[ 
https://issues.apache.org/jira/browse/CASSANDRA-12516?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15505259#comment-15505259
 ] 

Stefania commented on CASSANDRA-12516:
--------------------------------------

Thanks for the review, committed to 3.0 as 
4bc3aa93337474a29bc4ad43ff4438755c9c7ab8 and merged into trunk.

> Interned column identifiers can be overridden incorrectly
> ---------------------------------------------------------
>
>                 Key: CASSANDRA-12516
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-12516
>             Project: Cassandra
>          Issue Type: Bug
>          Components: Local Write-Read Paths
>            Reporter: Aleksey Yeschenko
>            Assignee: Stefania
>             Fix For: 3.10, 3.0.10
>
>
> As discovered by [~Stefania] in CASSANDRA-12423, our use of 
> {{ColumnIdentifier.internedInstances}} to map byte buffer column names to 
> column identifiers is incorrect. It's possible for two non-equal 
> {{ColumnIdentifier}} instances to have the same {{ByteBuffer}} as raw name, 
> and yet different text representations, due to different column name types, 
> leading to hard to debug correctness issues.
> Should either amend the map to have the type as part of the key, or limit the 
> use of that map.



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

Reply via email to