According to
https://glassfish.dev.java.net/issues/show_bug.cgi?id=2375 there hasn't been much progress...
In response to Tim's question on the bug - yes, the app was certainly undeployed when the output was captured.
This issue still happens for our whole development team. After 15-20 redeployments we have to restart the application server (and every redeploy gets slower until the app server finally throws a permgen error).
I also don't think testing this with one a test/tutorial ejb implementation is particularly useful. I think a real-world application that has dependencies on a lot of third party jar files would be more meaningful.
Admittedly it is more important that the app server run reliably without memory problems in the final deployment, but this is a really frustrating problem!
Cheers
Cameron
[Message sent by forum member 'cameronr' (cameronr)]
http://forums.java.net/jive/thread.jspa?messageID=225662