[
https://issues.apache.org/jira/browse/SOLR-5478?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
David Smiley updated SOLR-5478:
-------------------------------
Attachment: SOLR-5478 smiley fl.fieldCacheFields.patch
(Disclaimer; this patch isn't tested and in particular aside from porting it to
trunk from v4.2 just now, I didn't try it out)
The essentials of this patch is a modification in 3 places -- SolrReturnFields
is the start. Here, it detects that the only fields asked for are fields that
are in a whitelist of FieldCache'able fields. It defaults to just the
uniqueKey field. If this occurs, then a flag is set ("skipDocOptimization")
and a "DocTransformer" is added for each such field that is able to fetch the
field value from the FieldCache viewed as a ValueSource. Then, annoyingly this
code is in 2 places but in both BinaryResponseWriter and in TextResponseWriter
there's logic that fetches the lucene document. I modified it to see if the
skipDocOptimization flag is set and if so then to skip fetching the Document.
The transformers will run and add the id field and any others you might
configure.
I'd appreciate a look from others and recommendations on how to test it.
> Speed-up distributed search with high rows param or deep paging by
> transforming docId's to uniqueKey via memory docValues
> -------------------------------------------------------------------------------------------------------------------------
>
> Key: SOLR-5478
> URL: https://issues.apache.org/jira/browse/SOLR-5478
> Project: Solr
> Issue Type: Improvement
> Components: Response Writers
> Affects Versions: 4.5
> Reporter: Manuel Lenormand
> Fix For: 4.7
>
> Attachments: SOLR-5478 smiley fl.fieldCacheFields.patch,
> SOLR-5478.patch, SOLR-5478.patch
>
>
--
This message was sent by Atlassian JIRA
(v6.1.5#6160)
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org
