[
https://issues.apache.org/jira/browse/PHOENIX-1242?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14176175#comment-14176175
]
James Taylor commented on PHOENIX-1242:
---------------------------------------
Should have looked at this a little closer before, but why aren't you just
looking at the byte array, the offset, and the length and throwing up front if
the length is insufficient? Also, why are you catching a RuntimeException and
then doing an instanceof check and then rethrowing only if
ArrayIndexOutOfBoundsException when you could just catch an
ArrayIndexOutOfBoundsException and then throw it?
Also, FYI, if you throw a SQLException that's wrapped in a RuntimeException,
it'll unwrap it for you.
I'll tweak this a bit if you all don't mind.
> ArrayoutofBoundException Phoenix mapping to exisiting Hbase
> -----------------------------------------------------------
>
> Key: PHOENIX-1242
> URL: https://issues.apache.org/jira/browse/PHOENIX-1242
> Project: Phoenix
> Issue Type: Bug
> Affects Versions: 4.0.0
> Reporter: nicolas maillard
> Assignee: Alicia Ying Shu
> Priority: Minor
> Fix For: 5.0.0, 4.2, 3.2
>
> Attachments: Phoenix-1242-V1.patch, phoenix-1242.patch
>
>
> In a Phoenix mapping to an existing Hbase Table in case of mis-typing say try
> to cast a long an actual string type, Phoenix will break with:
> java.lang.ArrayIndexOutOfBoundsException: 324
> This is a user mis-use but the error message is misleading. The problem is
> trying to copy a string to a long a fixed Bytes[] in long decodeLong.
> Maybe a test to check if the translation is possible and a type error if size
> don't match would provide for a more explicit error message for the user to
> understand what he did wrong.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
