While the open source environment Modelio allows to develop extensions in the form of so-called 'modules' written in java, it does not provide support to develop non-trivial extensions in jython/python. This is exactly what the PyModelio framework is for: PyModelio allows to develop in jython what we call 'plugins' .

PyModelio is based on of the best features of Modelio, the jython interpreter. With modelio alone, one can write simple macros in jython. While this is handy for very simple task, macro development could rapidly becomes difficult when complexity raises. PyModelio aims to solve this and make it possible to offer the developer wishing to entend modelio to choose between:

• developing 'modules' in java,
• developing 'plugins' in jython.

The term "plugin" (which is not a modelio concept) is used here to refer to consistent set of jython scripts with possibly additional resources. While scripts are installed as "macros" in modelio some other are called by these macros. While macros are usually in a single file, we use here full fledged modular development as allowed by jython/python including the possibility to use arbitrary java libraries.

The alternative to develop what we called "plugin" in jython, is to develop modelio modules in java. This requires much more effort though.

Modelio provides a lot of features for the development of plugin in java but no facilities is given for developing in python beyond simple monolithic macro development.

This PyModelio framework supports the following features:

• support for code development/execution outside modelio workspace macro directory.
• support for 'ala' OCL operations thanks to PyAlaOCL python package.
• automatic instrumentation/extension of modelio metamodel with operations such as those to support AlaOCL.
• support so that the developer can themselves instrument modelio metamodel for adding additional methods computing derived information for instance.
• modular development of python modules with reloading features in development mode.
• inclusion of regular python and java libraries thanks to python/java path management.
• automatic definition of a global environment for jython interpreter, with globals symbols for project profiles.
• support for directory management allowing to develop code independently from execution.
• finding the local plugins directory where more features can be installed.
• extension of the python path and java path to reuse existing pure python or java libraries.
• access to modelio global variables (selection, selectedElements, modelingSession from inside modules.
• set the working directory to a local directory so that macros can read and write in a well defined place. When using Modelio the working directory is set to the Modelio installation directory. This is definitively not a good place to write files.
• allow the definition of modelio styles in plugins and automatically register these styles.