[jboss-jira] [JBoss JIRA] (WFLY-13730) WFLY-9440 fix breaks Galleon provisioning

[Darran Lofthouse (Jira)]

    [ https://issues.redhat.com/browse/WFLY-13730?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14348974#comment-14348974 ] 

Darran Lofthouse commented on WFLY-13730:
-----------------------------------------

TBH [~brian.stansberry] I think with the changes I have coming in we can remove the dependency on the JACC service from the application-security-domain service and instead use the boolan as an indicator JACC is required - maybe log a warning if not available and requested.

The long term plan is to get rid of these application-security-domain-resources entirely.


> WFLY-9440 fix breaks Galleon provisioning
> -----------------------------------------
>
>                 Key: WFLY-13730
>                 URL: https://issues.redhat.com/browse/WFLY-13730
>             Project: WildFly
>          Issue Type: Bug
>          Components: Build System, EJB
>            Reporter: Brian Stansberry
>            Assignee: Brian Stansberry
>            Priority: Major
>
> The WFLY-9440 fix breaks server provisioning in my EE9 branch. This is because the feature-spec Galleon generates for the resource doesn't comprehend that the value of the 'enable-jacc' attribute is not relevant to the required capability so it expects to find a 'org.wildfly.security.jacc-policy.false' capability.  
> I'm not sure why it doesn't break things elsewhere.
> There are a few possible ways to deal with this:
> 1) Don't register BooleanCapabilityReferenceRecorder with the ENABLE_JACC AttributeDefinition; instead manually do its work in the add/remove/write-attribute handlers. This essentially hides the capability relationship from Galleon by removing it from the resource description metadata and seems to work.
> 2) Have BooleanCapabilityReferenceRecorder correctly implement getRequirementPatternSegments to return an empty array. Perhaps this alone would fix it, but I doubt it.
> 3) Combine #2 with a change of some sort in WildFly Core's ReadFeatureDescriptionHandler to better understand and deal with this particular pattern.



--
This message was sent by Atlassian Jira
(v7.13.8#713008)