[
https://issues.apache.org/jira/browse/BSF-21?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13195760#comment-13195760
]
Rony G. Flatscher commented on BSF-21:
--------------------------------------
Checked in the fix on 2012-01-29, probably my mistake that the issue was
mistakingly declared to be resolved on 2012-01-15.
> AdapterClassLoader: does not attempt to use current Thread's context class
> loader, if ...
> -----------------------------------------------------------------------------------------
>
> Key: BSF-21
> URL: https://issues.apache.org/jira/browse/BSF-21
> Project: Commons BSF
> Issue Type: Bug
> Components: BSF-2.x
> Affects Versions: BSF-2.4
> Reporter: Rony G. Flatscher
> Assignee: Rony G. Flatscher
> Fix For: BSF-2.5
>
>
> When creating dynamically an event adapter in
> org.apache.bsf.util.event.generator.EventAdapterGenerator, an instance of the
> caching
> org.apache.bsf.util.event.generator.AdapterClassLoader is created and saved
> in a static variable.
> If a listener class has been supplied to EventAdapterGenerator and the class
> has been constructed dynamically, then AdapterClassLoader is used to
> define/load the class.
> Unfortunately, if OpenOffice.org is dispatching macros via BSF which need to
> have BSF create a new event adapter, then this fails with OpenOffice.org,
> starting with their version 2.3 (introduced fall 2007). The reason being that
> the OpenOffice.org defining class loader does not load any OpenOffice
> classes!
> Rather it becomes necessary with OpenOffice to set the current thread's
> context class loader to one of OpenOffice which resolves the OpenOffice
> classes. For BSF to be able to support this behaviour, it becomes necessary
> to try out the current thread's context class loader as well, otherwise the
> OpenOffice event type cannot be found when defining the dynamically class,
> causing an *error* (not an exception) which will kill the thread by default.
> A possible solution in the AdapterClassLoader for "defineClass" and
> "getLoadedClass" is to also attempt to use the thread's context class loader.
> This can be done by creating an inner class loader class for which instances
> a parent class loader is set to the thread's class loader.
> Have a local working version of this scheme, which I would apply after
> working through the current issues (next week, as I will be able to plan one,
> two days for tackling and resolving the reported BSF issues).
> ---rony
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
