https://issues.apache.org/bugzilla/show_bug.cgi?id=52563
--- Comment #5 from Konstantin Kolinko <knst.koli...@gmail.com> 2012-02-05 19:21:58 UTC --- Created attachment 28270 --> https://issues.apache.org/bugzilla/attachment.cgi?id=28270 2012-02-05_tc8_DirContextURLStreamHandler.patch (In reply to comment #4) > The current behaviour is intention. See r287546 So this behaviour was introduced 10 years ago and is well tested. Nobody ever complained about seeing IllegalStateException("Illegal class loader binding") failure enough to change the code. (In reply to comment #0) > Think that the code logic should be something like : > a. Check whether a value is binding on the current thread context classloader, > if does then return. > b. Check whether a value is binding with the current thread, if does then > return. > c. Check the classloader hierarchy to find a binding value. > I am against a.,b.,c., because it changes current behaviour. I would be OK with a.,c.,b.. With a,c,b it will have the same behavour as the current code, but will handle an additional use case without throwing an ISE. -> patch against trunk attached But I would like to see a valid use case for this new feature. Some sample code. Unused features are hard to maintain. > Geronimo has its own classloader creation and > destroying, and it did not has the WebAppLoader thing. Maybe a bug is there? It is hard to understand the whole picture from your short description in Comment 3. -- Configure bugmail: https://issues.apache.org/bugzilla/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are the assignee for the bug. --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org For additional commands, e-mail: dev-h...@tomcat.apache.org
