Extension in domain specific code generation with meta-model based aspect weaving

Abstract

Domain specific code generation improves software productivity and reliability. However, these advantages are lost if the generated code needs to be manually modified or adapted before deployment. Thus, the systematic extensibility of domain specific code generation becomes increasingly important to ensure that these advantages are maintained. However, the traditional extension approaches, like round-trip engineering, have their limitations in supporting certain code customization scenarios. In this thesis, we address this problem with aspect-oriented techniques. We first show that the meta-model and the code generator can be used to derive a domain specific aspect language whose join points are based on domain specific elements. We then show that a corresponding aspect weaver can be derived as well, provided a proper model tracing facility can be made available for the code generator. We demonstrate the viability of our approach on several concrete domain specific code generation case studies, respectively with the AUTOFILTER code generator, the ANTLR parser generator, and the CUP parser generator. We successfully construct a few Java program analysis tools as a result of these case studies.

More information

Identifiers

Catalogue record

Export record