You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@qpid.apache.org by Martin Ritchie <ri...@apache.org> on 2008/11/05 13:02:51 UTC
Re: svn commit: r711377 - /incubator/qpid/trunk/qpid/java/client/src/main/java/org/apache/qpid/client/AMQConnectionDelegate_0_10.java
Rafi,
Not sure why but our build systems have been failing since this
commit. I'm just writing up some other issues for the Java Client so
if you have a slower single core box to take a look. The quad core box
here seems to pass every so often but the 3.2Ghz P4 fails on every
run. With a bunch of test errors.
Cheers
Martin
2008/11/4 <rh...@apache.org>:
> Author: rhs
> Date: Tue Nov 4 12:13:07 2008
> New Revision: 711377
>
> URL: http://svn.apache.org/viewvc?rev=711377&view=rev
> Log:
> QPID-1430: fixed an NPE on connection close
>
> Modified:
> incubator/qpid/trunk/qpid/java/client/src/main/java/org/apache/qpid/client/AMQConnectionDelegate_0_10.java
>
> Modified: incubator/qpid/trunk/qpid/java/client/src/main/java/org/apache/qpid/client/AMQConnectionDelegate_0_10.java
> URL: http://svn.apache.org/viewvc/incubator/qpid/trunk/qpid/java/client/src/main/java/org/apache/qpid/client/AMQConnectionDelegate_0_10.java?rev=711377&r1=711376&r2=711377&view=diff
> ==============================================================================
> --- incubator/qpid/trunk/qpid/java/client/src/main/java/org/apache/qpid/client/AMQConnectionDelegate_0_10.java (original)
> +++ incubator/qpid/trunk/qpid/java/client/src/main/java/org/apache/qpid/client/AMQConnectionDelegate_0_10.java Tue Nov 4 12:13:07 2008
> @@ -206,7 +206,7 @@
> ConnectionException exc = exception;
> exception = null;
>
> - ConnectionClose close = exc.getClose();
> + ConnectionClose close = (exc == null ? null : exc.getClose());
> if (close == null)
> {
> try
>
>
>
--
Martin Ritchie
Re: svn commit: r711377 - /incubator/qpid/trunk/qpid/java/client/src/main/java/org/apache/qpid/client/AMQConnectionDelegate_0_10.java
Posted by Martin Ritchie <ri...@apache.org>.
2008/11/6 Rafael Schloming <ra...@redhat.com>:
> Martin Ritchie wrote:
>>
>> 2008/11/6 Rafael Schloming <ra...@redhat.com>:
>>>
>>> Martin Ritchie wrote:
>>>>
>>>> Rafi,
>>>>
>>>> Not sure why but our build systems have been failing since this
>>>> commit. I'm just writing up some other issues for the Java Client so
>>>> if you have a slower single core box to take a look. The quad core box
>>>> here seems to pass every so often but the 3.2Ghz P4 fails on every
>>>> run. With a bunch of test errors.
>>>
>>> That's bizarre, the change seems innocuous enough. Can you post the stack
>>> traces/logs?
>>
>> I haven''t had much time to investigate, but hope to get some time
>> tomorrow.
>>
>> The only thing jumps out is that in the tests they are often failing
>> due to lack of a broker. The InVM is mostly ok, (ok locally but CC has
>> errors) but Jva profile fails both localy and on CC box.
>>
>> Do you get a clean result with
>> ant clean build test -Dprofile=java
>> ?
>>
>> Will keep you posted, but agree change seems unlikely probably just a
>> coincidence.
>
> I think it's fixed now. I'll take the fifth on whether or not it was a
> coincidence and instead just say we *really* need the default profile to
> test against all the brokers. ;)
Great thanks, agreed about the profiles.
Perhaps default should be execute a test run for all testprofiles but
if you specify a profile then it only runs that.
That way we can do quick inVM tests whilst developing. The Java and
C++ profiles are so slow to run. Would be good to try and work out a
good way of annotating tests such that;
a) only Tests that actually work over TCP are run with Java/C++ ,
thinking of SimpleACLTest at a present as it ignores the QTC broker
and makes its own InVM to test against. If it configured the external
broker then that would be fine but really not sure if that is
required.
b) only 0-10 tests are run against C++ and 0-8/0-9 against Java. If
they were correctly annotated then we could remove most of the
excludes list.
c) if the calls to getQueue returned a unique queue (perhaps based on
the test name) then restarting the brokers between runs might not be
required. This would especially true if QTC required the tests to
correctly clean up after themselves. It would also give us the
advantage of having a longer test run against a broker. Currently our
testing is very short lived. The reason we have a fresh broker is
because the tests are so poor at clean up the only way to guarantee
there is no cross talk is to restart.
Just some points for thought.
If anyone is interested in this approach then I'll try and write it up
for further discussion about the approach.
Cheers
Martin
> --Rafael
>
>
--
Martin Ritchie
Re: svn commit: r711377 - /incubator/qpid/trunk/qpid/java/client/src/main/java/org/apache/qpid/client/AMQConnectionDelegate_0_10.java
Posted by Rafael Schloming <ra...@redhat.com>.
Martin Ritchie wrote:
> 2008/11/6 Rafael Schloming <ra...@redhat.com>:
>> Martin Ritchie wrote:
>>> Rafi,
>>>
>>> Not sure why but our build systems have been failing since this
>>> commit. I'm just writing up some other issues for the Java Client so
>>> if you have a slower single core box to take a look. The quad core box
>>> here seems to pass every so often but the 3.2Ghz P4 fails on every
>>> run. With a bunch of test errors.
>> That's bizarre, the change seems innocuous enough. Can you post the stack
>> traces/logs?
>
> I haven''t had much time to investigate, but hope to get some time tomorrow.
>
> The only thing jumps out is that in the tests they are often failing
> due to lack of a broker. The InVM is mostly ok, (ok locally but CC has
> errors) but Jva profile fails both localy and on CC box.
>
> Do you get a clean result with
> ant clean build test -Dprofile=java
> ?
>
> Will keep you posted, but agree change seems unlikely probably just a
> coincidence.
I think it's fixed now. I'll take the fifth on whether or not it was a
coincidence and instead just say we *really* need the default profile to
test against all the brokers. ;)
--Rafael
Re: svn commit: r711377 - /incubator/qpid/trunk/qpid/java/client/src/main/java/org/apache/qpid/client/AMQConnectionDelegate_0_10.java
Posted by Martin Ritchie <ri...@apache.org>.
2008/11/6 Rafael Schloming <ra...@redhat.com>:
> Martin Ritchie wrote:
>>
>> Rafi,
>>
>> Not sure why but our build systems have been failing since this
>> commit. I'm just writing up some other issues for the Java Client so
>> if you have a slower single core box to take a look. The quad core box
>> here seems to pass every so often but the 3.2Ghz P4 fails on every
>> run. With a bunch of test errors.
>
> That's bizarre, the change seems innocuous enough. Can you post the stack
> traces/logs?
I haven''t had much time to investigate, but hope to get some time tomorrow.
The only thing jumps out is that in the tests they are often failing
due to lack of a broker. The InVM is mostly ok, (ok locally but CC has
errors) but Jva profile fails both localy and on CC box.
Do you get a clean result with
ant clean build test -Dprofile=java
?
Will keep you posted, but agree change seems unlikely probably just a
coincidence.
Cheers
Martin
> --Rafael
>
--
Martin Ritchie
Re: svn commit: r711377 - /incubator/qpid/trunk/qpid/java/client/src/main/java/org/apache/qpid/client/AMQConnectionDelegate_0_10.java
Posted by Rafael Schloming <ra...@redhat.com>.
Martin Ritchie wrote:
> Rafi,
>
> Not sure why but our build systems have been failing since this
> commit. I'm just writing up some other issues for the Java Client so
> if you have a slower single core box to take a look. The quad core box
> here seems to pass every so often but the 3.2Ghz P4 fails on every
> run. With a bunch of test errors.
That's bizarre, the change seems innocuous enough. Can you post the
stack traces/logs?
--Rafael