[
https://issues.apache.org/jira/browse/LUCENE-6976?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15098150#comment-15098150
]
Adrien Grand commented on LUCENE-6976:
--------------------------------------
>Anyone have an opinion if BytesRef.deepCopyOf should handle null?
I would rather like it to not special-case null as it could hide bugs?
The patch looks good, even if it's a bit awkward that the second attribute is
instantiated through reflection in the test?
> BytesTermAttributeImpl.copyTo NPEs when the BytesRef is null
> ------------------------------------------------------------
>
> Key: LUCENE-6976
> URL: https://issues.apache.org/jira/browse/LUCENE-6976
> Project: Lucene - Core
> Issue Type: Bug
> Reporter: David Smiley
> Assignee: David Smiley
> Priority: Minor
> Attachments: LUCENE_6976.patch
>
>
> The BytesTermAttributeImpl class, not used much I think, has a problem in its
> copyTo method in which it assumes "bytes" isn't null since it calls
> BytesRef.deepCopyOf on it. Perhaps deepCopyOf should support null? And
> also, toString(), equals() and hashCode() aren't implemented but we can do so.
> This was discovered in SOLR-8541; the spatial PrefixTreeStrategy uses this
> attribute and the CachingTokenFilter when used on the analysis chain will
> call clearAttributes() in it's end() method and then capture the state so it
> can be replayed later. BytesTermAttributeImpl.clear() nulls out the bytes
> reference.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
