Aspect oriented programming as a formal framework for runtime monitoring

Aspect Oriented Programming (AOP) is a programming paradigm that aims to increase the modularity of a program by separating cross cutting concerns from the program's main logic. From early on, runtime monitoring has been one of the domains into which AOP has effectively been deployed. Yet to date AOP has not supported the full breadth of needs across the scope of widely varied runtime monitoring applications. This dissertation extends the domain of AOP with new concepts for supporting the breadth of runtime monitoring needs. Two main research issues are addressed. The first issue is the lack of weaving detail AOP currently encompasses, which limits the ways data can be collected for runtime monitoring. The second issue is the fact that existing AOP ideas and frameworks have limited the dimension of weaving to just the source code, excluding many monitoring needs that might not be easily translatable to code-centric instrumentations. This dissertation develops new ideas that extend AOP and solves these issues. These ideas were tested by realizing them in the context of an existing AOP framework. The experimental results showed that the proposed AOP extensions can efficiently and effectively be used for runtime monitoring needs.