How to invoke signal or message ?

[shashank-yelluri]

Hello @ralphhanna,
Good Evening !!

We are using your bpmn engine for solving the workflows. Now, i have one problem where i would need to offer the token to three different user tasks. For which, i have created an inclusive gateway. After any one user task gets executed or say 2 user tasks are executed, i want to trigger a message or signal to remove the created token for the user task 3. Is there any way or possibility for achieving this ? Or solving this problem in a better way ?

I've gone through the code and our engine is only having .start() .invoke() & .assign(). I could not find any other functionalities.

Your response would be highly appreciated !!
Thank You.

[shashank-yelluri]

@ALL - feel free to pitch in and help me out on how to resolve this. Any help would be highly appreciated !!

[ralphhanna]

Hello @shashank-yelluri ,
There are several ways to achieve this using various bpmn features without calling the engine.

1. Use Boundary Event to cancel, please see #https://bpmnserver.github.io/docs/examples/boundary-events/
2. Use Conditional flow #https://bpmnserver.github.io/docs/examples/conditionalFlow/
3. Use Event Based Gateway #https://bpmnserver.github.io/docs/examples/ebg/

Also please note the difference between inclusive and exclusive gateways #https://bpmnserver.github.io/docs/examples/gateways/

If you share with me your model and tell me more, I can surely help.
Thanks

[shashank-yelluri]

@ralphhanna - such a great reply and i feel you already pointed me to the right gateway. I was unaware of the event based gateway and may be this would solve my problem. Let me try to build the model using the same and update you acc. for your help. But thanks again for the very quick help.

[shashank-yelluri]

Ok. I've tried using the event gateway but my tokens are stuck with the "Receive Task". Ideally token shouldn't be waiting at this task right ? I'm sharing the model i've configured to test.
This is my requirement,
I have two user tasks offered to two different groups (GroupA & GroupB). If any 2 members approve from GroupA then i can consider the workflow to be closed OR if one person approves from GroupB then again it is closed. For that i've made the first user task to be "Multi Instance" by passing the collection as 2.

I thought of going with parallel gateway to solve the problem. But in this way, the token is already there with both the user tasks. Now i wanted a way to drop the already created token if any one approves from GroupA (at user task 1) OR if one person approves from GroupB(at user task 2) then again i need to drop the token offered at the GroupA.

Now after your help, i've tried configuring the "Event Based Gateway". I could see two problems,

1. I was not getting an option to link the gateway to user task directly.
2. Token is stuck with the "Message Task" only
3. What should i do to make the token to move forward and offer at user task ?

Usually we are starting the engine and after that using the .invoke by reading the last item. In this case, i'm not getting a better / efficient way to solve the problem Or i might be thinking and configuring my workflow as wrong. Please guide me as i'm new to the realm of this bpmn engine.

Attaching the model below,

Thank You

[shashank-yelluri]

In your given example, it was straight forward that after the configured time for the "Boundary Event", the token will eventually move forward to the user task and wait for the invoke. But let's say, i configure the boundary event to run after 1 hr. Now when will we go to the routes of 8,9,11 ?

what was the purpose of them ?

[shashank-yelluri]

Hello @ralphhanna,
Good Morning !!

I've ran into another issue where i'm trying to configure the workflow for the above mentioned problem. Which gateway to use, because both Exclusive & parallel are not fitting in solving my requirement.

1. Parallel is waiting for all the tokens to reach
2. XOR is taking one and moving into next (but i had the user task with multiple instance)

Any help would be highly appreciated !!
Thank You

[shashank-yelluri]

@ralphhanna - tried with the standard loop as well with the shared script at the docs. But the code is breaking as token is not having the loop defined in it.

Also, for every invoke for the standard token, we are creating two tokens. Any reason for that ? Ideally it should only be one token right ?

Thank You

[ralphhanna]

Please verify that the code is attached to a loop node.
Please update to latest release 2.2.16 just released.

[ralphhanna]

Hello @shashank-yelluri ,

Based on your requirements the attached model does the trick

Users Vote.bpmn.txt

Please update to release 2.2.15
sh npm update bpmn-server

[ralphhanna]

sorry, for late response, forgot to hit comment button

[shashank-yelluri]

Not a problem :)

[shashank-yelluri]

@ralphhanna - even after the update, it is not working :(

[shashank-yelluri]

Also, could you please explain what is this ?

if (data.inEscalation!==true) {
data.inEscalation=true;
return {escalation:'escalation code'};
}

I do not see a connection to the attached boundaryEvent. How does the bpmn understanding to escalate and cut out the process ? Because i need to build my workflow further by understanding this.

[shashank-yelluri]

@ralphhanna - I've updated the camunda:collection with some number and it worked 🥳
Now, my last and major problem is,

Let's say some one executed the task at the UserA, now the token offered at UserB should be removed. I've opened the discussion for the same and learnt a lot in the process. If there a way to do that, please help me out. Also please help with my previous comment on the inEscalation logic at the listener.

Thank You So Much. Saved my lot of time.

[ralphhanna]

It seems just calling return {escalation:'escalation code'}; causes a problem with recursive calls.
So avoid it I make sure it is only called once. This is being fixed in the new release 2.2.17

[shashank-yelluri]

@ralphhanna - sure. But how can i have my token deleted ? (which are already offered to multiple other tasks as i've configured inclusive gateway)

[ralphhanna]

Release 2.2.17 is now ready

Users Vote.bpmn.txt
and script is much simpler:

return {escalation:'escalation code'};
Also notice that Other Tasks have been cancelled: (grayed out in diagram and have no end date)

[shashank-yelluri]

@ralphhanna - Thank You for the efforts. But can i have the same escalation attached to the user task ? and not to the sub process ? Because i really have a complex workflow where it involves 6*6 user tasks and any one approval at say, row 1, then i need to delete the tokens or take the route of the row and ignore all the other rows. This rows i've configured as a parallel task. Like in our above example, the moment we execute the user task A once, then token should be removed at the Task B. How can we achieve this ?

[ralphhanna]

You can attach Escalation to any Task, however, Escalation will cancel the task and take the alternate path of event

when using loops you should place the loop inside a subprocess and attach the escalation

As you can see in the example above, First Escalation of Task 1 has no utility, since the task will end anyway.

In the second example, Escalation is can be called from any Task 3 instance, it will cancel all instances and subsequent flow inside the subprocess.

We simply use subprocess to define the scope of cancellation.

[shashank-yelluri]

Also do we support rest of the fields present in the "Multi-instance" section ?

[shashank-yelluri]

@ralphhanna - I've tried this as well,

$(item.data.rules[0].subRules[0].minimumApprovers)

Strange thing is, whenever i provide the dynamic thing to candidateGroups alone it is working. But the moment i try to assign some variable to the "camunda:collection", the code is breaking and blaming "candidateGroups".

error in script evaluation
var item=this;
var data=this.data;
var input=this.input;
var output=this.output;
var appDelegate=this.token.execution.appDelegate;
var appServices=this.token.execution.servicesProvider;
var appUtils=appDelegate.appUtils;

            return (data.rules[0].subRules[0].group);

TypeError: Cannot read properties of undefined (reading '0')

[ralphhanna]

Please submit your bpmn file for me to examine.
Most likely the problem is the scope of data see https://bpmnserver.github.io/docs/data/#instancedata-scope

[shashank-yelluri]

Here is the file,

matrix.txt

I'm stitching the rules to the data at the service task and expecting it to flow and have the multi collection value number.

[shashank-yelluri]

And do we have any documentation around this ? For dynamic values at multiple places,

1. For sequence, data.something is working
2. For candidate groups, $data.something is working
3. For service task, some $() is working
4. Now for camunda:collection, how should we pass ?

Do we have anything for reference ?

[ralphhanna]

As for loop fields are:

Only collection is used.

[shashank-yelluri]

Ha true true. But just curios to know whether we use any other fields ?

[shashank-yelluri]

Or we support any other ?

[ralphhanna]

Examples of valid collection:

$(item.data.records)

(item.data.records)

item.data.records

[1,2]

Same for Condition

[shashank-yelluri]

Sharing you the configuration,

[shashank-yelluri]

I've tried with this.data as well. Even in this case, my candidateGroups value is empty. Is there any issue that candidateGroups cannot be passed dynamic if multiple instance is enabled to the user task ? Because both of them are individually dynamic values are working fine.

[shashank-yelluri]

This is the exact thing i've tried for the candidateGroups -> $(this.data.candidateGroups)

[shashank-yelluri]

@ralphhanna - attaching the file again for your reference, please try to remove the one dynamic value from either loop and candidateGroups and try, the file will work on the other hand, having them both will not take the dynamic value from the data to the candidateGroups.

matrix.txt

[shashank-yelluri]

Also attaching the rules for your reference,

[ralphhanna]

for more complex scenarios you can use a method like this:

$appServices.getCollection(this)

and appServices will have a function:

    async getCollection(token) {
        console.log('get collection',token.currentNode.id);
        return ['A','B','C'];

    }

[shashank-yelluri]

@ralphhanna - apologies for the delay. I work in IST timezone.

[ralphhanna]

will investigate

[ralphhanna]

Here are my findings:
Works for me when I set the data at start event trigger

data.rules=[
	{subRules:[
		{group:'A',minimumApprovers:2}
		]
	}
	];

data.candidateGroups=['GroupA','GroupB'];

and Remove input/output from the Shashank task

[shashank-yelluri]

Oh nice. But i get the rules as an output of the service task :(
Why was it not working when we take the rules as an output from the service task ?

[shashank-yelluri]

But what is wrong with my file configuration ? Are we not allowed to set the data in that way ?

[ralphhanna]

sorry, here is the problem:

Each Instance of the loop has own token with own data, but the data.candidateGroups is declared at the parent token.
Bit confusing I know, can you recommend a better solution, would appreciate it

[shashank-yelluri]

Bit unclear, now is this the solution ? Also, then how come this was working well with setting the camunda:collection value dynamically ? Also, i was able to do the candidateGroups value also to be dynamic, but both of them together is having a problem, is it ? :(

[shashank-yelluri]

Also just curious to know, how did you crack this ? this.token.parentToken.data.candidateGroups ? This would help me to find better ways to the problem in future 😛

[shashank-yelluri]

@ralphhanna - anyways, thanks a lot for your efforts 🙌 . Anyways, i'll try with the above solution by reading the parent token. But now, let's say, i have a next user task also having a multi instance, then should i just do, this.token.parentToken.parentToken ?

But, if you can explain me on what is this, this.token thing ? How can i try things out on my own to debug things in future. If you can help me with thins thing, that would really help 🙏

[ralphhanna]

I simply used vs code to debug the app, and noticed that at the item level data was empty.
One way to work around this annoying problem is to inherit data from parent token, will try this in upcoming release.

[shashank-yelluri]

@ralphhanna - Perfect 🙌
Thank you for all the help. Let me try the above solution and update you shortly.

[shashank-yelluri]

Hello @ralphhanna,
I've tried one scenario today and the outer sub process multi instance collection was not working. Could you please verify the image and the shared bpmn file and let me know why after the parallel, the sequence is not working ?
Thank you in advance.

matrix.txt

[ralphhanna]

This should be fixed in new Release 2.2.18