[
https://issues.apache.org/jira/browse/LUCENE-6055?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14204796#comment-14204796
]
Shai Erera commented on LUCENE-6055:
------------------------------------
I don't know, but it could be. I.e. in regular indexing, the payload is
consumed before the next document populates it ... unless you use
TeeSinkTokenFilter and then the payload of the first may control the values for
all. But it's hard to tell - need a concrete example.
> PayloadAttribute.clone() should deep clone its BytesRef
> -------------------------------------------------------
>
> Key: LUCENE-6055
> URL: https://issues.apache.org/jira/browse/LUCENE-6055
> Project: Lucene - Core
> Issue Type: Bug
> Components: core/index
> Reporter: Shai Erera
> Assignee: Shai Erera
> Fix For: Trunk, 5.x
>
> Attachments: LUCENE-6055.patch, LUCENE-6055.patch
>
>
> PayloadAttribute.clone() does a shallow clone, unlike e.g. CharTermAttribute.
> Attributes should deep clone, otherwise capturing state isn't correct. In
> addition, both PA's and CTA's .clone() falsely documents that they do shallow
> cloning on purposes, so need to fix that too.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
