[
https://issues.apache.org/jira/browse/LUCENE-1805?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Shai Erera updated LUCENE-1805:
-------------------------------
Attachment: LUCENE-1805.patch
Remove assert v != null and added a test case.
Funny, but it's as if CTL itself could cause this bug. If you call ctl.get()
w/o setting anything first, it creates a WeakReference and set the initValue(),
which is ... null !. The first call to get() will return null, the second would
fail. :)
BTW, TestCloseableThreadLocal is under o.a.l.index. Shouldn't it be under
o.a.l.util? Can whoever assigns this to himself move it?
> CloseableThreadLocal should allow null Objects
> ----------------------------------------------
>
> Key: LUCENE-1805
> URL: https://issues.apache.org/jira/browse/LUCENE-1805
> Project: Lucene - Java
> Issue Type: Bug
> Components: contrib/analyzers
> Affects Versions: 2.4.1
> Reporter: Shai Erera
> Fix For: 2.9
>
> Attachments: LUCENE-1805.patch
>
>
> CloseableThreadLocal does not allow null Objects in its get() method, but
> does nothing to prevent them in set(Object). The comment in get() before
> assert v != null is irrelevant - the application might have passed null.
> Null is an important value for Analyzers. Since tokenStreams (a ThreadLocal
> private member in Analyzer) is not accessible by extending classes, the only
> way for an Analyzer to reset the tokenStreams is by calling
> setPreviousTokenStream(null).
> I will post a patch w/ a test
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
---------------------------------------------------------------------
To unsubscribe, e-mail: java-dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: java-dev-h...@lucene.apache.org
