Model-driven model repository with textual save and load operations

Abstract

The article defines the concept of model-driven software, that is beyond the mass of application data, the software model of this data is also stored, modified, saved and/or loaded. This part of the software is called the model level, and the corresponding component is called the model repository. The idea, that the model repository itself should be also designed in a model driven way, has similar advantages, like ordinary model-driven software has. That is, its model level can be modified in run-time, and the architecture enables a greater flexibility and customizability, than the monolithic way of software development. The lowest three of the four layer metamodel structure of UML is straightforward –the application data, its software model and the metamodel. The point of the fourth layer can be just understood when we want to create a model-driven model-repository. The model level of such a software is the metamodel itself, that means the meta-metamodel is the layer that has to be programmed in, hard-coded and unchangeable. The article analyses the structure of such software, and as an example presents two model-driven operations: the textual loading and saving of models.