[
https://issues.apache.org/jira/browse/OAK-11827?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18011127#comment-18011127
]
Joerg Hoh commented on OAK-11827:
---------------------------------
I discussed that topic with [~angela] offline; the bottom line is that the
current implementation operates on structures, which already have permissions
checks enforced for properties. But this API call {{isNodeType}} has to return
the correct values, independent of having read access to the jcr:mixin property
or not.
For that the implementation should be changed, so that it can directly access
the raw NodeState. In that case {{canReadMixinTypes()}} is not required any
more.
> speedup NodeImpl.isNodeType() for negative checks
> -------------------------------------------------
>
> Key: OAK-11827
> URL: https://issues.apache.org/jira/browse/OAK-11827
> Project: Jackrabbit Oak
> Issue Type: Task
> Components: jcr
> Affects Versions: 1.84.0
> Reporter: Joerg Hoh
> Priority: Major
>
> We have code which iterates recursively down a Node tree, and it decides to
> iterate a specific child and its children by checking its nodetype. If that
> childnode does *not* have that specific nodetype, it's iterated.
> When checking such a situation, we have frequently stacktraces like this:
> {noformat}
> at
> org.apache.jackrabbit.oak.jcr.security.AccessManager.hasPermissions(AccessManager.java:54)
> at
> org.apache.jackrabbit.oak.jcr.session.NodeImpl.canReadMixinTypes(NodeImpl.java:1346)
> at
> org.apache.jackrabbit.oak.jcr.session.NodeImpl.getMixinTypeNames(NodeImpl.java:1328)
> at
> org.apache.jackrabbit.oak.jcr.session.NodeImpl$24.lambda$perform$0(NodeImpl.java:958)
> at
> org.apache.jackrabbit.oak.jcr.session.NodeImpl$24$$Lambda$1777/0x0000001801d10440.get(Unknown
> Source)
> at
> org.apache.jackrabbit.oak.plugins.nodetype.ReadOnlyNodeTypeManager.isNodeType(ReadOnlyNodeTypeManager.java:297)
> at
> org.apache.jackrabbit.oak.jcr.session.NodeImpl$24.perform(NodeImpl.java:958)
> at
> org.apache.jackrabbit.oak.jcr.session.NodeImpl$24.perform(NodeImpl.java:953)
> at
> org.apache.jackrabbit.oak.jcr.delegate.SessionDelegate.perform(SessionDelegate.java:230)
> at
> org.apache.jackrabbit.oak.jcr.session.ItemImpl.perform(ItemImpl.java:113)
> at
> org.apache.jackrabbit.oak.jcr.session.NodeImpl.isNodeType(NodeImpl.java:953)
> {noformat}
> That means that despite the optimizations done in OAK-10639, we need to
> consistently resolve all the mixins to be sure to return "false".
> (It is not possible to rewrite that code to perform a positive check, which
> would benefit then from the optimization in OAK-10639.)
> I see at least 1 way to optimize this, because NodeImpl.canReadMixinTypes
> performs always the same check on the same Mixin-Tree, irrespective of the
> actual NodeImpl instance it is invoked on. For that the result of the
> AccessManager.hasPermission call could be cached on a session level (and
> invalidated on {{Session.refresh()}}).
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
