Thoughts on updating the modeler

Just kicking off some discussion on who can contribute and how to creating a new modeler app…

We think using the Oryx core libraries still make sense, but again open to any input on better options that are completely unrestricted.


Hello, bpmn-js proved itself as a extensible library for bpmn, cmmn and dmn modeling. Most users doesn’t care about bpmn-io logo in the right-bottom corner.

We are going to start a new project with flowable and bpmn-js with modeling tool based on bpmn-js.

IMHO, Camunda bad with provided api, but good at tooling (modeler, unit testing libraries). Flowable good with api, architecture, but very poor with modeler.

If Camunda will provide better architecture and api, our company will turn back to Camunda.

1 Like

Building on the suggested features listed here: Flowable Modeler vs Activiti Modeling App

Suggested Features

  • Open Source License: Apache License 2.0
  • Embeddable: Developers would like to embed the ‘Modeler’ in their applications
  • Extensible via Plugins

I aggree with @maratkalibek. We also use flowable as our bpmn engine but use for providing our own modelling solution to our customers.

So something similar like would be great - but starting a similar community project feels cumbersome. Would there be a way to participate in for flowable?

There’s a complete blocker for using bpmn io because it’s not Apache licensed. Everything in the Flowable project is unemcumbered Apache 2.0 and it’s not acceptable to us or many other Flowable users to have restrictions on use.

Consequently, we have no choice. We think we can do a lot based on the Oryx libraries we currently use. We’re still interested in any other input :slight_smile:


(OP for the Flowable Modeler vs Activiti Modeling App post that kicked this off)

Having scoured the interwebs and found nothing better, we’re considering either sticking with KIS BPM (Yuck!) or the same approach as @donkon : extending and using bpmn-js. The main part missing from the bpmn-js modeler is property support for activiti: prefixed properties. There’s a camunda extension in github but no activiti equivalent.
@donkon - would you consider opensourcing yours?

On the license issue raised by @PHH : Makes sense, but is a damn shame, since bpmn-js extensions would be such a simple approach. It might be worth trying to a) negotiate those license terms with Camunda (since it is mostly MIT except for the logo) or starting with diagram-js which is MIT. (search for license questions on the bmpn io forum)


One last point: We started this journey with the intention of integrating the current Flowable Modeler.
Even though it is Angular 1, so not simply embeddable in our Angular 8 app, we could have treated it like a separate static single-page-editor, which is what we did with the KIS editor.

But the flowable modeler blocks that by requiring 2 separate web apps (backends) to manage the loading, saving and property-editing of processes. With KIS this was all in one page - including the properties (with script editors and form assignment tables and all).

A small effort to enable that level of separation in the current Flowable Modeler would help a lot

@chrisw @donkon

The Flowable Modeler’s index.html:


<script src="editor-app/editor/oryx.debug.js" type="text/javascript"></script>


Is oryx.js comparable to bpmn-js?

What about making oryx.js available as a Web Component and therefore usable by any framework - Angular, React, Vue, …

And, to take advantage of the “shadow” DOM so that you can keep the element’s features private, so they can be scripted and styled without the fear of collision with other parts of the document.

For example, the Material Web Components (MWC) are a collection of Web Components maintained by Google that implement Material Design. And, there’s an Angular wrapper for the MWC.

We’re going to do some experimentation with the oryx core, as we think that is a richer framework to start from. It may take a bit of time before we have some conclusion there, but it’s where we’ve decided to begin.


I find this repo is great.

But not support all the flowable properties, in short it is a good example

1 Like


How is the research on oryx core going ? Is it still the plan to continue using it for the flowable modeler ?



It’s going ahead but as a background activity - the approach looks promising but we’ve a lot going on so no timeline quite yet.


Any new status about updating the modeler ? Kind regards.

As I just said, it’s moving forward, but we’ve got lots of other things going on as well.


Sorry Paul, I misinterpreted the date of your previous answer.

What about mxgraph it is licensed under the Apache 2.0 license:

Yes, it’s interesting. However, we think we can get a long way stripping back the current libraries and keep a lot of the bpmn/cmmn pieces. It’s just going to take time between other efforts.

Thanks for the pointer.