That sounds like a map, or a document structure that acts much like one. There 
is some earlier discussion of the problem at 
http://markmail.org/message/oxhlwm5uzscvlniw - reading through it might help, 
if you haven't already.

-- Mike

On 4 Jun 2013, at 05:43 , Paul M <[email protected]> wrote:

> Say I have documents in the form
> <doc>
> <datanode>
> <lookup-nodes>
> <lookup-nodes>
> <lookup-nodes>
> <lookup-nodes>
> …
> <lookup-nodes>
> 
> 
> Where I may have thousands of lookup-nodes. Usually, I would split-out the 
> lookup nodes into separate documents along with datanode to keep fragment/doc 
> size small, while inserting datanode-key to point back to datanode. Then when 
> I search for a lookup-node, I get  datanode-key, and return datanode. 
> However, what if I almost always return/utilize entire document doc. Still 
> split-out lookup-nodes? Just leave as large document and limit paging size. 
> This is a java/xquery setup using xcc.  Seems odd splitting apart, search on 
> lookup-nodes, find datanode-keys,  then loop and query matching on 
> datanode-keys to get all the pieces of the doc for merging together in java 
> layer. Trying to see when/how split a document up. 
> 
> _______________________________________________
> General mailing list
> [email protected]
> http://developer.marklogic.com/mailman/listinfo/general

_______________________________________________
General mailing list
[email protected]
http://developer.marklogic.com/mailman/listinfo/general

Reply via email to