On Mon, Jul 7, 2008 at 7:17 PM, Niclas Hedhman <[EMAIL PROTECTED]> wrote: > On Tue, Jul 8, 2008 at 12:22 AM, Stuart McCulloch > <[EMAIL PROTECTED]> wrote: > >> again, this wouldn't solve the Pax-Logging issue because >> then the application will have all it's logging calls bound to >> the original Log4J jar, which lacks the OSGi support that >> Pax-Logging provides (various other things will then fail) > > Correct. I suspect that the Log4J classes from the WebAppClassloader > is "leaking" and what causes the problems. So, either it is in Craig's > configuration "somewhere" that the Framework should expose those > packages from the parent CL, or we are touching on a bug in Felix.
That is what triggered me as well. I did look into it a little bit but no other explanation as of now. Is it somehow possible that log4j is doing some classloader magic itself and maybe is using the context classloader? The context classloader would be the WebAppClassloader I think ... regards, Karl >> I believe Niclas has some ideas on how to proceed... > > I have received the whole project from Craig, and will try to look at > it later when I am back "home"... Stay tuned. > > > Cheers > Niclas > -- Karl Pauls [EMAIL PROTECTED]
