On 21.12.2009 17:16, Mark Thomas wrote:
On 21/12/2009 15:59, Konstantin Kolinko wrote:
2009/12/21<ma...@apache.org>:
Author: markt
Date: Mon Dec 21 14:30:46 2009
New Revision: 892834
URL: http://svn.apache.org/viewvc?rev=892834&view=rev
Log:
Proposal
+
+* Fix https://issues.apache.org/bugzilla/show_bug.cgi?id=48214
+ Add logging when a context fails to unregister a JDBC driver. Don't
unregister
+ the jdbc-obdc bridge driver that is loaded by the system classloader.
+ http://svn.apache.org/viewvc?view=revision&revision=885231
+ +1: markt
+ -1:
What is so different with the OBDC bridge? I wonder whether you can
test the class loader of that class to determine if it was loaded by
the System class loader.
Best I can tell, DriverManager treats it as a special case. You should
be able to look at the class loader of the driver. I'll need to do some
testing.
Does it affect classes that are loaded through the Common classloader?
a) if there is a datasource in<GlobalResources>
b) if there is a datasource in<GlobalResources> and another in webapp's context
It shouldn't but depending on configuration, it may be possible to
configure a global resource but end up with a web app triggering the
loading of the driver. That would cause issues and is worthy of further
testing. On the plus-side, if anyone did get hit by that there are easy
work-arounds.
We already applied a lot of last minute patches. Maybe we don't need to
put so much pressure into those more complex leak prevention patches for
6.0.21, and instead let JFC know, when the really important ones are
done and he can proceed tagging etc. Since JFC is now in the boat as 6.0
RM, we might wan to do another release in the not so distant future?
Regards,
Rainer
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org
For additional commands, e-mail: dev-h...@tomcat.apache.org
