[
https://issues.apache.org/jira/browse/OAK-1592?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13944792#comment-13944792
]
Michael Dürig commented on OAK-1592:
------------------------------------
AFAICS this patch "reflects" the session's permission into the
{{ObservationManager}} by passing it the respective {{PermissionProvider}}. I
think this is a good thing generally. However I'm a bit worried about threading
issues. Are {{PermissionProvider}} implementations thread safe and if so, will
sharing them across threads not result in lock contentions?
> Performance of Session#hasPermission
> ------------------------------------
>
> Key: OAK-1592
> URL: https://issues.apache.org/jira/browse/OAK-1592
> Project: Jackrabbit Oak
> Issue Type: Improvement
> Components: jcr
> Reporter: angela
> Assignee: angela
> Fix For: 0.20
>
> Attachments: ConcurrentHasPermissionTest_20140321_150212.csv,
> ConcurrentHasPermissionTest_20140321_174520.csv, OAK-1592.patch
>
>
> the benchmarks i added recently for Session#hasPermission show a poor
> performance of Session#hasPermission compared to regular permission
> evaluation and jr2.
> i suspect that the most prominent reason for this is the fact that the
> AccessManager#hasPermission *always* refreshes the underlying
> permissionprovider even if the session has neither been modified nor
> refreshed.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
