Re: [ecd-pmc] Resolve CQ 15939

[Oliver Kopp <kopp.dev@xxxxxxxxx>]

Hi,

I EMO-IP from the CC until we have a result. Is that OK with you guys? :)

> Following this strategy we would like to include the Flowable Modeler with
> as less modifications as possible. It depends for its end from the Oryx
> Editor [3] which has been forked from the Google Code repository to GitHub
> and the license is changed to MIT.

(The most updated mirror is
https://github.com/zeisschoice/oryx-editor, which did not modify
anything, but shows the license switch. There are active developments
going on, such as Gemsbok (https://github.com/j5int/Gemsbok), but
still relying on Ext JS)

> @Oliver, who is one of the contributors of Oryx project, can share more
> information about the project status especially on the dependencies section
> e.g. ExtJS usage, etc. Also it will be great to know whether there are other
> alternatives for BPMN modelling in the browser that we can use in Dirigible.

The code of Oryx heavily relies on Ext JS 2.0.2. There are 421 calls
to "Ext.". It is a fundamental part of Oryx, so replacing is not easy.

When preparing Winery for Eclipse in 2012 all our graphical modeling
code was based on Oryx. Cloud Computing Topologies and Management
plans. We evaluated alternatives
(https://stackoverflow.com/a/15117083/873282):

- Oryx: unmaintained
- bpmn.io - looking similar to Oryx, maintained, but non-free license,
because it has to be linked back to "bpmn.io"
(https://github.com/bpmn-io/bpmn-js/blob/master/LICENSE)
- Eclipse Stardust - now archived
- own implemention using jsPlumb

We managed to rewrite the topology modeler part using jsplumb
(https://github.com/eclipse/winery/tree/master/org.eclipse.winery.topologymodeler;
part of the initial code contribution;
https://www.eclipse.org/proposals/soa.winery/) and are now working on
an Angular port (nearly finished -->
https://github.com/eclipse/winery/pull/196)

Regarding the BPMN part, we have a very restricted subset of BPMN
working. The required subset has "scientifically" be developed:
https://link.springer.com/chapter/10.1007/978-3-642-33155-8_4 (full
PDF: http://www.iaas.uni-stuttgart.de/RUS-data/INPROC-2012-23%20-%20BPMN4TOSCA.pdf).
The paper shows the Oryx-based prototype :). The prototype was
extended (http://dx.doi.org/10.18419/opus-3497)

 ZTE (from China) reimplemented it using jsplumb and Angular
(https://github.com/eclipse/winery/tree/master/org.eclipse.winery.workflowmodeler).

This is a good start for a BPMN modeler, but far away from being
complete. Oh, and we use JSON and not the standard XML format. We
convert it to WS-BPEL for execution.

Which elements do you need? Maybe, investing energy in that code would
help both projects?


Stepping back one more step:

So, is Flowable a *must* or do is it possible to use something else?

- Eclipse Stardust is a solid BPMN engine, which is currently
archived, but maybe we can revive it? It was started in Germany and I
know the lead architect.
- We have a working toolchain of BPMN modeling, converting to BPEL and
executing it on Apache ODE. Would that be another possibility?
- We worked on integrating our modeler with the Camunda BPMN Engine.
- In Eclipse.SOA, there are more workflow projects:
https://projects.eclipse.org/projects/soa: Eclipse Java Workflow
Tooling (JWT) is the only actively maintained.

We would be happy if we could join forces in bringing one of the other
options forward. We are most experienced in Apache ODE. - For a
complete description of our system: http://www.opentosca.org/

Cheers,

Oliver