[
https://issues.apache.org/jira/browse/HBASE-3387?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12982671#action_12982671
]
Jesse Yates commented on HBASE-3387:
------------------------------------
This was introduced to correct behavior in software on my side (e.g. non-HBase
core) that for all intents and purposes should have worked, but didn't because
of the previous definition of Pair.equals().
As far as the equals/hashCode issue, we could also overwrite the hashCode in
Pair to ensure that the equals and hashCode methods work out to the same
equality.
This change was made because they underlying arrays really should check the
underlying types in Pairs, to check if they really would be the same thing. If
we fix the hashCode method, then I think we end up with a system that is really
more flexible than just comparing the pointer addresses.
. Is there somewhere in HBase where this breaks the current model and I just
missed it?
> Pair does not deep check arrays for equality.
> ----------------------------------------------
>
> Key: HBASE-3387
> URL: https://issues.apache.org/jira/browse/HBASE-3387
> Project: HBase
> Issue Type: Bug
> Components: util
> Affects Versions: 0.90.1
> Environment: Any (discovered in Ubuntu 10.10 using TRUNK).
> Reporter: Jesse Yates
> Priority: Minor
> Fix For: 0.90.1, 0.92.0
>
> Attachments: HBASE-3387.patch
>
> Original Estimate: 0h
> Remaining Estimate: 0h
>
> Pair does not deep check arrays for equality. It merely does x.equals(y) for
> the sent Object. However, with any type of array this is merely going to
> compare the array pointers, rather than the underlying data structure.
> It requires a rewriting of the private equals method in Pair to check for
> elements being an array, then checking the underlying elements.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
