StringBuilder.append長string時出問題。

今天PROD遇到問題了，查詢大量數據，而後返回XML， 在StringBuilder.append時發生錯誤，並且發生錯誤是在append一個小String時發生的錯誤。若是內存設置也很大，而且出示capacity的邏輯也比較麻煩，那麼就要從設計上下功夫了，好比看看爲何在這一短期訪問量這麼大，能不能均衡一下。String在expandCapacity會進行一下currentsize*2 + 2,是否大於要追加的string的長度，若是小於就擴展當前長度+要追加的String的長度。

ArrayList使用currentsize*3/2 + 1和追加的數據長度來比較

java.lang.OutOfMemoryError: Java heap space
at java.util.Arrays.copyOf(Arrays.java:2882)
at java.lang.AbstractStringBuilder.expandCapacity(AbstractStringBuilder.java:100)
at java.lang.AbstractStringBuilder.append(AbstractStringBuilder.java:390)
at java.lang.StringBuilder.append(StringBuilder.java:119)

A:

http://www.captaincasademo.com/forum/posts/list/1503.page

Hi, 

today we were forced to dive into the StringBuffer class - after looking into a OutOfMemory problem in our client. 

It's "amazing" that the following code causes an OutOfMemory - event though the virtual machine is started with -Xmx256m: 

Code:

public static void main(String[] args)
     {
         try
         {
             char[] chars = new char[30000000];
             for (int i=0; i<chars.length; i++) chars[i] = 'A';
             String s = new String(chars);
             StringBuffer sb = new StringBuffer();
             System.out.println("Now appending s");
             sb.append(s);
             System.out.println("Now appending Hallo");
             sb.append("Hallo!"); 
             // ----
             System.out.println("Finished!");
         }
         catch (Throwable t)
         {
             t.printStackTrace();
         }
     }
 
 
 OUTPUT:
 Now appending s
 Now appending Hallo
 java.lang.OutOfMemoryError: Java heap space
 	at java.util.Arrays.copyOf(Arrays.java:2882)
 	at java.lang.AbstractStringBuilder.expandCapacity(AbstractStringBuilder.java:100)
 	at java.lang.AbstractStringBuilder.append(AbstractStringBuilder.java:390)
 	at java.lang.StringBuffer.append(StringBuffer.java:224)
 	at org.eclnt.client.ztest.TestMem.main(TestMem.java:22)

A String with the length of 30.000.000 characters is added to a stringbuffer - that's it. It's even more astonishing that the error is NOT thrown when appending the "big" string, but when appending "Hallo". 

The reason is the internal processing of StirngBuffer. StringBuffer keeps everyhing in an char-array, and there is a certain method that increases the char-array-size when finding out that the array does not fit anymore. The method increases the size by at least "2 * currentSize + 2". This may make sense for small sizes, but it's completely a wrong decision for big sizes. In our case the appending of "Hello" means that the total array size is increased by 60.000.002 as minimum, so that the minimum size is 90.000.002 afterwards. And now, the previous char array with length of 30.000.000 needs to be copied into the one of 90.000.002 - this means both arrays co-exist for a short duration of time, resulting in 120.000.002 characters being in memory (and of course there is some additional memory blocked by the array object). So the appending of "Hallo" means a lot... - enough to cause out of memory. 

When concatenating long strings, then it is as consequence essential to carefully work with StringBuffers. We eliminated our OutOfMemory problems by using the following utitlity method: 

Code:

public static String concatenateStrings(List<String> items)
     {
         if (items == null)
             return null;
         if (items.size() == 0)
             return "";
         int expectedSize = 0;
         for (String item: items)
             expectedSize += item.length();
         StringBuffer result = new StringBuffer(expectedSize);
         for (String item: items)
             result.append(item);
         return result.toString();
     }

When now changing the test code to: 
Code:

public static void main(String[] args)
     {
         try
         {
             char[] chars = new char[30000000];
             for (int i=0; i<chars.length; i++) chars[i] = 'A';
             String s = new String(chars);
             List<String> buffer = new ArrayList<String>();
             System.out.println("Now appending s");
             buffer.add(s);
             System.out.println("Now appending Hallo");
             buffer.add("Hallo!");
             String all = concatenateStrings(buffer);
             // ----
             System.out.println("Finished!");
         }
         catch (Throwable t)
         {
             t.printStackTrace();
         }

...then no error occurs anymore. 

you can set an inital capacity for StringBuffer i guess. 

Also i would recommend to use StringBuilder instead of StringBuffer (in case you don't need a garanteed synchronisation). 

this test (link) resulted in a 34% better perfomance for StringBuilder. 

http://littletutorials.com/2008/07/16/stringbuffer-vs-stringbuilder-performance-comparison/

Hi, 

you are 100% right...! 
The setting of an internal capacity is by the way the "trick" behind the concatenateStrings-method int the text above. Before appending things in the StringBuffer, its size is calculated (variable "expectedSize"). 

Regards, Björn