jackrabbit-oak-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Vikas Saurabh (JIRA)" <j...@apache.org>
Subject [jira] [Comment Edited] (OAK-4601) In case of branch commits, nodeCache gets entries for branch revision without "br" prefix
Date Tue, 26 Jul 2016 12:04:20 GMT

    [ https://issues.apache.org/jira/browse/OAK-4601?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15393686#comment-15393686
] 

Vikas Saurabh edited comment on OAK-4601 at 7/26/16 12:03 PM:
--------------------------------------------------------------

Further investigation showed that a test like \[0] gets 1 branch with base={{rBase}}, 1commit={rCommit}}.
The cache entries in node cache show up for
# {{childX\@brCommit}}
# {{/\@brCommit}}
# {{/\@rBase}}
# {{/\@brBase}}

[~mreutegg] pointed out that technically, both #3 and #4 are correct semantically although
they carry the same information from node state pov and could potentially be avoided. But,
that's too minor a space waste to worry about.... resolving as not a problem.

\[0]:
{code}
@Test
public void nodeCacheForBranchCommit() throws Exception {
    DocumentNodeStore ns = builderProvider.newBuilder().getNodeStore();
    NodeBuilder b1 = ns.getRoot().builder();
    final int NUM_CHILDREN = 3*DocumentRootBuilder.UPDATE_LIMIT + 1;
    //this would push cache entries -> #nodes=3*update_limit
    for (int i = 0; i < NUM_CHILDREN; i++) {
        b1.child("child" + i);
    }
    //this would pull in cache entries
    for (int i = 0; i < NUM_CHILDREN; i++) {
        b1.getChildNode("child" + i);
    }
}
{code}


was (Author: catholicon):
Further investigation showed that a test like \[0] gets 1 branch with base={{rBase}}, 1commit={rCommit}}.
The cache entries in node cache show up for
# {{childX\@brCommit}}
# {{/\@brCommit}}
# {{/\@rBase}}
# {{/\@brBase}}

Technically, #3 and #4 carry the same information from node state pov and could potentially
be avoided. But, that's too minor a space waste to worry about.

\[0]:
{code}
@Test
public void nodeCacheForBranchCommit() throws Exception {
    DocumentNodeStore ns = builderProvider.newBuilder().getNodeStore();
    NodeBuilder b1 = ns.getRoot().builder();
    final int NUM_CHILDREN = 3*DocumentRootBuilder.UPDATE_LIMIT + 1;
    //this would push cache entries -> #nodes=3*update_limit
    for (int i = 0; i < NUM_CHILDREN; i++) {
        b1.child("child" + i);
    }
    //this would pull in cache entries
    for (int i = 0; i < NUM_CHILDREN; i++) {
        b1.getChildNode("child" + i);
    }
}
{code}

> In case of branch commits, nodeCache gets entries for branch revision without "br" prefix
> -----------------------------------------------------------------------------------------
>
>                 Key: OAK-4601
>                 URL: https://issues.apache.org/jira/browse/OAK-4601
>             Project: Jackrabbit Oak
>          Issue Type: Bug
>            Reporter: Vikas Saurabh
>            Assignee: Vikas Saurabh
>            Priority: Minor
>
> Similar to OAK-4600, nodeCache also gets cached entries without {{br}} prefix. Still
investigating when/where are those entries created.
> Btw, unlike, OAK-4600, this affects trunk too.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message