Language for modeling Harel StateCharts scxml

While you may surely define a modeling language based on the XML schema (by mapping schema elements to different elements of GOPPRR types) your modeling language does not need to be developed based on the expected output. Instead you may have totally other kind of language constructs and then use the generator to produce the desirable output, like the one that follows the XML schema mentioned above.

 

I would instead start to look what kind of phone applications I would be developing: a browser, particular type of game, enterprise app (e.g. http://www.metacase.com/cases/ucssm/), or own administrative software (e.g. sending electricity consumption bill, remote controlling heating in a house as in http://www.metacase.com/cases/phone_example.html) as they advocate different kind of language concepts. If we look the problem this way, then we don’t even need to focus on one particular platform (like Android): from higher-level models we can target different platforms (output formats). For example, the S60Phone example in the evaluation version that generates Python could already now be extended with another generator that produces state machine parts as given in the XML schema. However, if we can fix the target platform, like Android/XML schema, the development of the language becomes easier.

 

Still your question remains: how we can find good language concepts or at least candidate concepts if we don’t take them from the XML schema (or other expected output). Based on our experiences there are different kind of sources for the language concepts and I would suggest checking the SPLC article (available at http://users.jyu.fi/~jpt/TolKelSPLC2005.pdf) as we reviewed 23 cases and identified different ways to develop the language. One way was following the output (or XML schema) but there are also others. We extended the analysis later in the book on Domain-Specific Modeling (see www.dsmbook.com).