[
https://issues.apache.org/jira/browse/LUCENE-3496?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13122942#comment-13122942
]
Martijn van Groningen commented on LUCENE-3496:
-----------------------------------------------
Up until now I only checked the difference between BYTES_VAR_SORTED and
BYTES_FIXED_SORTED. The search time and memory usage seem to be more or less
the same. I executed all queries on 30M index and the performance with grouping
by IDV is somewhere between 500 ms and 550 ms. The same test with grouping by
indexed values have search times somewhere between 300 ms and 330 ms.
bq. We might need to round up the packedint sizes inside IDV for lookup perf
and pay the prize for a slightly larger size on disk / memory.
I think we need this. Just like we have with FC#getTermsIndex(). A boolean
option that either uses PackedInt or DirectInt. I think the ~200 ms difference
is b/c IDV is using PackedInt and FC uses DirectInt.
> Support grouping by IndexDocValues
> ----------------------------------
>
> Key: LUCENE-3496
> URL: https://issues.apache.org/jira/browse/LUCENE-3496
> Project: Lucene - Java
> Issue Type: New Feature
> Components: modules/grouping
> Reporter: Martijn van Groningen
> Attachments: LUCENE-3496.patch
>
>
> Although IDV is not yet finalized (More particular the SortedSource). I think
> we already can discuss / investigate implementing grouping by IDV.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org
