[
https://issues.apache.org/jira/browse/OAK-5229?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15858098#comment-15858098
]
Alex Parvulescu commented on OAK-5229:
--------------------------------------
attaching new version of the patch [^OAK-5229-v3.patch].
Based on an offline discussion with [~anchela], I updated the patch again:
* added a lightweight verification of the provided primary type (very similar
to what is happening to the mixins) while at the same time full validation of
the transient space is not covered (validation is still based on the
commithook).
* revisited changing the primary type to add the missing autocreated properties
(similar to the mixin situation).
* made adding the same primary type again to a node into a noop.
Benchmarks attached. Even though the patched version is a lot better off, I
noticed very large variance between subsequent runs of the same bench, so I'm
not convinced these numbers actually mean anything, but this might merit a
dedicated discussion.
Unpatched version
{noformat}
# ConcurrentReadWriteTest C min 10% 50% 90% max
N
Oak-Segment-Tar 1 45 107 125 151 264
705
# ConcurrentWriteReadTest C min 10% 50% 90% max
N
Oak-Segment-Tar 1 3 6 16 32 141
5259
# ConcurrentWriteTest C min 10% 50% 90% max
N
Oak-Segment-Tar 1 78 81 85 94 175
1038
# CreateManyChildNodesTest C min 10% 50% 90% max
N
Oak-Segment-Tar 1 445 455 480 532 581
149
# CreateManyNodesTest C min 10% 50% 90% max
N
Oak-Segment-Tar 1 134 138 154 164 196
593
# ConcurrentCreateNodesTest C min 10% 50% 90% max
N
Oak-Segment-Tar 1 512 525 552 598 641
163
# SequentialCreateNodesTest C min 10% 50% 90% max
N
Oak-Segment-Tar 1 211 222 243 258 336
350
{noformat}
Patched version
{noformat}
# ConcurrentReadWriteTest C min 10% 50% 90% max
N
Oak-Segment-Tar 1 40 98 117 139 252
761
# ConcurrentWriteReadTest C min 10% 50% 90% max
N
Oak-Segment-Tar 1 3 5 16 30 158
5524
# ConcurrentWriteTest C min 10% 50% 90% max
N
Oak-Segment-Tar 1 77 80 82 90 154
1070
# CreateManyChildNodesTest C min 10% 50% 90% max
N
Oak-Segment-Tar 1 471 481 539 567 612
138
# CreateManyNodesTest C min 10% 50% 90% max
N
Oak-Segment-Tar 1 144 147 154 171 201
577
# ConcurrentCreateNodesTest C min 10% 50% 90% max
N
Oak-Segment-Tar 1 496 506 521 557 605
171
# SequentialCreateNodesTest C min 10% 50% 90% max
N
Oak-Segment-Tar 1 209 214 222 243 271
375
{noformat}
[~tripod] does this cover everything?
> Using Node.setPrimaryType() should fail if non-matching childnodes
> ------------------------------------------------------------------
>
> Key: OAK-5229
> URL: https://issues.apache.org/jira/browse/OAK-5229
> Project: Jackrabbit Oak
> Issue Type: Bug
> Components: core
> Affects Versions: 1.5.14
> Reporter: Tobias Bocanegra
> Assignee: Alex Parvulescu
> Priority: Critical
> Fix For: 1.8, 1.6.1
>
> Attachments: OAK-5229.patch, OAK-5229-tests.patch, OAK-5229-v2.patch,
> OAK-5229-v3.patch
>
>
> 1. Assume the following:
> {noformat}
> /testNode [nt:unstructured]
> /unstructured_child [nt:unstructured]
> {noformat}
> 2. setting "/testNode".setPrimaryType("nt:folder")
> 3. save the session.
> Altering the primary type works, thus leaving the repository in an
> inconsistent state.
> Interestingly, subsequent calls to
> "/testNiode/unstructured_child".setProperty() will fail:
> {noformat}
> javax.jcr.nodetype.ConstraintViolationException: OakConstraint0001:
> /test_node[[nt:folder]]: No matching definition found for child node
> unstructured_child with effective type [nt:unstructured]
> {noformat}
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
