Question

I think the common idiom for creating instances of java.util.logging.Logger is this:

public class SomeClassName {

    private static final Logger LOG = Logger.getLogger(SomeClassName.class.getName());

}

My IDE will manage changing the line appropriately when I refactor my code (change the name of the class, for example). It still bugs me that I have to repeat the name of the class, though. What I’d really like to do is something like Logger.getLogger(getName()) or Logger.getLogger(class.getName()), but this isn’t legal Java in a static initilization.

Is there a better way of getting at a logger that doesn’t involve repeating myself?

Answer 1

Issue 137 of The Java Specialists’ Newsletter deals with this problem. It recommends applying a logger factory, which can detect the actual class name e.g. by generating an exception and analysing the call stack.

I personally find this worse than the original problem, but this is just my 2 cents. At any rate, technically it is interesting, so here it is:

public class LoggerFactory {
  public static Logger make() {
    Throwable t = new Throwable();
    StackTraceElement directCaller = t.getStackTrace()[1];
    return Logger.getLogger(directCaller.getClassName());
  }
}

...

public class BetterApplication {
  private final static Logger logger = LoggerFactory.make();

  ...
}