[
https://issues.apache.org/jira/browse/ARIES-976?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Jeremy Hughes reopened ARIES-976:
---------------------------------
The fix in ARIES-847 is ok, but if another unchecked exception comes out of the
cleanup we swallow it. I'll add debug so at least we can see what's happening.
> BlueprintContainerImpl swallowing NPE prevents error being logged
> -----------------------------------------------------------------
>
> Key: ARIES-976
> URL: https://issues.apache.org/jira/browse/ARIES-976
> Project: Aries
> Issue Type: Bug
> Components: Blueprint
> Reporter: Jeremy Hughes
> Assignee: Jeremy Hughes
>
> I'm seeing a timing issue in the case where a blueprint bundle (the client)
> has a service reference to a service in a second bundle (the service). But
> the service bean is for a class that can't be found. When the service bundle
> is processed first I get a helpful stacktrace of ClassNotFoundException. In
> the case where the client bundle is processed first, the GRACE_PERIOD event
> is sent then the service bundle is processed. After the
> ClassNotFoundException is thrown, the service bundle blueprint container is
> cleaned up - in the catch (Throwable t) at the end of
> BlueprintContainerImpl.doRun(). Eventually the BlueprintRepository.destroy()
> is called which gets a null back from BlueprintRepository.getInstance().
> BlueprintRepository.destroy() passes that to BundleRecipe.destroy() which
> casts it and NPEs. The NPE gets swallowed by the
> BlueprintContainerImpl.doRun() and I never see the error message holding the
> original ClassNotFoundException.
> I'll commit a fix later.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
