为什么会出现那么多比较String和StringBuffer的文章?
原因在于当改变字符串内容时,采用StringBuffer能获得更好的性能。既然是为了获得更好的性能,那么采用StringBuffer能够获得最好的性能吗?
答案是NO!
为什么?
如果你读过《Think in Java》,而且对里面描述HashTable和HashMap区别的那部分章节比较熟悉的话,你一定也明白了原因所在。对,就是支持线程同步保证线程安全而导致性能下降的问题。HashTable是线程安全的,很多方法都是synchronized方法,而HashMap不是线程安全的,但其在单线程程序中的性能比HashTable要高。StringBuffer和StringBuilder类的区别也在于此,新引入的StringBuilder类不是线程安全的,但其在单线程中的性能比StringBuffer高。如果你对此不太相信,可以试试下面的例子:
package com.hct.test;
import java.util.ArrayList;
import java.util.Iterator;
import java.util.List;
/**
* @author: chengtai.he
* @created:2009-12-9 上午09:59:57
*/
public class StringBuilderTester {
private static final String base = " base string. ";
private static final int count = 2000000;
public static void stringTest() {
long begin, end;
begin = System.currentTimeMillis();
String test = new String(base);
for (int i = 0; i < count/100; i++) {
test = test + " add ";
}
end = System.currentTimeMillis();
System.out.println((end - begin)
+ " millis has elapsed when used String. ");
}
public static void stringBufferTest() {
long begin, end;
begin = System.currentTimeMillis();
StringBuffer test = new StringBuffer(base);
for (int i = 0; i < count; i++) {
test = test.append(" add ");
}
end = System.currentTimeMillis();
System.out.println((end - begin)
+ " millis has elapsed when used StringBuffer. ");
}
public static void stringBuilderTest() {
long begin, end;
begin = System.currentTimeMillis();
StringBuilder test = new StringBuilder(base);
for (int i = 0; i < count; i++) {
test = test.append(" add ");
}
end = System.currentTimeMillis();
System.out.println((end - begin)
+ " millis has elapsed when used StringBuilder. ");
}
public static String appendItemsToStringBuiler(List list) {
StringBuilder b = new StringBuilder();
for (Iterator i = list.iterator(); i.hasNext();) {
b.append(i.next()).append(" ");
}
return b.toString();
}
public static void addToStringBuilder() {
List list = new ArrayList();
list.add(" I ");
list.add(" play ");
list.add(" Bourgeois ");
list.add(" guitars ");
list.add(" and ");
list.add(" Huber ");
list.add(" banjos ");
System.out.println(StringBuilderTester.appendItemsToStirngBuffer(list));
}
public static String appendItemsToStirngBuffer(List list) {
StringBuffer b = new StringBuffer();
for (Iterator i = list.iterator(); i.hasNext();) {
b.append(i.next()).append(" ");
}
return b.toString();
}
public static void addToStringBuffer() {
List list = new ArrayList();
list.add(" I ");
list.add(" play ");
list.add(" Bourgeois ");
list.add(" guitars ");
list.add(" and ");
list.add(" Huber ");
list.add(" banjos ");
System.out.println(StringBuilderTester.appendItemsToStirngBuffer(list));
}
public static void main(String[] args) {
stringTest();
stringBufferTest();
stringBuilderTest();
addToStringBuffer();
addToStringBuilder();
}
}
上面的程序结果如下:
5266 millis has elapsed when used String.
375 millis has elapsed when used StringBuffer.
281 millis has elapsed when used StringBuilder.
I play Bourgeois guitars and Huber banjos
I play Bourgeois guitars and Huber banjos
从上面的结果来看,这三个类在单线程程序中的性能差别一目了然,采用String对象时,即使运行次数仅是采用其他对象的1/100,其执行时间仍然比其他对象高出25倍以上;而采用StringBuffer对象和采用StringBuilder对象的差别也比较明显,前者是后者的1.5倍左右。由此可见,如果我们的程序是在单线程下运行,或者是不必考虑到线程同步问题,我们应该优先使用StringBuilder类;当然,如果要保证线程安全,自然非StringBuffer莫属了。
除了对多线程的支持不一样外,这两个类的使用几乎没有任何差别,上面的例子就是个很好的说明。appendItemsToStringBuiler和appendItemsToStirngBuffer两个方法除了采用的对象分别为StringBuilder和StringBuffer外,其他完全相同,而效果也完全相同。
参考资料:http://www.cnblogs.com/Fskjb/archive/2010/04/19/1715176.html