ONE task output is required when trying to setup actions on outputs of multiple tasks

  • 3
  • Idea
  • Updated 8 months ago
  • Not Planned
  • (Edited)
I was working with a services architect and we tried creating a rule that had two dataset tasks that fell under the same general umbrella of what I wanted this rule to do - even though they were two disparate sets of data.   I didn't need to merge the data, I just needed the two sets to be available. 

When I got to the Setup Action screen, I got an error message:  "It is mandatory to have only one output task to execute actions. Ensure to use Merge task, if there are more than one output tasks."

The documentation isn't clear, but engineering told me that you have to only have ONE task output in order to create actions.  But as long as you have only ONE task output - you can still select any of the dataset tasks upon which to perform an action.  Which seems like you're forced to take an extra step of creating a merge, when you don't really need the data to be merged to accomplish your goal.

Incidentally, if you have two datasets + 1 merge, and two more datasets + 1 merge, you then have to merge the two merges.   THEN you can select any of the dataset tasks or merge tasks and perform an action off of them. 

As long as there is only one ultimate output from all the Tasks, you can use any task to deliver an action.

REQUEST #1:  Can the documentation be updated to more clearly explain this requirement?

QUESTION:  Is there a reason why that single output requirement is necessary?  

I will add that when I have only two dataset tasks (which would result in two outputs) I get the error message cited above BUT I can still select one of those two tasks, create an action, and attempt to run the rule (which subsequently fails immediately because of the dual task outputs).

REQUEST #2:  If the single output requirement remains a requirement, then you should probably restrict someone from being able to do what I did above.  If there are multiple outputs from the Tasks, then I shouldn't be able to select a task upon which to create an action.
Photo of Jeff Kirkpatrick

Jeff Kirkpatrick, Champion

  • 26,636 Points 20k badge 2x thumb

Posted 10 months ago

  • 3
Photo of Marcelo

Marcelo, Champion

  • 3,540 Points 3k badge 2x thumb
@Jeff: if you do a dummy merge (to avoid the error), can you still execute actions on the individual data sets without using the dummy merge? Being that the functionality of picking the dataset you want to act on is new, I'm wondering if the requirement for the merge is something that is dragged from the past release and not removed.
Photo of Jeff Kirkpatrick

Jeff Kirkpatrick, Champion

  • 26,636 Points 20k badge 2x thumb
Hey Marcelo! Yes you can still execute actions on the individual data sets without using the dummy merge - but the dummy merge is still required.

Engineering said that was intentional to prevent people from using a single rule to perform actions on unrelated data sets, assumedly because it could create complexity.  I get what they are saying - but in my use case they were two datasets but they were being used to accomplish a similar goal.  (GS basically said those should be split out into two rules.) That use case is probably not common but I do think it can come up from time to time.  

I didn't get that requirement from reading the docs, so hopefully this will help someone else from encountering the same.
Photo of Sai Ram Pulluri

Sai Ram Pulluri, Official Rep

  • 13,270 Points 10k badge 2x thumb
Thanks Jeff, we will update the document.