[
https://issues.apache.org/jira/browse/NET-532?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13970353#comment-13970353
]
Sebb commented on NET-532:
--------------------------
I realise that, but as already noted I cannot reproduce the NPE with you test
data.
I tried with both the tar.gz attachment and the header-only listing in the
comments (BTW, one reference is truncated in the comments).
Can you provide a complete test case that fails?
Note that the class was originally designed for use with NNTP, in particular
XOVER responses.
It would be useful to know if these responses can trigger the bug, or if the
response is adjusted to remove any missing threads so the problem does not
occur.
> NPE in nntp.Threader.thread() when all messages in a derived thread are
> missing.
> ---------------------------------------------------------------------------------
>
> Key: NET-532
> URL: https://issues.apache.org/jira/browse/NET-532
> Project: Commons Net
> Issue Type: Bug
> Components: NNTP
> Affects Versions: 3.0, 3.3
> Reporter: Elliot Shank
> Attachments:
> messages-referencing-canh-dxkxpwcgncyp_tfvh0yxb6__mkbzfbcnopeyunz_arv...@mail.gmail.com.tar.gz,
> threader-3.3-debug.patch, threader.patch
>
>
> I'm threading emails using the References: header. When I've got an
> incomplete set of messages (in this specific case, it's from one of the C++
> standards committee lists) such that I'm missing an entire thread, I get
> exceptions like this:
> {noformat}
> Exception in thread "main" java.lang.NullPointerException
> at org.apache.commons.net.nntp.Threader.gatherSubjects(Threader.java:342)
> at org.apache.commons.net.nntp.Threader.thread(Threader.java:81)
> at org.apache.commons.net.nntp.Threader.thread(Threader.java:47)
> {noformat}
> What happens is that the graph returned by findRootSet() at line 71 will look
> something like
> {noformat}
> ...
> <ThreadContainer A with Threadable>
> ...
> <ThreadContainer B without Threadable>
> <ThreadContainer ... without Threadable>
> <ThreadContainer ... without Threadable>
> <ThreadContainer ... without Threadable>
> <ThreadContainer ... without Threadable>
> <ThreadContainer ... without Threadable>
> <ThreadContainer ... without Threadable>
> <ThreadContainer C with Threadable>
> ...
> ...
> {noformat}
> And, after pruneEmptyContainers() has had its way with things on line 75, the
> graph will be
> {noformat}
> ...
> <ThreadContainer A with Threadable>
> ...
> <ThreadContainer B without Threadable>
> <ThreadContainer C with Threadable>
> ...
> ...
> {noformat}
> when it should be
> {noformat}
> ...
> <ThreadContainer A with Threadable>
> ...
> <ThreadContainer C with Threadable>
> ...
> ...
> {noformat}
> I.e., ThreadContainer B should be eliminated entirely.
> As it is, with ThreadContainer B present, but not containing any children or
> a Threadable, gatherSubjects() chokes.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
