[xwiki-notifications] [Issue] Commented: (XWATCH-116) Use Lucene as the underlying query engine

[Anca Paula Luca (JIRA)]

    [ http://jira.xwiki.org/jira/browse/XWATCH-116?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=28408#action_28408 ] 

Anca Paula Luca commented on XWATCH-116:
----------------------------------------

Since Lucene is a full text indexing engine and Watch (and XWiki) data is organized by a relational model, the implementation of Lucene search for Watch queries is not very straightforward. The standard way in which databases are mapped for Lucene indexing is denormalization of database and creating one Lucene document per denormalized record, with the big drawback that it requires a lot more time on index creation and update. There are two approaches for object querying using Lucene indexes:
- either we change the Lucene plugin and allow storing each property in a field so that we can perform structured searches.
- or we create a specific watch indexing plugin, that would index and search Watch specific data . This last option would prevent Watch distribution as xar.