You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@lucene.apache.org by "Shai Erera (JIRA)" <ji...@apache.org> on 2012/05/16 09:02:07 UTC
[jira] [Updated] (LUCENE-4060)
DirectoryTaxonomyWriter.addTaxonomies does not work well in a
multi-threaded environment
[ https://issues.apache.org/jira/browse/LUCENE-4060?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Shai Erera updated LUCENE-4060:
-------------------------------
Attachment: LUCENE-4060.patch
Patch does:
* Fix the bug
* Renames addTaxonomies to addTaxonomy
* Greatly simplifies addTaxonomy logic
* Overhauls TestAddTaxonomies to be simpler and test what it needs to test - for some reason it tested that after addTaxonmy the new categories are sorted in a lexicographic order, which is a side effect of addTaxonomy, but does not need to be like that.
NOTE: before applying the patch you should run
{code}
svn mv lucene/contrib/facet/src/test/org/apache/lucene/facet/taxonomy/directory/TestAddTaxonomies.java lucene/contrib/facet/src/test/org/apache/lucene/facet/taxonomy/directory/TestAddTaxonomy.java
{code}
One nice side effect of this improvement is that testBig, with the really huge taxonomy, takes 12 seconds faster to run after the change. While this is not a pure benchmark, it does show that the modified implementation is faster than the older.
I will commit this shortly and port to trunk.
> DirectoryTaxonomyWriter.addTaxonomies does not work well in a multi-threaded environment
> ----------------------------------------------------------------------------------------
>
> Key: LUCENE-4060
> URL: https://issues.apache.org/jira/browse/LUCENE-4060
> Project: Lucene - Java
> Issue Type: Bug
> Components: modules/facet
> Reporter: Shai Erera
> Assignee: Shai Erera
> Fix For: 4.0, 3.6.1
>
> Attachments: LUCENE-4060.patch
>
>
> DirTaxoWriter.addTaxonomies may lead to a corrupt taxonomy index if addCategory is called in parallel. While reviewing the code, the following changes seemed to be required:
> * Fix the bug
> * Simplify addTaxonomies code -- it's very complicated, seems rather inefficient, and seems to take care of more than is needed.
> * The intention to support multiple taxonomies is unclear, especially since FacetsPayloadProcessor can handle only one Directory and one OrdinalMap at a time. So I'd like to change the method to addTaxonomy. I anyway don't see the reason to add multiple taxonomies at once, one can call addTaxonomy several times.
> I will post a patch soon.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@lucene.apache.org
For additional commands, e-mail: dev-help@lucene.apache.org