[dward]

Developer forum thread started here: "Filtering Context Properties".

[djharte]

Thanks very much for the update on this issue. Look forward to trying out the solution

[dward]

Hi David (H),

I want to test your app with my fix, and I have it all deployed, but honestly I'm not a REST guy.  Can you give me an example raw HTTP request that would be made to the example REST binding?

Thanks,

David (W)

[dward]

Nevermind, I got it recreated. Now to drop in my fix and re-test.

[dward]

Okay, please check out the attached file.  Notes:

1. My serialization fix worked.  I've issued a pull request on the associated jira, and Keith said he would be pushing it later today.  So, this means you will have to wait for a nightly build of 0.8.0-SNAPSHOT.  Additionally...
2. The binding.remote element got replaced with the new binding.sca element. (Thanks Keith!)
3. Once I got past the serialization issue, I did hit another intersesting issue with your app.  Because both deployed applications ("example" and "profitability") shared the same namespace ("urn:...resteasy-binding:1.0"), a strange deployment-ordering problem surfaced.  Basically, if you deploy profitibility, then deploy example, then test, you're fine.  But then if you re-deploy profitability without also re-deploying profitability, then an error occurs "cannot find operation getProfit on ProfitModel".  Keith can give a better explanation of why this happens than I can.  However, I then changed the two applications to have different namespaces (as well as adding the targetNamspace to "profitability" from "example"), then I could hot-redeploy either application in any order with no errors.  I think Keith said with the old binding.remote, they had to be in the same namespace.  With the new binding.sca, they should be in different namespaces (and then the binding.sca needs to add a targetNamespace attribute as previously described).
4. I did enable MessageTracing in both switchyard xml files.
5. I did fix a spelling mistake in a couple places ("positioning" was mispelled).

[djharte]

Hi David,

Thanks very much for the fix and detailed explanation of the problem and solution.

Unfortunately though I have been unable to get my environment updated with the fix. In summary, I updated my environment as follows:

• project's pom files refer to Switchyard version 0.8.0-SNAPSHOT.
• In order to update my Jboss AS's Switchyard module I leveraged the latest installer available under: http://ci.jboss.org/jenkins/job/SwitchYard-Release/lastSuccessfulBuild/, The logs state: "Building SwitchYard: Distribution 0.8.0-SNAPSHOT"

The updated project didn't work out of the box (the profit model wasn't being called) and I noticed that the namespace was incorrect and didn't reflect the reference specified in the switchyard.xml file I then updated the namespace and the remote service was called but I ended up with the same problem as before, leaving me to believe I don't have the updated code.

Also, I notice that the ticket is still in an 'Unresolved' state: https://issues.jboss.org/browse/SWITCHYARD-1325.

Where have I gone wrong?

Thanks,

David

[kcbabo]

The problem lies with me - I didn't get a chance to push David's change on Friday so the nightly you used didn't have the fix.  I pushed everything upstream just now and kicked a Jenkins release build, so build #768 should have it.  It will probably take a few hours for the build to sync with the external Hudson server on jboss.org, so keep that in mind if you don't see it right away.

[djharte]

Thanks for confirming, will give it a whirl tomorrow so.

[djharte]

I just tried the latest build (#767) and it doesn't appear to have the changes applied; I still can't get the fix to work. You mentioned that it should be available in build #768 but I don't see this build, did it fail? Can you double check the status please and let me know.

Thanks,

David

[kcbabo]

Unfortunately, it looks like there is a problem with the synchronization between our internal build servers and our external Hudson instance.  The latest build on the internal server is #771. I have spoken to the people that manage that environment and it looks like it will be addressed in the next maintenance cycle.  In the meantime, you can always build a release by doing the following:

> git clone git://github.com/jboss-switchyard/release.git

> cd release

> mvn -U -DskipTests clean install

Skipping the test should speed up the release build considerably.  First time could be a bit slow depending on what's in your local Maven repository, but subsequent builds should be 2-3 minutes without tests, IIRC.  Once the build completes, you can find the distribution in:

release/jboss-as7/standalone/dist/target/switchyard-as7.zip

Not as easy as downloading from our server, but it does get you the latest and greatest on demand. :-)  Sorry about the Jenkins issues, but there a bit beyond my control at this point.

[kcbabo]

Quick update on the nightly build status.  I just heard from the team that manages publication of builds to hudson.jboss.org and the synchronization problem should be fixed this weekend. 

Another recent development is that a clustering demo has been added for 0.8.  It can be found here:

https://github.com/jboss-switchyard/quickstarts/commit/faa3366a60f6cd63b2ecf2c1908dd3f6e5dd4c23

If everything goes as planned, you should be able to pick up a nightly on Monday that has the fix you need for your application and the new clustering quickstart.

[kcbabo]

FYI - Nightly builds have recovered and are available on the public Hudson server:

https://hudson.jboss.org/hudson/job/SwitchYard-Release/