Modeler Performance

We are trying to create Camunda DMN using Modeler for the following scenario
We have a questionnaire with different questions. So with respect to answers to questions, our DMN calculate the score of each question and returns as LIST. It worked well with few questions but cant able to load in Modeler to edit or add more input entries/columns Is there any simple method to add more entries as inputs in “Camunda” engine “DMN” file dynamically?
Any help should be appreciated

What do you mean by “add more entries as inputs”? Can you model a small sample and upload it (or its screens exported as PNG) to highlight your issue?

There are some issues with DMN performance. As far as I know our modeler team is currently working on fixing those, so it should be fixed with some future release.

As for dynamic creation. You can always create XML file dynamically using some programming language and xml libraries.

Thanks, @mariusz.sielski and @cmensch for responding. I can explain further
I am trying to create a different questionnaire with a different set of questions.Please see sample in screen and DMN attached. Now I added three columns as input.But in my actual scenario, we have hundreds and more columns. so we cant create it in modeler and it hangs with no further processing. We chose these kind of implementation because we can send the inputs while submit the questionnare Task.

score.dmn (2.9 KB)

If you want hundreds of columns and just can’t add these in the modeler: check the XML and create the DMN via XML directly.

If you have multiple questionnaire then it is likely best to use DRD features by placing each questionnaire in a single DMN and calculate the score for it directly (you’d likely use COLLECT -> SUM which will be shown as “C+”) and use the results from all DMNs as input for a new DMN that again uses COLLECT -> SUM and just has input Q1 -> Output ${Q1}, input Q2 -> Output ${Q2}, etc. leading to this “final” DMN providing a single result with the calculated points.
You can group various DMNs to sub-groups to provide both visual groupings (mainly in DRD) and to keep the number of input columns low.

In the engine you still have all the input variables set (which could come from User-Tasks (one for each questionnaire) providing direct input possibility) and reference only the final “sum DMN” - the necessary input DMNs (including their own input DMNs if you group these more) are called directly by the engine.

1 Like

@geo We’re working on improving the Camunda Modeler to handle the case of hundreds of columns and / or rows. See related bug ticket.