You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@lucene.apache.org by "Michael McCandless (JIRA)" <ji...@apache.org> on 2016/11/01 23:22:58 UTC
[jira] [Reopened] (LUCENE-7501) Do not encode the split dimension
in the index in the 1D case
[ https://issues.apache.org/jira/browse/LUCENE-7501?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Michael McCandless reopened LUCENE-7501:
----------------------------------------
I think there's a small back compat bug here, that looks like corruption.
I'll attach a quick patch, but maybe there's a cleaner way e.g. to factor out the address++ / splitDim logic.
I also hit some silly crabs in CheckIndex.
> Do not encode the split dimension in the index in the 1D case
> -------------------------------------------------------------
>
> Key: LUCENE-7501
> URL: https://issues.apache.org/jira/browse/LUCENE-7501
> Project: Lucene - Core
> Issue Type: Improvement
> Reporter: Adrien Grand
> Assignee: Adrien Grand
> Priority: Minor
> Fix For: master (7.0), 6.3
>
> Attachments: LUCENE-7501.patch
>
>
> When there is a single dimension, the split dimension is always 0, so we do not need to encode it in the index of the BKD tree. This would be 33% memory saving for half floats, 20% for ints/floats, 11% for longs/doubles and 6% for ip addresses.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@lucene.apache.org
For additional commands, e-mail: dev-help@lucene.apache.org