[
https://issues.apache.org/jira/browse/FELIX-978?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Richard S. Hall closed FELIX-978.
---------------------------------
Resolution: Fixed
I have committed a patch to address this issue, so I will close it. I have also
added a test case to the BND testing harness in my sandbox to check for this
situation in the future.
> Resolver does not clean up properly on a failed recursive attempt to resolve
> ----------------------------------------------------------------------------
>
> Key: FELIX-978
> URL: https://issues.apache.org/jira/browse/FELIX-978
> Project: Felix
> Issue Type: Bug
> Components: Framework, Specification compliance
> Affects Versions: felix-1.4.1
> Reporter: Richard S. Hall
> Assignee: Richard S. Hall
> Fix For: felix-1.6.0
>
>
> When the resolver is calculating the set of potential candidates for module
> being resolved, it uses a map to store the potential candidates associated
> with each module that needs to be resolved. It also uses this map to detect
> cycles. In the case there are no potential candidates to resolve a
> dependency, the attempt to populate candidates for the given module fails.
> However, the failed module is not removed from the candidates map. The result
> is in certain situations, the resolver may end up wiring to the failed module
> since it still has candidates in the candidate map. This typically can only
> happen if there are multiple dependencies on the failed module with at least
> one of them being optional.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
