[
https://issues.apache.org/jira/browse/FELIX-6161?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16885870#comment-16885870
]
David Jencks commented on FELIX-6161:
-------------------------------------
Our comments crossed... I think that FELIX-4964 results in 2 events when
references match different ObjectClasses from a single service. I wonder if
there's a way to avoid responding to a 2nd or later event for the same service
change, perhaps by checking whether the existing reference is currently the
best? This might allow exposing the actual filters to the framework, as Arnold
wants, and having only one component cycling in even more cases than at present.
> SCR: Method of resolving references limits Service ListenerHook
> implementations
> -------------------------------------------------------------------------------
>
> Key: FELIX-6161
> URL: https://issues.apache.org/jira/browse/FELIX-6161
> Project: Felix
> Issue Type: Bug
> Components: Declarative Services (SCR)
> Affects Versions: scr-2.1.16
> Reporter: Arnoud Glimmerveen
> Priority: Major
> Attachments: FELIX-6161.patch
>
>
> When experimenting with creating a Service ListenerHook to realise a
> [providing a service on demand
> pattern|https://osgi.org/specification/osgi.core/7.0.0/framework.servicehooks.html#d0e45721],
> I noticed that references from components managed by Felix SCR were not
> showing up as expected: When a reference declares a target filter, that
> target filter is not included in the ListenerInfo provided to the
> ListenerHook. As a consequence the ListenerHook is unable to create a
> matching service.
> Source of the observed behaviour is how SCR tracks references, specifically
> that it opens a ServiceListener for just the className and performs the
> matching of the target filter inside the ServiceListener. This approach
> unfortunately hides the target filter from the Service ListenerHook and thus
> the Service ListenerHook will have insufficient information to create a
> matching service.
> Also look at the discussion on the dev mailing list:
> [https://www.mail-archive.com/dev@felix.apache.org/msg48657.html]
--
This message was sent by Atlassian JIRA
(v7.6.14#76016)
