tuscany-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Simon Nash <n...@hursley.ibm.com>
Subject Re: Conversation from a non-SCA application
Date Thu, 31 Jan 2008 21:36:32 GMT

Simon Nash wrote:

> Nishant Joshi wrote:
>> Thanks Simon for your sample....
>> I have tried with your sample, but still i m getting same error...
>> there is only change i have done is in pom.xml... you have used
>> 1.2-incubating-SNAPSHOT and i have 1.1-incubating from 1.1-RC3...
>> Do you have any clue whats the problem ? can you try this sample with
>> 1.1-RC3 ?
>> I guess that there is a problem with jar that i m using(1.1-RC3) and what
>> you have used (1.2-SNAPSHOT)...
>> Please guide me on this regard... ?
> I tried this and I now see the same error that you are seeing.  I am
> extremely surprised as I did not think there was any difference between
> the trunk and the release in this area.  I will investigate further and
> let you know what I find.
OK, I have the rest of the sad story now.  The difference between the
release and the trunk is related to the fix to TUSCANY-1965.  This was
added to the release at the last minute, and I checked in different
code to the release and the trunk.  The trunk version had a more
extensive restructure and included caching, and I did not put this
more complex code change into the release because I thought it would
introduce higher risk at this late stage of closing down the release.

Now for the sad story part.  The more extensive code restructure
fixed a bug (this "/" problem) that I didn't even know was there.
So the "safe" version of the code change for the release had a bug
(actually a bug that was there all along), and the "risky" version
of the code change did not have this bug.

The only workaround for the bug is to go back to using injected
callbacks.  You need to copy the injected callback into a safe
place (a Java assignment is OK) before the second client has made
a forward call to the service method.  It might work even if you
don't make the copy, but it's not a good idea to depend on it.
When you move up to the next release of Tuscany, you can change
your code to do this the correct way.  At some point you will have
to make this change, because OASIS has decided not to allow injected
callbacks for a composite-scoped component.

My apologies for all the difficulties this has caused you.  Thanks
for your patience in helping to debug the problem.


To unsubscribe, e-mail: tuscany-dev-unsubscribe@ws.apache.org
For additional commands, e-mail: tuscany-dev-help@ws.apache.org

View raw message