Background info:

• We want to be able to find content not only at page level but in components nested in areas and sub-areas
• We do not want to use JCR_SQL2 joins between page nodes and their descendants in searches b/c they are too slow

The solution to the above problems is to use JR indexing configuration, in particular aggregate indexing. http://wiki.apache.org/jackrabbit/IndexingConfiguration
This works out pretty well in terms of speed however it returns more results than what one expects.

To reproduce in demo-project:

• setup indexing in website workspace by adding

   <param name="indexingConfiguration" value="${wsp.home}/indexing_configuration.xml"/> 

  to workspace.xml in the SearchIndex part

• copy https://git.magnolia-cms.com/gitweb/?p=ce-bundle.git;a=blob;f=magnolia-empty-webapp/src/main/webapp/WEB-INF/config/repo-conf/indexing_configuration.xml;h=5969539c3380f0a336e06f2183a725ec206390d8;hb=30599019d4e0d13b44cc6ede6948f3c30111d794
  to the workspace folder (same level as the workspace.xml)
• delete the index folder and (re-)start Magnolia
• perform a search for "white-paper"
• this will return two hits/pages /demo-project/about/subsection-articles/article and /demo-project/about/subsection-articles

The term we searched for is actually present in page /demo-project/about/subsection-articles/article namely in a component (named 0) found in the sub-area /extras/extras1. So, the first hit (the article page) is correct, whereas the second one (the subsection-articles page) is not.

Trying with text only present in the same page in a component under the content area would instead return 3 hits!

• /demo-project/about/subsection-articles/article (the correct one)
• /demo-project/about/subsection-articles
• /demo-project/about

This might possibly have to do with this JR issue https://issues.apache.org/jira/browse/JCR-2947