[ https://issues.apache.org/jira/browse/GERONIMO-4470?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12656865#action_12656865 ]
Kevan Miller commented on GERONIMO-4470: ---------------------------------------- I noticed the behavior while looking at the code. It seemed inconsistent to me. So, created a Jira to track. for <hidden-classes> <filter>foo</filter></hidden-classes>, we will *never* search a parent for a foo.* classes. for <non-overridable> <filter>bar</filter></non-overridable>, we won't load a bar.* classes, if we find the class in a parent. However, if we don't find a bar.LocalOnly in a parent, we'll also try to load from the local ClassLoader. This seems inconsistent to me. And as you note, it's unlikely to behave properly... I would expect the local ClassLoader to never be searched, in this case. Prolly a relatively minor issue. We can argue semantics... Perhaps you have an argument for why the current behavior is a good thing? > non-overridable filters working properly > ---------------------------------------- > > Key: GERONIMO-4470 > URL: https://issues.apache.org/jira/browse/GERONIMO-4470 > Project: Geronimo > Issue Type: Bug > Security Level: public(Regular issues) > Affects Versions: 2.0.4, 2.1.4, 2.2 > Reporter: Kevan Miller > Fix For: 2.2 > > > If we're unable to load a non-overridable class from a parent classloader and > inverse classloading is configured, looks like we'll try to load the class > from the local ClassLoader. I don't think this is correct. If we're unable to > load from a parent classloader, should always return a > ClassNotFoundException... -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.