[
https://issues.apache.org/jira/browse/FELIX-3129?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13113442#comment-13113442
]
Richard S. Hall commented on FELIX-3129:
----------------------------------------
As OBR moves toward the new RFC 112 resolver API, this should become a little
more obvious since the "environment" must also model existing wires. I am
shortly going to try to start working on creating the new RFC 112 resolver
implementation from the Felix framework implementation.
> OBR resolver should not attempt to resolve local resources that are already
> resolved by the Framework
> -----------------------------------------------------------------------------------------------------
>
> Key: FELIX-3129
> URL: https://issues.apache.org/jira/browse/FELIX-3129
> Project: Felix
> Issue Type: Improvement
> Components: Bundle Repository (OBR)
> Reporter: Neil Bartlett
> Labels: obr
>
> As a result of issue FELIX-2465, an interesting issue arose due to a
> difference in resolver behaviour between OBR and the framework itself.
> Scenario:
> 1) From an OBR repository I want to install bundle B, which imports package
> "org.foo"
> 2) Package "org.foo" is already exported by a resolved bundle in the
> framework, let's call it A.
> 3) For some reason OBR thinks that A is not resolvable... even though it is
> actually resolved by the framework. This could happen as a result of bug
> FELIX-2465, or because the framework's resolver is a little different to the
> OBR resolver (possible if running on Equinox or KF).
> In this scenario, OBR will report that bundle B cannot be resolved, due to
> the failed resolution of A.
> OBR should not attempt to resolve local resources that are already in the
> RESOLVED state according to the running framework. I think this could be
> achieved by not generating any Requirements for those resources.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
