Remy,
Remy Maucherat wrote:
Jan Luehe wrote:
It's a little risky. We should use the container logger only while the classloader is initialized.
It's probably far from perfect ;)
Then how about RealmBase.authenticate()?
RealmBase.authenticate(String username, String credentials) uses Container.getLogger(), whereas the other RealmBase.authenticate() methods use LogFactory.getLog(). Shouldn't this be consistent? This is where my confusion has stemmed from.
Well, yes, it's a problem. It's simple: as soon as the container is started, all the container logging should go to container.getLogger(), as it used to in Tomcat 4.x.
That was my idea.
If you agree, I can help make the inconsistent cases consistent.
Sure.
In parallel, I want to start a commons component about a j.u.logging implementation keyed per classloader (as seen in bug 33143; it may be expanded a little later to allow per classloader configuration).
I'm looking for committers for the proposal :)
Sounds interesting!
Actually, it's not a very interesting project ;)
The support for j.u.logging is bad at the moment, as there is no (to my knowledge) implementation of j.u.logging which can realistically be used in a server like Tomcat. I hope the core (log manager + rotatable logger) can fit inside a JAR smaller than c-l-api.
Before moving to commons (svn is scaring me ;) ), I think I should start putting the code together in j-t-c/juli (as mentioned, juli means Java Util Logging Implementation). Is that ok ? (and is the name ok ?)
Rémy
--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
