Using Ada source code generators in a large project

Abstract

In large software projects, especially for heterogeneous systems, good support for parallel Application Function development is essential. For this reason the RNLN/CAWCS has developed a concept in which there is a clear split in system-wide software infrastructure (related to ISO layers 2–6) and multiple, domain specific Application Functions.

The approach taken comprises building information models for the complete hardware architecture, the software architecture, and the system level information (emanating from the Application Function Information Modeling activities). Major standards such as the OSI stack have been adopted. The second step in the process was to build one large database application from these models. Code generators then were produced for the different end products, such as ADA source code and IATEX source files for documentation.

The adopted strategy has already proven itself for CAWCS-built systems, especially in the areas of a rapidly maturing process for the code generator, high quality code and superior flexibility in a System Integration phase. The produced software components are specific in the sense that only the required infrastructure is generated for each Application Function in each computer, instead of one generic general purpose package.