[
https://issues.apache.org/jira/browse/POOL-188?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13031668#comment-13031668
]
Sebb commented on POOL-188:
---------------------------
Turns out that the method is overridden by the subclass
PoolUtils.ErodingPerKeyKeyedObjectPool.numIdle(K key) which calls
getKeyedPool().getNumIdle(key) - i.e. behaves the same as the parent
genNumIdle().
Very confusing (though it appears not to be used).
If ErodingKeyedObjectPool.numIdle(key) really is supposed to ignore the key,
then either this should be carefully documented, or (better) the caller should
use the appropriate method in the first place.
> Confusing PoolUtils.ErodingKeyedObjectPool.numIdle(K key) method; bug in
> ErodingKeyedObjectPool.returnObject(K key, V obj) ?
> ----------------------------------------------------------------------------------------------------------------------------
>
> Key: POOL-188
> URL: https://issues.apache.org/jira/browse/POOL-188
> Project: Commons Pool
> Issue Type: Bug
> Affects Versions: 2.0
> Reporter: Sebb
>
> numIdle(key) is equivalent to PoolUtils.ErodingKeyedObjectPool.getNumIdle() -
> unless the method getKeyedPool() is overridden.
> It's confusing that the key parameter is not used - this may be the cause of
> a bug, as the method is used internally by
> ErodingKeyedObjectPool.returnObject(K key, V obj).
> It looks like returnObject() should probably call getNumIdle(key) rather than
> numIdle(key).
> In which case the confusing numIdle method can be deleted.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
