[
https://issues.apache.org/jira/browse/SLING-5371?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15053551#comment-15053551
]
Marc Pfaff commented on SLING-5371:
-----------------------------------
Sorry, I was not aware of the plain JCR API constraint. Yes, this approach uses
the Jackrabbit API to register the listener. The benefit is that it allows to
register one listener on multiple paths (installer root paths) in this case,
and this could save us some Oak observation queues that need to be processed.
I will give it a try with a plain JCR API version only, which would end in one
move listener registered per root path. But AFAIU that's still more efficient
than one move listener on the root node.
> Improving JcrInstaller's move listener registration
> ---------------------------------------------------
>
> Key: SLING-5371
> URL: https://issues.apache.org/jira/browse/SLING-5371
> Project: Sling
> Issue Type: Improvement
> Components: Installer
> Reporter: Marc Pfaff
> Attachments: SLING_5371.patch
>
>
> The JcrInstaller is currently registering a JCR listener for move events on
> the root node with setting the 'deep' property to true. If it comes to
> processing the underlaying Oak observation queue, this combination is very
> inefficient, as it does not allow Oak to filter changes by path.
> AFAIU this move listener is covering the use case where an installer artifact
> is moved from/to an installer root folder. I therefore wanted to propose to
> register the move listener on the configured root folders only.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
