[
https://issues.apache.org/jira/browse/VXQUERY-188?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15207419#comment-15207419
]
Steven Jacobs commented on VXQUERY-188:
---------------------------------------
Our current model involves pre-indexing sources, not indexing at runtime. We
have "build index" statement that creates an index for a given collection.
Since our Lucene index is path-based, the index is actually capable of
returning the results for the query.
Right now the query can be written in two forms: query the collection, or query
the index. The improvement would be to dynamically decide whether the index
would make the execution faster or not, i.e.
query from Cli ---> Decide whether to use the index ----> Executing the query
in system accordingly.
The third task comes from the fact that the Lucene index is build manually
offline. This means that changes to the XML files will not automatically
reflect in the index. This task would be to build some sort of an update
mechanism (faster than a simple rebuild of the index), whether adding a manual
rebuild query or possibly finding a way to automatically update before use
(keeping in mind that VXQuery is not an "online" system, meaning that it can't
have an active monitoring task).
> Fully integrate Lucene Indexing into VXQuery
> --------------------------------------------
>
> Key: VXQUERY-188
> URL: https://issues.apache.org/jira/browse/VXQUERY-188
> Project: VXQuery
> Issue Type: Bug
> Reporter: Steven Jacobs
> Labels: GSOC2016, HDFS, JSONiq, Lucene
>
> Currently, the indexing project for VXQuery is able to perform two tasks:
> 1) Create a Lucene index from an XML collection (folder).
> 2) Specify in a query to use this index to perform the query.
> There are several more desired capabilities to fully integrate Lucene
> indexing:
> 1) Enable queries to dynamically decide at runtime when to use indexes.
> 2) Extend indexing to HDFS folders.
> 3) Allow updates to collection indexes (when Adding/Deleting/Modifying XML
> files).
> 4) As JSONiq becomes integrated into VXQuery, extend indexing to work with
> either XQuery or JSONiq.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
