From do-not-reply at jboss.com Fri Mar 30 20:44:36 2007
Content-Type: multipart/mixed; boundary="===============2017000335198467812=="
MIME-Version: 1.0
From: shea.phillips <do-not-reply at jboss.com>
To: jboss-user at lists.jboss.org
Subject: [jboss-user] [JBoss Seam] - can't access non-Seam variables from jPDL
 decision condition
Date: Fri, 30 Mar 2007 20:44:36 -0400
Message-ID: <15694439.1175301876724.JavaMail.jboss@colo-br-02.atl.jboss.com>

--===============2017000335198467812==
Content-Type: text/plain; charset="utf-8"
MIME-Version: 1.0
Content-Transfer-Encoding: quoted-printable

Hello,

After significant amount of troubleshooting, I have discovered that the  or=
g.jboss.seam.jbpm.SeamVariableResolver is used when resolving variables in =
a expression inside jBPM jPDL when the process is started inside of Seam.

This is fine and very useful in many cases, however, in my scenario, I am d=
oing the following:

- starting the process from Seam with a single BUSINESS_PROCESS scope Strin=
g var
-  invoking Seam components from embedded jPDL scripts, which set process c=
ontext variables (non-Seam), that are subsequently referenced in condition =
expressions
- the condition expression evaluation uses the SeamVariableResolver and, as=
 a result does not see the variable I  exposed from my script, although oth=
er scripts do (apparently they use some different variable resolution strat=
egy)

Here is my jPDL snippet:


  |  <node name=3D"establishInterpretationProcess">
  |         <transition to=3D"decideOnInterpretation">
  |             <script>
  |                 <expression>
  |                     import org.jboss.seam.Component;
  |                     import com.foo.ECGInterpretationProcessNameResolver=
Service;
  |                     =

  |                     System.out.println("ECG ID is '" + ecgId + "'");   =
                 =

  | =

  |                     lookup =3D Component.getInstance(ECGInterpretationP=
rocessNameResolverService.NAME, true);
  | =

  |                     interpretationName =3D lookup.determineInterpretati=
onProcessName(ecgId);
  | =

  |                     System.out.println("Interpretation name in establis=
hInterpretationProcess is '" + interpretationName + "'");
  |                 </expression>
  |                 <variable name=3D"ecgId" mapped-name=3D"ecgId" access=
=3D"read"/>
  |                 <variable name=3D"interpretationName" mapped-name=3D"in=
terpretationName" access=3D"read,write"/>                =

  |             </script>
  |         </transition>
  |     </node>
  | =

  |     <decision name=3D"decideOnInterpretation">        =

  |         <transition name=3D"interpret" to=3D"doInterpretation"/>       =
     =

  |         <transition name=3D"dontInterpret" to=3D"assignForManualInterpr=
etation">
  |             <condition>#{interpretationName =3D=3D null}</condition>
  |             <script>System.out.println("interpretationName was null!");=
</script>
  |         </transition>
  |     </decision>
  | =


The specific behaviour here is that the expression #{interpretationName =3D=
=3D null} always evaluates to null, so the associated transition is always =
taken, regardless of the value of the interpretationName process context va=
riable (as set in the script)

I tried to outject a value with BUSINESS_PROCESS scope from my component in=
voked in the script, but this also was not visible by the expression.

My design goal is to keep the process and the components loosely coupled (t=
he current bsh notwithstanding :P ) such that either can potentially be reu=
sed, which is  why I am wanting to hold some minimal state in the process c=
ontext.

Does anyone have any suggestions about how I might work around this?  I hav=
e an idea how I might modify the SeamVariableResolver to look in the jBPM c=
ontexts as well as its own scopes, but this seems like a big hammer.

thanks,

shea.

View the original post : http://www.jboss.com/index.html?module=3Dbb&op=3Dv=
iewtopic&p=3D4033401#4033401

Reply to the post : http://www.jboss.com/index.html?module=3Dbb&op=3Dpostin=
g&mode=3Dreply&p=3D4033401

--===============2017000335198467812==--