[ 
https://issues.apache.org/jira/browse/PHOENIX-1989?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14551987#comment-14551987
 ] 

James Taylor commented on PHOENIX-1989:
---------------------------------------

The serialized binary representation in Phoenix is always UTF8. We'd convert 
the string being searched for to the same binary representation. Then we can 
walk from the beginning and skip the correct number of bytes (based on UTF8 
encoding rules) until we find the bytes or reach the end the byte array.

> Implement byte-based INSTR instead of serializing into String
> -------------------------------------------------------------
>
>                 Key: PHOENIX-1989
>                 URL: https://issues.apache.org/jira/browse/PHOENIX-1989
>             Project: Phoenix
>          Issue Type: Bug
>            Reporter: James Taylor
>
> The current implementation of INSTR serializes the arguments as Strings. It'd 
> be much more efficient to leave them as bytes and do the in-string search 
> based on bytes.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Reply via email to