Defining a modelling language is a big task, which requires careful thinking at different levels of abstraction. We can easily end up in endless discussions about details, unless we make sure that we first have a shared view of the big picture.
My suggestion is to arrange our topics into an hierarchy, like this:
- Pragmatic level
- Conceptual level
- Syntactic and semantic levels
- Lexical level
As agreed with Bob (and Karim too), I have written up a summary of dataflowml, my framework for implementing Dynaplan Smia. Here is a link to the document: http://www.myrtveit.com/papers/A_framew ... models.pdf. It will give you an idea how I have worked with the different levels of the language definition, and it also contains many suggestions and recommendations that I think can be worth taking into account in our work.
I have included my definition of "core system dynamics", which is my first attempt to pick the concepts and features that I believe belong to the standard we are working on.
The document also contains a specification of an equations format (model listing), which seems to be missing in the current XMILE specification.
Comments and corrections are welcome. I hope you enjoy the reading.