java - 我应该使用 Iterator 还是 for 循环来迭代？

Question

我不喜欢必须调用 hasNext() 和 next() 的想法，并且看看我如何需要一个循环来调用它们，我并没有真正看到使用列表和映射中提供给我们的迭代器有什么意义。

只有当您需要从列表中提取一个时才需要吗？或者，如果您要从列表中取出一项，并且必须花费无法衡量的时间来完成该操作？我在 Google 或 Stack 上找不到此内容，请帮助我。

我并不是专门讨论增强的 for 循环(for-each 循环)。

我也理解 foreach 循环在性能方面表现出色，但这更多的是一个“为什么它存在”的问题

编辑:意识到我只是在谈论集合，而不是数组。 Foreach 循环在这种情况下没有限制。

Answer 1

一个foreach相当于一个迭代器——它是同一件事的语法糖。所以你应该总是选择foreach尽可能使用迭代器，只是因为它很方便并且可以生成更简洁的代码。

我在另一个答案中写过这一点: https://stackoverflow.com/questions/22110482/uses-and-syntax-for-for-each-loop-in-java/22110517#22110517

正如 @RonnyShapiro 所说，在某些情况下您需要使用迭代器，但在许多情况下 foreach应该足够了。请注意 foreach是不正常的for环形。正常for当需要访问索引时需要循环。尽管您可以使用 foreach 手动创建单独的索引 int 变量这并不理想，来自 variable-scope观点。

这里有一些更多信息: Which is more efficient, a for-each loop, or an iterator?

访问集合时，foreach是 significantly faster比基本for循环的数组访问。然而，当访问数组时(至少对于原始数组和包装数组来说)，通过索引访问速度方式更快。见下文。

---

访问 int 时，索引比迭代器快 23-40%或Integer数组。以下是以下测试类的输出，它将 100 个元素的原始 int 数组中的数字相加(A 是迭代器，B 是索引):

[C:\java_code\]java TimeIteratorVsIndexIntArray 1000000
Test A: 358,597,622 nanoseconds
Test B: 269,167,681 nanoseconds
B faster by 89,429,941 nanoseconds (24.438799231635727% faster)

[C:\java_code\]java TimeIteratorVsIndexIntArray 1000000
Test A: 377,461,823 nanoseconds
Test B: 278,694,271 nanoseconds
B faster by 98,767,552 nanoseconds (25.666236154695838% faster)

[C:\java_code\]java TimeIteratorVsIndexIntArray 1000000
Test A: 288,953,495 nanoseconds
Test B: 207,050,523 nanoseconds
B faster by 81,902,972 nanoseconds (27.844689860906513% faster)

[C:\java_code\]java TimeIteratorVsIndexIntArray 1000000
Test A: 375,373,765 nanoseconds
Test B: 283,813,875 nanoseconds
B faster by 91,559,890 nanoseconds (23.891659337194227% faster)

[C:\java_code\]java TimeIteratorVsIndexIntArray 1000000
Test A: 375,790,818 nanoseconds
Test B: 220,770,915 nanoseconds
B faster by 155,019,903 nanoseconds (40.75164734599769% faster)

[C:\java_code\]java TimeIteratorVsIndexIntArray 1000000
Test A: 326,373,762 nanoseconds
Test B: 202,555,566 nanoseconds
B faster by 123,818,196 nanoseconds (37.437545972215744% faster)

完整的测试类:

   import  java.text.NumberFormat;
   import  java.util.Locale;
/**
   <P>{@code java TimeIteratorVsIndexIntArray 1000000}</P>

   @see  <CODE><A HREF="https://stackoverflow.com/questions/180158/how-do-i-time-a-methods-execution-in-java">https://stackoverflow.com/questions/180158/how-do-i-time-a-methods-execution-in-java</A></CODE>
 **/
public class TimeIteratorVsIndexIntArray  {
   public static final NumberFormat nf = NumberFormat.getNumberInstance(Locale.US);
   public static final void main(String[] tryCount_inParamIdx0)  {
      int testCount;
      //Get try-count from command-line parameter
         try  {
            testCount = Integer.parseInt(tryCount_inParamIdx0[0]);
         }  catch(ArrayIndexOutOfBoundsException | NumberFormatException x)  {
            throw  new IllegalArgumentException("Missing or invalid command line parameter: The number of testCount for each test. " + x);
         }

      //Test proper...START
         int[] intArray = new int[] {1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100};

         long lStart = System.nanoTime();
            for(int i = 0; i < testCount; i++)  {
               testIterator(intArray);
            }
         long lADuration = outputGetNanoDuration("A", lStart);

         lStart = System.nanoTime();
            for(int i = 0; i < testCount; i++)  {
               testFor(intArray);
            }
         long lBDuration = outputGetNanoDuration("B", lStart);

         outputGetABTestNanoDifference(lADuration, lBDuration, "A", "B");
   }
      private static final void testIterator(int[] int_array)  {
         int total = 0;
         for(int i = 0; i < int_array.length; i++)  {
            total += int_array[i];
         }
      }
      private static final void testFor(int[] int_array)  {
         int total = 0;
         for(int i : int_array)  {
            total += i;
         }
      }
      //Test proper...END

//Timer testing utilities...START
   public static final long outputGetNanoDuration(String s_testName, long l_nanoStart)  {
      long lDuration = System.nanoTime() - l_nanoStart;
      System.out.println("Test " + s_testName + ": " + nf.format(lDuration) + " nanoseconds");
      return  lDuration;
   }

   public static final long outputGetABTestNanoDifference(long l_aDuration, long l_bDuration, String s_aTestName, String s_bTestName)  {
      long lDiff = -1;
      double dPct = -1.0;
      String sFaster = null;
      if(l_aDuration > l_bDuration)  {
         lDiff = l_aDuration - l_bDuration;
         dPct = 100.00 - (l_bDuration * 100.0 / l_aDuration + 0.5);
         sFaster = "B";
      }  else  {
         lDiff = l_bDuration - l_aDuration;
         dPct = 100.00 - (l_aDuration * 100.0 / l_bDuration + 0.5);
         sFaster = "A";
      }
      System.out.println(sFaster + " faster by " + nf.format(lDiff) + " nanoseconds (" + dPct + "% faster)");
      return  lDiff;
   }
//Timer testing utilities...END
}

我还运行了这个 Integer数组和索引仍然是明显的赢家，但速度只快了 18% 到 25%。

对于List的Integers但是，迭代器速度更快。只需将上面代码中的int-array更改为

List<Integer> intList = Arrays.asList(new Integer[] {1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100});

并对测试函数进行必要的更改( int[] 到 List<Integer> 、 length 到 size() 等)

[C:\java_code\]java TimeIteratorVsIndexIntegerList 1000000
Test A: 3,429,929,976 nanoseconds
Test B: 5,262,782,488 nanoseconds
A faster by 1,832,852,512 nanoseconds (34.326681820485675% faster)

[C:\java_code\]java TimeIteratorVsIndexIntegerList 1000000
Test A: 2,907,391,427 nanoseconds
Test B: 3,957,718,459 nanoseconds
A faster by 1,050,327,032 nanoseconds (26.038700083921256% faster)

[C:\java_code\]java TimeIteratorVsIndexIntegerList 1000000
Test A: 2,566,004,688 nanoseconds
Test B: 4,221,746,521 nanoseconds
A faster by 1,655,741,833 nanoseconds (38.71935684115413% faster)

[C:\java_code\]java TimeIteratorVsIndexIntegerList 1000000
Test A: 2,770,945,276 nanoseconds
Test B: 3,829,077,158 nanoseconds
A faster by 1,058,131,882 nanoseconds (27.134122749113843% faster)

[C:\java_code\]java TimeIteratorVsIndexIntegerList 1000000
Test A: 3,467,474,055 nanoseconds
Test B: 5,183,149,104 nanoseconds
A faster by 1,715,675,049 nanoseconds (32.60101667104192% faster)

[C:\java_code\]java TimeIteratorVsIndexIntList 1000000
Test A: 3,439,983,933 nanoseconds
Test B: 3,509,530,312 nanoseconds
A faster by 69,546,379 nanoseconds (1.4816434912159906% faster)

[C:\java_code\]java TimeIteratorVsIndexIntList 1000000
Test A: 3,451,101,466 nanoseconds
Test B: 5,057,979,210 nanoseconds
A faster by 1,606,877,744 nanoseconds (31.269164666060377% faster)

在一项测试中，它们几乎是等效的，但迭代器仍然获胜。