You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@cxf.apache.org by Peter Jones <pj...@progress.com> on 2009/09/11 21:04:05 UTC
JMSConduitTest failure with ibm jdk
Hi there,
I'm seeing the JMSConduitTest.testJMSMessageMarshal() fail
regularly when run with the ibm jdk 5. It seems that for
some reason the JMSConduit is finalized in the midst of the
jmsTemplate.execute() call in that test. This closes the
connection and the test gets an error like this:
"org.springframework.jms.IllegalStateException: The connection is already closed;"
Adding an assertNotNull(conduit) after the
jmsTemplate.execute() call prevents the JMSConduit from
being finalized before the test method completes, and so
makes the test pass every time. But, it seems odd to me
since the conduit really should still be in scope while
jmsTemplate.execute() is called. Just wondering if anyone
had any thoughts on that.
Anyway, I'll commit that work-around for the ibm jdk as
long as there are no moral objections.
Regards,
Peter
----------------------------------------
Peter Jones
Progress Software
570 Newfoundland Drive
St. John's, NL, Canada A1A 5B1
Tel: +1 (709) 738-3725 ext 321
Internal wiki: http://wiki.progress.com/display/PRODENG/Home
RE: JMSConduitTest failure with ibm jdk
Posted by Peter Jones <pj...@progress.com>.
Hi Christian,
For sure, will do.
Cheers,
Peter
----------------------------------------
Peter Jones
Progress Software
570 Newfoundland Drive
St. John's, NL, Canada A1A 5B1
Tel: +1 (709) 738-3725 ext 321
Internal wiki: http://wiki.progress.com/display/PRODENG/Home
-----Original Message-----
From: Christian Schneider [mailto:chris@die-schneider.net]
Sent: Fri 9/11/2009 6:20 PM
To: dev@cxf.apache.org
Subject: Re: JMSConduitTest failure with ibm jdk
Hi Peter,
that is really strange. I also have no idea why the conduit is finalized.
You should write a good comment in front of the assert to avoid that
someone will later remove it again because he thinks it is unnecessary.
Greetings
Christian
Peter Jones schrieb:
> Hi there,
>
> I'm seeing the JMSConduitTest.testJMSMessageMarshal() fail
> regularly when run with the ibm jdk 5. It seems that for
> some reason the JMSConduit is finalized in the midst of the
> jmsTemplate.execute() call in that test. This closes the
> connection and the test gets an error like this:
>
> "org.springframework.jms.IllegalStateException: The connection is already closed;"
>
> Adding an assertNotNull(conduit) after the
> jmsTemplate.execute() call prevents the JMSConduit from
> being finalized before the test method completes, and so
> makes the test pass every time. But, it seems odd to me
> since the conduit really should still be in scope while
> jmsTemplate.execute() is called. Just wondering if anyone
> had any thoughts on that.
>
> Anyway, I'll commit that work-around for the ibm jdk as
> long as there are no moral objections.
>
> Regards,
> Peter
>
> ----------------------------------------
> Peter Jones
> Progress Software
> 570 Newfoundland Drive
> St. John's, NL, Canada A1A 5B1
> Tel: +1 (709) 738-3725 ext 321
>
> Internal wiki: http://wiki.progress.com/display/PRODENG/Home
>
>
>
>
Re: JMSConduitTest failure with ibm jdk
Posted by Christian Schneider <ch...@die-schneider.net>.
Hi Peter,
that is really strange. I also have no idea why the conduit is finalized.
You should write a good comment in front of the assert to avoid that
someone will later remove it again because he thinks it is unnecessary.
Greetings
Christian
Peter Jones schrieb:
> Hi there,
>
> I'm seeing the JMSConduitTest.testJMSMessageMarshal() fail
> regularly when run with the ibm jdk 5. It seems that for
> some reason the JMSConduit is finalized in the midst of the
> jmsTemplate.execute() call in that test. This closes the
> connection and the test gets an error like this:
>
> "org.springframework.jms.IllegalStateException: The connection is already closed;"
>
> Adding an assertNotNull(conduit) after the
> jmsTemplate.execute() call prevents the JMSConduit from
> being finalized before the test method completes, and so
> makes the test pass every time. But, it seems odd to me
> since the conduit really should still be in scope while
> jmsTemplate.execute() is called. Just wondering if anyone
> had any thoughts on that.
>
> Anyway, I'll commit that work-around for the ibm jdk as
> long as there are no moral objections.
>
> Regards,
> Peter
>
> ----------------------------------------
> Peter Jones
> Progress Software
> 570 Newfoundland Drive
> St. John's, NL, Canada A1A 5B1
> Tel: +1 (709) 738-3725 ext 321
>
> Internal wiki: http://wiki.progress.com/display/PRODENG/Home
>
>
>
>