On Dec 17, 2010, at 1:14 PM, Alex Dean wrote:
>
> On Dec 16, 2010, at 5:33 PM, Alex Dean wrote:
>
>>
>> I'll get a useful test written up tomorrow morning, and create a new JIRA
>> issue if Jeff doesn't beat me to it.
>
> I've created https://issues.apache.org/jira/browse/AMQ-3094 for this problem.
>
> My testing so far is inconsistent. If I publish 5 messages quickly via
> JavaScript, then a separate JavaScript client sometimes receives 1 of the
> messages, and sometimes two of them.
>
> If I publish 3 messages in a JUnit test, my http client can either receive 0
> messages or all 3 messages. That's after re-running the same tests multiple
> times, with no code changes inbetween test runs. Sometimes I get messages,
> and sometimes I don't.
>
> JavaScript & Java test files & sample output are attached to the ticket. I
> don't feel like I'm any closer to a solution, but hopefully this helps others
> reproduce the issue.
I think the issue may be that AjaxListener's constructor does not set an
initial value for lastAccess. By setting "this.lastAccess =
System.currentTimeMillis();" in the constructor, my test always pass. I want
to do more testing, since I don't have very good test coverage yet, but this
seems promising.
One question for the core guys... Why does the code always surround debugging
log statements with an if()?
if (LOG.isDebugEnabled()) {
LOG.debug( "blah" );
}
Is it OK to just call "LOG.debug()" without the if()?
alex
