Modeling collapsed subprocesses


All the samples you have here, all Camunda samples, all Camunda documentation only talks about expanded suprocesses? Also, Camunda cycle doesn’t include referenced subprocesses (always used for collapsed subprocess).

Are you not supporting in any way the hierarchical modeling method suggested by Bruce Silver? I think it is a strategic error to ignore this critical need of modelers.

Thank you,


The camunda Modeler supports expanding and collapsing. Nor sure what cycle may be missing.

Other than that I can only talk about the project here. Sooner or later we will support collapsed sub-processes and the ability to expand / collaps them. Even if there wasn’t Bruce Silver it is still an important feature in complex modeling scenarios.

It is a complex issue on the technical level that has to be thought through and implemented well.

I think you are actually talking about call activities here (in the BPMN sense)?

Thank you, nikku.
I was told Camunda’s strategic modeler will become, so I assumed there is some coordination between you and Camunda. Considering their Eclipse based modeler doesn’t really behave when dealing with collapsed subprocesses, I tried to model using, but you guys have the same limitation. Then, I tried Signavio modeler, which is more mature, being an older modeler. When I export from Signavio to Camunda using their cycle, there is no option to include referenced subprocesses, Camunda confirmed. I wasn’ talking at all about call activity.

Of course whether Bruce Silver says it or not, modeling collapsed subprocesses still has the same relevance. I pointed to him to underscore the importance of having this feature.

Is there any time horizon for adding collapsed subprocess modeling capabilities to

As an idea, you can use a new tab/window to allow modeler to elaborate the subprocess contents, such as Bizagi’s modeler, for example. This way, you don’t disturb the items surrounding the collapsed subprocess by expanding it in place.

Thank you,


Hi Cristian,

Thank you for your question/feedback on this matter. I think I have to ask a little further in order to understand it better.

Let me clarify some aspects first:

  1. is a project that belongs to Camunda, it is just a separate open source project and therefore has its own naming.

  2. The referenced subprocess does not exists anymore in BPMN 2.0. In the older version of BPMN (I think 1.x) there was a concept of embedded and referenced subprocesses. This has been replaced in BPMN 2.0 with subprocesses (always being embedded) and called processes, always being referenced.

Still a subprocess can be expanded or collapsed, which basically means that its content is either always visible or hidden. This is still the case with BPMN 2.0.

As I understand it you are asking for the ability to model/view the content of a collapsed subprocess separately from the rest of the process. This is a perfectly understandable requirement and we are planning to add that in the future. However there is still some work to be done beforehand, like collaborations (pools), lanes, boundary events, etc.

Does that help to clarify your question?


Thank you, Robert.

Yes, I am talking about what you define in the paragraph of your reply starting with “As I understand”.

In my posts I used the term “referenced subprocess” in the context of talking about Signavio modeler, assuming you are familiar with their modeler. I guess my assumption was wrong or I did not explain clearly enough. I did not mean to talk about BPMN 1.X at all. When Signavio modeler starts its exporting function, it allows modeler to include the referenced subprocesses in the exported model. By referenced subprocesses they mean the processes referenced by their like-named property of any collapsed subprocesses that were part of exported process.

My whole point is this: unless a modeler is able to handle decently the case of collapsed subprocesses (meaning: allowing modeling person to provide subprocess contents not as in-place, but out-of-place), that modeler cannot claim it supports the hierarchical method of modeling. And if a process model may not be modeled like this, it means the whole process is a monolith and that is not ideal at all. So my question was: when will your provide this functionality? And I got an answer in your paragraph referenced above, modeler does not yet support it, but I kind of knew that.

I am looking forward to using it once available, that would be really great. It represents the main instrument for handling process complexity, in a top-down modular manner, so that modeling person doesn’t end up with a diagram as big as the Black Forest. Would you like to work on a screen as wide as 12 meters? Kidding…it depends, right?

Thank you,


Your input is very much appreciated, Cristian.

We have quite a long way to go to complete BPMN 2.0 modeling support (kind of what the camunda Modeler provides at the moment). From then on we will focus on advanced topics. Hierarchical modeling is on of them. You said it, it is a deal breaker for a professional modeling tool.

Thanks again,