jackrabbit-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Ard Schrijvers (JIRA)" <j...@apache.org>
Subject [jira] Created: (JCR-2311) Improve indexing performance by reusing already create Lucene Documents within a single indexing transaction
Date Fri, 18 Sep 2009 08:26:57 GMT
Improve indexing performance by reusing already create Lucene Documents within a single indexing
transaction
------------------------------------------------------------------------------------------------------------

                 Key: JCR-2311
                 URL: https://issues.apache.org/jira/browse/JCR-2311
             Project: Jackrabbit Content Repository
          Issue Type: Improvement
          Components: indexing
    Affects Versions: 2.0-alpha9, 1.6.0, 1.5.7
            Reporter: Ard Schrijvers
            Assignee: Ard Schrijvers
            Priority: Minor
             Fix For: 1.5.8, 1.6.1, 2.0-alpha11


When using indexing aggregates, it can become very expensive when for example binary data
like pdfs are involved. Within a single MultiIndex transaction, we should be save to reuse
the created Lucene Document's. This is useful for aggregates, as this might currently involve
reindexing the same node multiple times now in a single transaction. 

In the mail discussion below I describe to use WeakReferences, though I think SoftReferences
are in this case better suited (we do not want the created lucene docs to be gc-ed, unless
memory gets low...)

Also see http://www.nabble.com/Improve-indexing-performance-wrt-Aggregates-td25471828.html


-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


Mime
View raw message