No problem.

I don't really want getUser() to have its own logging

am concerned about is the end user experience. If I am

I want to see DEBUG for the login process, I shouldn't have to set DEBUG levels on more than one category. I should set DEBUG for "org.widgets.login" and see all of the debug available for the login process. Plus, getUser() would display DEBUG messages every time logout() is called which is not what I wanted either.

Now, as the server developer, I could implement getUser() like this:

void getUser( Logger logger ) { }

and just pass the logger in every time I call it. That would accomplish the same thing. But, I don't really want to

the code that calls getUser() and pass in a Logger.

So, to me, the cleanest solution would be to set a Logger for the current thread. Then getUser() can just grab the "current" logger and use that. It doesn't care because it is only part of a larger process anyway. This is an implementation detail but, you would also need a default category (maybe the root logger) to log messages to in case the caller hasn't set a logger context. Either way, getUser() relies on someone else to tell it what its logger should be.

Tony

To unsubscribe, e-mail: log4...@logging.apache.org For additional commands, e-mail: log4...@logging.apache.org