I suppose Tomcat shouldn't probably do an indiscriminate flushCaches().
It would probably have to selectively clear only the cache entries for
classes loaded by the webapp classloader?

In the end though, while this fixed by contrived web app I used in the
testcases, it still doesn't fix my real web-app.  I'm onto the next
library (hibernate) that seems to be pinning the webapp classloader in
memory.  I'm begining to think that this is a lost cause.  All it takes
is one static reference to the classloader from any code in the web app
and the leak will occur.  I wish the classloader class had a kill()
method that would take care of the problem ;-)

Thanks for your feedback.

-Jay


> It could.  I was considering that when I attached the info 
> you posted to the relevant Bugzilla issues yesterday.  But 
> it's not trivial, because Tomcat might be using cached 
> BeanInfos from the Introspector (directly or indirectly).  
> The flushCaches would have to be done in the webapp's 
> classloader of course.  So because it's not trivial, and I 
> didn't have time to think further, I punted on the issue for 
> yesterday ;)
> 
> Yoav

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to