[ 
https://issues.apache.org/jira/browse/JS2-1198?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Woonsan Ko resolved JS2-1198.
-----------------------------

    Resolution: Fixed

Upgraded Lucene to the latest one, 3.0.2 with the following improvements:
- made the test case to not open http connection during testing; instead load 
html content from the local classpath urls.
- added 'synthetic' field (which is indexed but not stored) to simplify query 
without using MultiFieldQueryParser.
- improved not to add multiple documents for the same portlet.
- improved resource usages to avoid possible leaks.
- added max top count in search API because Lucene has that limitation in API 
as well.
- added portlet registry search test case.
- added new search method to allow to set the default search field.
- improved the search in toolbox on keywords.
- refactoring in search engine impl.
- optimization during adding or removing by reducing index-rewriting calls.

By the way, I didn't change the default search index directory by RAMDirectory
because the file system directory is very useful to investigate with Luke.
However, the SearchEngineImpl has additional constructors to allow directory 
injection; it's possible to use RAMDirectory by configuring spring assembly 
manually.

One thing to note is that the search index directory format has been changed to 
3.0; so the existing search index directory in the previous version couldn't be 
compliant any more. They should be recreated.

> Improve search engine usage
> ---------------------------
>
>                 Key: JS2-1198
>                 URL: https://issues.apache.org/jira/browse/JS2-1198
>             Project: Jetspeed 2
>          Issue Type: Improvement
>          Components: Portlet Registry
>    Affects Versions: 2.2.1
>            Reporter: Woonsan Ko
>            Assignee: Woonsan Ko
>             Fix For: 2.2.2
>
>
> I think we can improve the search engine usages with the recent lucene (3.0.1 
> or later) like the followings:
>  - Use explicit term queries instead of string based query which could have 
> made problems to not distinguish portlet app and portlet def.
>  - By default, we can configure RAM Directory instead of FS Directory.
>    Because the current search index store is small-sized for portlet 
> registry, the in-memory search index could be no problem by default.
>    Of course, we could provide a configuration and spring assembly to allow 
> search index directory customizations.
>  - Currently, we don't have options to check or reinitialize the search index 
> directory. If we do, it would be great because we don't have to restart the 
> servlet container to reinitialize the search index.
>  

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


---------------------------------------------------------------------
To unsubscribe, e-mail: jetspeed-dev-unsubscr...@portals.apache.org
For additional commands, e-mail: jetspeed-dev-h...@portals.apache.org

Reply via email to