[
https://issues.apache.org/jira/browse/OAK-8249?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16819056#comment-16819056
]
Alex Deparvu commented on OAK-8249:
-----------------------------------
This is a wip evaluation branch [0].
preliminary benchmark results look good (almost 4x faster), but the test is
obviously biased because it only does primary type lookup. still good to get a
feel for some raw numbers:
{noformat}
(trunk)
# IsNodeTypeTest C min 10% 50% 90% max
N
Oak-Segment-Tar 1 85 87 92 98 130
644
(patch)
# IsNodeTypeTest C min 10% 50% 90% max
N
Oak-Segment-Tar 1 21 22 23 24 32
2590
{noformat}
Unfortunately there is an Interator<> passed around so I had to wrap into a
custom lazy loading iterator (not sure if something like this exists already,
it is generic enough to go into some sort of commons code if needed).
Problem 2: dealing with the RepositoryException from the lazy loader is a pain,
the current solution is very hacky.
[~mreutegg] what do you think?
[0] https://github.com/apache/jackrabbit-oak/compare/trunk...stillalex:OAK-8249
> NodeImpl#isNodeType could load mixin info lazily
> ------------------------------------------------
>
> Key: OAK-8249
> URL: https://issues.apache.org/jira/browse/OAK-8249
> Project: Jackrabbit Oak
> Issue Type: Improvement
> Components: jcr
> Reporter: Alex Deparvu
> Assignee: Alex Deparvu
> Priority: Minor
>
> The current isNodeType check loads both primary type and all mixins eagerly.
> I'm wondering how often is the case where someone only needs a primary type
> check, and if loading the mixins lazily would improve the throughput.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
