[
https://issues.apache.org/jira/browse/HBASE-13634?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14531040#comment-14531040
]
Sergey Shelukhin commented on HBASE-13634:
------------------------------------------
Hmm... My memory is rusty on this. Could these checks be working because
everything ever passed to compare is actually the same constant?
It's not like we expect people to pass random empty/magic arrays that just
happen to be the same as constant value.
Patch looks safe but I'm not sure it's necessary... I could be wrong though.
Let me take a look later this week.
> Unsafe reference equality checks to EMPTY_START_ROW
> ---------------------------------------------------
>
> Key: HBASE-13634
> URL: https://issues.apache.org/jira/browse/HBASE-13634
> Project: HBase
> Issue Type: Bug
> Components: Compaction, Scanners
> Reporter: Dave Latham
> Attachments: HBASE-13634.patch
>
>
> While looking to see if there was a standard method in the code base for
> testing for the empty start and end row, I noticed some cases that are using
> unsafe reference equality checks and thus may have incorrect behavior in
> boundary cases:
> ScanQueryMatcher.checkPartialDropDeleteRange
> StripeStoreFileManager.findStripeForRow
> It looks like both are intended to support stripe compaction
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
