Research On Aspect Weaving For Aspect-oriented Programming

The rapid development of the computer industry makes programming idea has undergone lots of changes. With the complexity and size of software system increasing, the relationships between various modules of software system become more complex. Traditional software development method will bring code tangling, code scattered and many other issues, this has led to the challenges of software customization and reusability. Aspect-oriented programming came into being under this circumstance.At present, aspect-oriented programming (AOP) technology has begun to gradually be used in various software systems development. From the AOP perspective, the software system is the realization of a series of concerns. Concerns can be divided into two categories according to their characteristics, one is the core concerns which achieve the main functions and objectives of the software; the other is the cross-cutting concerns which have crosscutting effect on the core concerns. AOP allows the programmer to achieve the core and crosscutting concerns respectively and then weave them together. So, we can use aspect-oriented programming to reduce duplication of system code, the coupling between modules and improve the system maintainability.Weaving efficiency of aspect-oriented programming is a key factor affecting performance. This paper mainly studied several weaving issues based on deep understand of aspect-oriented techniques and their implementation technology. Based on current research results that raise weave-on-demand, weave history mechanisms and weaving approach implementation based on Log,this paper presented a log-based incremental weaving mechanism implementation. This method gives an aspect storage structure, based on this structure we can quantitatively describe current weave state, compute the change of weave state, and achieve incremental weaving to improve the efficiency of weaving. At the same time we designed a more efficient log storage structure for the recovery of aspect weaving state, and then propose a more efficient recovery mechanism for aspect weaving state.Finally, we made three sets of comparative experiments. Through comparing weaving time of aspect-oriented programming and compile-time of object-oriented programming, we verified the weaving time of aspect-oriented programming is just slightly larger than the compile time of object-oriented programming, so performance gap is not significant. Through log-based recovery weaving state and completely re-weaving without log-based recovery analogy, we verified the log-based recovery can greatly improve weaving efficiency when the system collapsed. At the same time we tested the run time whether using incremental weaving mechanism or not to verify the log-based incremental weaving mechanism can greatly improve the weaving efficiency.