Search

Automatic Program Repair (APR) techniques can promisingly help reducing the cost of debugging. Many relevant APR techniques follow the generate-and-validate approach, that is, the faulty program is iteratively modified with different change operators and then validated with a test suite until a plausible patch is generated. In particular, Kali is a generate-and-validate technique developed to investigate the possibility of generating plausible patches by only removing code. Former studies show that indeed Kali successfully addressed several faults. This paper addresses the case of code-removal patches in automated program repair investigating the reasons and the scenarios that make their creation possible, and the relationship with patches implemented by developers. Our study reveals that code-removal patches are often insufficient to fix bugs, and proposes a comprehensive taxonomy of code-removal patches that provides evidence of the problems that may affect test suites, opening new opportunities for researchers in the field of automatic program repair., New version of the manuscript

Field monitoring techniques can collect data about the behavior of software applications while running in the field, with real users and real data. Developers can exploit the information extracted from the field to timely improve, tune, and fix their systems, anticipating feedback by users. It is, however, challenging to extract a relevant amount of information from field executions without introducing significant overhead. This paper addresses this challenge by studying how to inexpensively trace data in-memory while postponing save operations to idle time, so that the operations requested by users are exposed to a negligible overhead only. In particular, this paper presents delayed saving, a technique that efficiently traces references to objects, opportunistically saving information only when the monitored application is not serving any user request. Storing references and postponing save operations may introduce inaccuracy in the collected data, that is, a lately saved object might be in a different state compared to the state of the object at the time it was traced. Our evaluation shows that the level of inaccuracy introduced by delayed saving is limited compared to its efficiency and low intrusiveness.

Program repair techniques can dramatically reduce the cost of program debugging by automatically generating program fixes. Although program repair has been already successful with several classes of faults, it also turned out to be quite limited in the complexity of the fixes that can be generated. This Ph.D. thesis addresses the problem of cost-effectively generating fixes of higher complexity by investigating how to exploit failure information to directly shape the repair process. In particular, this thesis proposes Failure-Driven Program Repair, which is a novel approach to program repair that exploits its knowledge about both the possible failures and the corresponding repair strategies, to produce highly specialized repair tasks that can effectively generate non-trivial fixes.

In recent years, research on techniques to identify and classify activities of daily living (ADLs) has significantly grown. This is justified by the many application domains that benefit from the application of these techniques, which span from entertainment to health support. Usually, human activities are classified by analyzing signals that have been acquired from sensors. Inertial sensors are the most commonly employed, as they are not intrusive, are generally inexpensive and highly accurate, and are already available to the user because they are mounted on widely used devices such as fitness trackers, smartphones, and smartwatches. To be effective, classification techniques should be tested and trained with datasets of samples. However, the availability of publicly available datasets is limited. This implies that it is difficult to make comparative evaluations of the techniques and, in addition, that researchers are required to waste time developing ad hoc applications to sample and label data to be used for the validation of their technique. The aim of our work is to provide the scientific community with a suite of applications that eases both the acquisition of signals from sensors in a controlled environment and the labeling tasks required when building a dataset. The suite includes two Android applications that are able to adapt to both the running environment and the activities the subject wishes to execute. Because of its simplicity and the accuracy of the labeling process, our suite can increase the number of publicly available datasets.