| Author |
Message
|
| devin |
 Posted: Thu May 26, 2005 10:26 am Post subject: Modeler Validation Error |
 |
|
Voyager
Joined: 27 Oct 2003
Posts: 82
|
I am seeing this error when trying to do MQ Workflow Validation in WBI Workbench.
Error is as follows
" Invalid Connections
Error Message
'ActivityA' is connected more than once to 'ActivityB' through different conditions."
ActivityA is indeed connected to ActivityB two times thru different conditions. But why is that an error ??
Interesting thing is the process model validates fine without any errors when I replaces the ActivityB with a subprocess 'ProcessC'
Has anybody seen this error before?Any ideas?
Anyhelp would be great.
Thanks
_________________
IBM Certified Solution Designer -
WebSphere MQ Workflow 3.4 |
|
| Back to top |
|
 |
| jmac |
| Posted: Thu May 26, 2005 11:16 am Post subject: |
|
|
Jedi Knight
Joined: 27 Jun 2001
Posts: 3081
Location: EmeriCon, LLC
|
I've not seen that in the past.
If you want to zip up the org and proc files and send to me I will take a look.
Being a Buildtime user more than Modeler, I can tell you that Buildtime will not allow two control connections from Activity A to Activity B. So I would assume Modeler should do the same.
_________________
John McDonald
RETIRED |
|
| Back to top |
|
|
| mqmhr |
| Posted: Thu May 26, 2005 7:57 pm Post subject: |
|
|
Centurion
Joined: 28 Dec 2004
Posts: 105
|
I have seen this validation error occurr in Workbench and this is in fact due to more than one control flow paths between two tasks. For e.g. in case you have modeled something like this
Task1 ---> Phi ---> Task2
Task1 ---> Decision1 (Yes) ---> Task2
MQ Workflow validation of this would say "Task1 is connected more than once to Task 2 through different conditions".
Strangely enough, even if there is only data flow between Task 1 and Task2, MQ Workflow validation fails. But I am not able to understand that bit. Do send me too the model if possible. I would be able to provide some inputs. |
|
| Back to top |
|
|
| jmac |
| Posted: Fri May 27, 2005 5:19 am Post subject: |
|
|
Jedi Knight
Joined: 27 Jun 2001
Posts: 3081
Location: EmeriCon, LLC
|
It sounds to me like this should be opened as a PMR... I suspect you will get back WAD, but it never hurts to inquire, especially about modeler.
_________________
John McDonald
RETIRED |
|
| Back to top |
|
|
| mlf |
| Posted: Fri May 27, 2005 6:59 am Post subject: |
|
|
Newbie
Joined: 27 May 2005
Posts: 2
|
I don't think this is an error. In this case of:
Task1 ---> Phi ---> Task2
Task1 ---> Decision1 (Yes) ---> Task2
When the process is translated to FDL, you'd have 2 control connectros from Task1 to Task2 (one of which has a transition condition). However, this is not allowed in FDL. If you try to build it in Buildtime, you'll get an error too.
If this is something you absolutely need to do, then you can use a NOOP activity:
Task1 ---> Phi ---> NOOP ---> Phi ---> Task2
Task1 ---> Decision1 (Yes) ---> Task2
Now it will pass validation. |
|
| Back to top |
|
|
| mqmhr |
| Posted: Mon May 30, 2005 6:01 am Post subject: |
|
|
Centurion
Joined: 28 Dec 2004
Posts: 105
|
In Modeler, Task1 ---> Phi ---> Task2 can mean two things.
If there is a sold line between the tasks and the Phi, it means both control and data are transferred from Task1 to Task2. If there is a dotted line, it means only data is transferred between the tasks.
Now, Workflow validation failure is expected in the first case as there will be two control connectors from Task1 to Task2 and indeed happens so. But how do you explain Modeler reporting a Workflow validation error when only data is passed from Task 1 to Task 2 through the Phi, in other words, there is only one control path between these two tasks? |
|
| Back to top |
|
|
| devin |
| Posted: Tue May 31, 2005 7:30 am Post subject: |
|
|
Voyager
Joined: 27 Oct 2003
Posts: 82
|
Thanku all very much for your inputs.
there are infact two control paths from Task A to Task B in my models.
I replaced the control paths to just one by mixing the decision choices and it is working fine now.
But I dont understand clearly why Buildtime restricts this.
And also the error doesnt happen when there are two control paths from Task A with subprocess Proc B.
Any thoughts on this ?
_________________
IBM Certified Solution Designer -
WebSphere MQ Workflow 3.4 |
|
| Back to top |
|
|
| jmac |
| Posted: Tue May 31, 2005 7:46 am Post subject: |
|
|
Jedi Knight
Joined: 27 Jun 2001
Posts: 3081
Location: EmeriCon, LLC
|
Devin:
This has always been the rule since the FlowMark days (i.e only one control connection between any 2 tasks) so I have to admit I am having a hard time seeing why you would need more that one connection.
Also, I can not duplicate what I understand you to be saying here:
| devin wrote: |
| And also the error doesnt happen when there are two control paths from Task A with subprocess Proc B. |
When I try this I get the same duplicate connector message, (as I think I should). I am running this test on MQWF BT for 3.4.0.5
_________________
John McDonald
RETIRED |
|
| Back to top |
|
|
| devin |
| Posted: Tue May 31, 2005 7:53 am Post subject: |
|
|
Voyager
Joined: 27 Oct 2003
Posts: 82
|
Well, I was using more than one connection to make the model look more clear and also instead of using a big expression in a single choice , I spread it across mutiple decisions and choices.
Anyway, I ran my test on WBI Workbench 4.2.4.
It might be that Workbench doesnt realize about the error.
Thanks for your help.
_________________
IBM Certified Solution Designer -
WebSphere MQ Workflow 3.4 |
|
| Back to top |
|
|
| jmac |
| Posted: Tue May 31, 2005 7:58 am Post subject: |
|
|
Jedi Knight
Joined: 27 Jun 2001
Posts: 3081
Location: EmeriCon, LLC
|
| devin wrote: |
Well, I was using more than one connection to make the model look more clear and also instead of using a big expression in a single choice , I spread it across mutiple decisions and choices.
|
This is what I expected your answer to be.... Thanks for confirming.
_________________
John McDonald
RETIRED |
|
| Back to top |
|
|
|
|
