openjpa-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Donald Woods (JIRA)" <j...@apache.org>
Subject [jira] Commented: (OPENJPA-1755) MultiClassLoader relies on its underlying ClassLoaders for hashCode
Date Tue, 21 Sep 2010 20:59:33 GMT

    [ https://issues.apache.org/jira/browse/OPENJPA-1755?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12913247#action_12913247
] 

Donald Woods commented on OPENJPA-1755:
---------------------------------------

Patrick, was hashCode() throwing an Exception for you in your GAE scenario?  If so, catching
the exception, log a msg and then return 1 might be an acceptable compromise....


> MultiClassLoader relies on its underlying ClassLoaders for hashCode
> -------------------------------------------------------------------
>
>                 Key: OPENJPA-1755
>                 URL: https://issues.apache.org/jira/browse/OPENJPA-1755
>             Project: OpenJPA
>          Issue Type: Bug
>            Reporter: Patrick Linskey
>         Attachments: OPENJPA-1755.txt
>
>
> The MultiClassLoader relies on its underlying ClassLoaders in its hashing algorithm.
This is problematic if the underlying classloaders refuse to return a hash or if the classloader
list changes over time.
> I ran into a problem that demanded a resolution to this issue when doing some work with
OpenJPA in a Google AppEngine context. I don't have a stack trace handy.
> An effective-yet-ugly fix was to simply change the hashCode() implementation to return
a constant value (I chose '1'). This clearly is a crappy hashing algorithm, but will always
respect the hashing contract. And I don't think that we really put MultiClassLoaders into
big maps or anything.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


Mime
View raw message