[
https://issues.apache.org/jira/browse/GORA-555?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16807455#comment-16807455
]
Kevin Ratnasekera edited comment on GORA-555 at 4/2/19 7:06 AM:
----------------------------------------------------------------
[~c.uent] The idea here is, right now in current implementation whatever the
Key type is, it's getting converted to java.lang.String before the query
construction and execution. Purpose of this ticket is to avoid that conversion,
use that Key type native as it is ( without conversion ), using Lucene query to
take performance benefits. ( If Lucene query does support that for given Key
type )
Also you need to check whether current implementation seamlessly work for Key
types other than String Eg:- Long once you do that changes. We will review once
you come up PR.
[~lewismc] Is there anything we should add here?
was (Author: djkevincr):
[~c.uent] The idea here is, right now in current implementation whatever the
Key type is, it's getting converted to java.lang.String before the query
construction and execution. Purpose of this ticket is to avoid that conversion,
use that Key type native as it is ( without conversion ), using Lucene query to
take performance benefits. ( If Lucene query does support that for given Key
type )
Also you need to check whether current implementation seamlessly work for Key
types other than String Eg:- Long once you do that changes. We will review once
you come up PR/
[~lewismc] Is there anything we should add here?
> Improve Lucene query implementation with NumericRangeQuery
> -----------------------------------------------------------
>
> Key: GORA-555
> URL: https://issues.apache.org/jira/browse/GORA-555
> Project: Apache Gora
> Issue Type: Improvement
> Reporter: Kevin Ratnasekera
> Priority: Major
>
> There are performance benefits around NumericRangeQuery. Please notice
> comment on LuceneQuery implementation.
> {code}
> //TODO: Change this to a NumericRangeQuery when necessary (it's faster)
> String lower = null;
> String upper = null;
> if (getStartKey() != null) {
> //Do we need to escape the term?
> lower = getStartKey().toString();
> }
> if (getEndKey() != null) {
> upper = getEndKey().toString();
> }
> if (upper == null && lower == null) {
> q = new MatchAllDocsQuery();
> } else {
> q = TermRangeQuery.newStringRange(pk, lower, upper, true, true);
> }
> {code}
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
