[
https://issues.apache.org/jira/browse/VFS-548?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14905174#comment-14905174
]
Bernd Eckenfels commented on VFS-548:
-------------------------------------
another interpretation of this code
> SoftRefReleaseThread might leak
> --------------------------------
>
> Key: VFS-548
> URL: https://issues.apache.org/jira/browse/VFS-548
> Project: Commons VFS
> Issue Type: Bug
> Affects Versions: 2.0
> Reporter: Bernd Eckenfels
> Priority: Minor
> Labels: leak, thread
>
> The lock-free but non-strict initiallisation of the SoftRefReleaseThread
> might in rare condition construct but not start Thread objects. This is
> actually a leak in older java versions (and I think causes a threadgroup to
> not be destroyed in newer versions). In any case, this optimization is not
> needed as it is on a cold path, so using synchronized to set the new value is
> perfectly fine. (especially as the thread constructor uses synchronized
> methods anyway).
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
