previous_group previous up next next_group
Previous: 3. Implementation-related Problems Up: 2. HQ Interface Next: 3. Related Work


6. Conclusion

In this section, we have analyzed in details some important abstract operators, and proposed an interface called HQ. It reveals lots of compatibility problems among the four existing interfaces used in static analyzers. Compatibility problems have several origins:

Our work aims at unifying those interfaces, which requires some adaptations since simple wrapping is not enough. Then wrapping layers for the common interface, through which existing libraries can be reused efficiently, will be implemented.

The most visible application of our work is that an analyzer can use, through a wrapping API, the polyhedral domain, which is more precise, and the octagonal domain, which is less expensive. Thus we can adapt the behavior of the analyzer in static analyses and transformation of programs. Then, the impacts of using different abstract domains can be observed. One of the perspectives of this common interface is to be able to automatically change from one domain to another in an intelligent way, in order to obtain the best compromise between precision and speed.

The beginning of the APRON project, described in the next section, as well as the workshops VMCAI 2005 and NSAD 2005, do not have great impact on our HQ interface, since the workshops did not focus on the problem, and the APRON project has a different point of view to ours.


previous_group previous up next next_group
Previous: 3. Implementation-related Problems Up: 2. HQ Interface Next: 3. Related Work
Nguyen Que Duong
2006-09-16