[
https://issues.apache.org/jira/browse/PHOENIX-1287?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14271801#comment-14271801
]
James Taylor commented on PHOENIX-1287:
---------------------------------------
Here's a bit more detail on what would be required for this one:
- Introduce a RegExEval interface to abstract the regular expression
compilation/evaluation class.
- Create two implementations: one that maintains the existing usage of
java.util.regex and one that uses this new library. For the evaluation method,
the input would be the child Expression and an ImmutableBytesWritable that
contains the already evaluated child expression. The java.util.regex would
convert the byte pointer to a String while the other wouldn't.
- Create a new boolean config parameter in QueryServices with a default value
in QueryServicesOptions that'll be used by the RegExEval factory to determine
which implementation gets used.
- Modify LikeExpression, RegexpReplaceFunction, RegexpSplitFunction,
RegexpSubstrFunction to go through the ReExEval interface.
> Use the joni byte[] regex engine in place of j.u.regex
> ------------------------------------------------------
>
> Key: PHOENIX-1287
> URL: https://issues.apache.org/jira/browse/PHOENIX-1287
> Project: Phoenix
> Issue Type: Bug
> Reporter: James Taylor
> Assignee: Andrew Purtell
>
> See HBASE-11907. We'd get a 2x perf benefit plus it's driven off of byte[]
> instead of strings.Thanks for the pointer, [~apurtell].
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
