[ 
https://issues.apache.org/jira/browse/LUCENE-7752?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16051061#comment-16051061
 ] 

Daniel Jelinski commented on LUCENE-7752:
-----------------------------------------

Okay, I think I got to the bottom of this.
The FlattenGraphFilter is a bit lazy when it comes to freeing visited graph 
nodes. When it's done processing nodes 1,3,5, it frees nodes <1 (=0). When it's 
done processing nodes 2,4,6 it frees nodes <2 (=1). In order to process node 7 
it loads the entire graph up to node 14, pushing the number of concurrently 
buffered nodes up to 13, which triggers test failure.
I modified the code to free nodes eagerly, so that releasing unneeded nodes 
happens before processing, not after. This made the beast happy. Anyway, this 
problem probably deserves a separate bug.

> Findbugs: Array.equals is equivalent to comparing references
> ------------------------------------------------------------
>
>                 Key: LUCENE-7752
>                 URL: https://issues.apache.org/jira/browse/LUCENE-7752
>             Project: Lucene - Core
>          Issue Type: Sub-task
>            Reporter: Daniel Jelinski
>            Priority: Minor
>         Attachments: failing_graph.png, LUCENE-7752.patch
>
>




--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to