当我们在用LoggerFactory.getLogger(xyz.class)时我们在干什么

我们在使用logback一类的logging system的时候

总是在代码的前面部分放上这段代码private final Logger log = LoggerFactory.getLogger(AdController.class

这段代码调用了org.slf4j.LoggerFactory line:280

1. public static Logger getLogger(Class clazz) {
2. return getLogger(clazz.getName());
3. }

1. public static Logger getLogger(String name) {
2. ILoggerFactory iLoggerFactory = getILoggerFactory();
3. return iLoggerFactory.getLogger(name);
4. }

进入 ch.qos.logback.classic.LoggerContext  line:105

 

   Logger childLogger = (Logger) loggerCache.get(name);

loggerCache定义处

private Map<String, Logger> loggerCache;

初始化代码

1. public LoggerContext() {
2. super();
3. this.loggerCache = new ConcurrentHashMap<String, Logger>();
5. this.loggerContextRemoteView = new LoggerContextVO(this);
6. this.root = new Logger(Logger.ROOT_LOGGER_NAME, null, this);
7. this.root.setLevel(Level.DEBUG);
8. loggerCache.put(Logger.ROOT_LOGGER_NAME, root);
9. initEvaluatorMap();
10. size = 1;
11. this.frameworkPackages = new ArrayList<String>();
12. }

可以看到logger里通过name获取logger的实际上是访问了loggerCache这个支持高并发的hashmap.

以下50行代码 做了最重要的工作,请大家仔细阅读,就会明白了.其实很简单.

1. public final Logger getLogger(final String name) {
3. if (name == null) {
4. throw new IllegalArgumentException("name argument cannot be null");
5. }
7. // if we are asking for the root logger, then let us return it without
8. // wasting time
9. if (Logger.ROOT_LOGGER_NAME.equalsIgnoreCase(name)) {
10. return root;
11. }
13. int i = 0;
14. Logger logger = root;
16. // check if the desired logger exists, if it does, return it
17. // without further ado.
18. Logger childLogger = (Logger) loggerCache.get(name);
19. // if we have the child, then let us return it without wasting time
20. if (childLogger != null) {
21. return childLogger;
22. }
24. // if the desired logger does not exist, them create all the loggers
25. // in between as well (if they don't already exist)
26. String childName;
27. while (true) {
28. int h = LoggerNameUtil.getSeparatorIndexOf(name, i);
29. if (h == -1) {
30. childName = name;
31. } else {
32. childName = name.substring(0, h);
33. }
34. // move i left of the last point
35. i = h + 1;
36. synchronized (logger) {
37. childLogger = logger.getChildByName(childName);
38. if (childLogger == null) {
39. childLogger = logger.createChildByName(childName);
40. loggerCache.put(childName, childLogger);
41. incSize();
42. }
43. }
44. logger = childLogger;
45. if (h == -1) {
46. return childLogger;
47. }
48. }
49. }

如果获取不了,就进入一个循环遍历

比如 我们给的name是com.colamachine.calendar.CalendarService

那么他会第一次会去找com 如果没有com就创建com的logger

再去寻找com.colamachine 同上创建

再去找并创建com.colamachine.calendar

再去找并创建com.colamachine.calendar.CalendarService.

好看完了代码,这只是我一厢情愿的猜测.现在来开始

果然如此 

这样的机制 在内存的开销上是否过大

对于只有6个模块的小系统,轻轻松松的mapsize达到了532个