Community
Participate
Working Groups
Our current metamodel has the structure of the XMI metamodel provided by the BPMN2 specification, enriched with ExtendedMetadata where a direct mapping to the XML schema is feasible. However, there are some places where the metamodel significantly differs between XMI and XSD as far as serialization is concerned. I think there are two main areas: 1. extension mechanisms of the specification, and 2. attribute in XMI, but mixed content in XSD used to store some information. I opened this bug to provide an overview of the differences and will create separate dependent bugs for the individual areas. I think that we should let users work with the metamodel as it appears in the specification, that is, the API should stay as it is now. Behind the scenes, we should strive to produce correct XML and XMI serialization. Generally, I can imagine different levels of support: 1) serialized models generated by our implementation ought to be valid 2) read valid XML models (produced by other tools) 3) process all elements of valid models and expose their content to clients 4) keep content during editing 5) switch between XML and XMI seamlessly 6) allow editing and creation of special elements