 0.21875
Dict3 for each key: 0.328125
Coll1 for x to y: 0.234375
Coll2 for each item: 0.234375
最佳答案
除非您确实知道给定语句或过程的执行时间是一个问题,否则您不应该在优化上浪费人力。首先设计和调试,然后如果您认为事情太慢(它们可能不会),分析,然后才优化(执行时间通常会浪费在与您想象的完全不同的地方)。
For Each
结构很好、简洁、整洁。不使用它的唯一原因是如果您要从循环的集合中删除项目。那么你就有可能跳过某些项目。如果您打算删除项目,请向后循环索引。
关于performance - 迭代 Scripting.Dictionary/Collection 对象,我们在Stack Overflow上找到一个类似的问题: https://stackoverflow.com/questions/11122074/