Scope of our paper is the software engineering process of a special class of important software systems, namely large, heterogeneous, distributed information systems based on several underlying autonomous databases and applications. Increasingly, these systems are subject to integration and interoperability requirements using middleware (e.g. CORBA) and/or the Internet as integration platforms. One major characteristics in this scenario is the premise that no development will start from scratch. Always a huge amount of existing legacy data and code will have to be included, yielding our notion of 'continuous software engineering' which will be explained in the paper.Within the software process, we claim a consistent development of appropriate meta data structures for two different purposes: firstly, on the information level a methodology for the development of meta information components in order to support an efficient information retrieval process over such distributed heterogeneous environments, but, secondly, also the software engineering process itself to be improved by corresponding meta data structures and repositories, to be materialized using CORBAs meta object facility.