Here is a suggestion of what may be a better implementation:


Currently, when entering an object (eg model class), all ML does is offer the user a property list.  This is a very complex, hard to use, and fault prone system.

With ML's current implementation I would estimate a 10 to 1 in-efficiency between entering ML (UML) code and generating Modelica code, and just entering Modelica code through MDT.


It is suggested:

The correct ML profile should be attached automatically to a project when it is created, as part of the project creation process.

UML object attributes should be split into two classes, required, optional

When a UML object (eg. model class) is placed on a diagram, the system should automatically prompt for the required parameters, and allow the user as well, to enter optional parameters if necessary.

In the same way that MDT prompts with attribute completion, ML should do so likewise, offering input attribute lists and alternative entries.

Pop up help would be great if the user hovers over a potential entry.

ML should learn from the user's usage, such that when the next object (eg model class) is entered, ML remembers which attributes the user entered, and offers this list first (and of course allows optional additional parameter entry).  ie. ML is an expert or AI system

This learning system is always based on the last entry of a particular UML object.

Of course entry/modification of attributes can still be accessed if necessary through the properties list.

This improvement should make ML much much more usable.


Love and peace,

Sol.