Support for extensibility and customization in architecture description languages (ADLs) is essential for expressing varying stakeholder concerns. Several ADLs provide this support, to a certain extent. However, they all have limitations, such as the type of possible extensions and tool support for extending the language features.

Malavolta provides a concise overview of: requirements for next-generation extensible and customizable ADLs, points where the requirements fall short, and an envisioned approach for fulfilling all the requirements with an architectural framework that utilizes model-driven engineering techniques.

The content of the paper is relevant to language designers and researchers interested in software architecture documentation and analysis. Although some initial results and contributions are reported, the framework is currently under development; as such, it is not yet mature enough to be useful to software architects.

The approach relies mainly on model-driven engineering techniques as an enabling technology. In future work, Malavolta should consider the utilization of other technologies and methodologies. For example, the approach could inspire and borrow concepts from software product line engineering, in order to exploit common language features and facilitate extensive reuse. As another example, aspect-oriented software development techniques could be used for handling crosscutting concerns, views, and extensions.