[jboss-user] [jBPM] - Re: Human task as service in jbpm5.2

jemmerling do-not-reply at jboss.com
Mon Jan 16 13:26:00 EST 2012

jemmerling [https://community.jboss.org/people/jemmerling] created the discussion

"Re: Human task as service in jbpm5.2"

To view the discussion, visit: https://community.jboss.org/message/647036#647036

Well, I decided to take a closer look at this WAR module. As previously stated, I had inferred its functionality from its name. On close inspection, I can't say that my previous inference was necessarily correct.

First, I have confirmed that, for the demo set-up, the Demo Task Server acts as the Human Task Service for both the Eclipse Tooling and the GWT Console demo.

However, given the name of this server, it is strongly implied that its use should be restricted to a limited situation such as the demo. So what is not clear is, what does jBoss recommend as an "enterprise class" implementation of the human task service?

So (as I said) I took a closer look at the WAR, what I see is there is one Servlet class that rejects GET and POST requests, when it initializes it executes more or less the same logic as the Demo Task Server. To wit, it kicks off a MinaTaskServer as a separate thread running in the same JVM as the servlet. So there are really just 2 differences I can discern between this and the demo server:

1.) It executes within a JEE container.
2.) It uses a JEE data source rather than a JDBC connection.

Of course I would consider point 2.) above as an absolute minimum requirement for an enterprise solution.

However, what I would think of as the "elephant in the living room" is that I don't think this is what the OP had in mind. My suspicion is that the OP was looking for an implementation of WS-HumanTask exposed via JAX-WS over SOAP (or similar words to that effect). I can find no evidence that anything like this is included in the jBPM 5.2 distribution.

Of course, some of us have been led to believe that it is something of a sin to create one's own threads within any JEE container. Perhaps that is less of a problem in jBoss than other platforms. A question (I honestly don't know the answer off the top of my head): Suppose I deploy this WAR to jBoss, then using jBoss' management console, I should be able to stop and restart the associated webapp. In that case, will the separate thread that was kicked off when the servlet starts up also terminate, or will it live on? If the foregoing is not a problem in jBoss, is it possibly still a problem with some other AS?

The other issue has to do with MINA and HornetQ. I had not been exposed to these technologies before evaluating jBPM. So the WS-HumanTask specification seems describe a command pattern, and it doesn't seem strictly necessary to expose this over SOAP or REST. So when using either the MINA or HornetQ implementation of the Human Task Service, I suppose it is reasonable to expect the developer not to care about protocol, you simply plug it in and it "just works" (unless it doesn't). Is that the case?

Thank you for your forebearance,


Reply to this message by going to Community

Start a new discussion in jBPM at Community

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.jboss.org/pipermail/jboss-user/attachments/20120116/d4acadf3/attachment.html 

More information about the jboss-user mailing list