Machine learning methods have achieved great success in many software engineering tasks. However, as a data-driven paradigm, how would the data quality impact the effectiveness of these methods remains largely unexplored. In this paper, we explore this problem under the context of just-in-time obsolete comment detection. Specifically, we first conduct data cleaning on the existing benchmark dataset, and empirically observe that with only 0.22% label corrections and even 15.0% fewer data, the existing obsolete comment detection approaches can achieve up to 10.7% relative accuracy improvement. To further mitigate the data quality issues, we propose an adversarial learning framework to simultaneously estimate the data quality and make the final predictions. Experimental evaluations show that this adversarial learning framework can further improve the relative accuracy by up to 18.1% compared to the state-of-the-art method. Although our current results are from the obsolete comment detection problem, we believe that the proposed two-phase solution, which handles the data quality issues through both the data aspect and the algorithm aspect, is also generalizable and applicable to other machine learning based software engineering tasks.

Code snippets are prevalent on websites such as Stack Overflow and are effective in demonstrating API usages concisely. However they are usually difficult to be used directly because most code snippets not only are syntactically incomplete but also lack dependency information, and thus do not compile. For example, Java snippets usually do not have import statements or required library names; only 6.88% of Java snippets on Stack Overflow include import statements necessary for compilation.

Generating natural language messages for source code changes is an essential task in software development and maintenance. Existing solutions mainly treat a piece of code difference as natural language, and adopt seq2seq learning to translate it into a commit message. The basic assumption of such solutions lies in the naturalness hypothesis, i.e., source code written by programming languages is to some extent similar to natural language text. However, compared with natural language, source code also bears syntactic regularities. In this paper, we propose to simultaneously model the naturalness and syntactic regularities of source code changes for commit message generation. Specifically, to model syntactic regularities, we first enlarge the input with additional context information, i.e., the code statements that have dependency with the variables in the code difference, and then extract the paths in the corresponding ASTs. Moreover, to better model code difference, we align the two versions of code before and after the committed code change at token level, and annotate their differences with fine-grained edit operations. The context and difference are simultaneously encoded in a learning framework to generate the commit messages. We collected from GitHub a large dataset containing 480 Java projects with over 160k commits, and the experimental results demonstrate the effectiveness of the proposed approach.

Real-world e-commerce systems need large cache capacities. Persistent memory (PM) can be employed to enlarge JVMs’ cache capacities, meanwhile they incur heavy write slowdowns and garbage collection overheads. This paper proposes JPDheap, a JVM heap design for PM-DRAM memories. A JPDheap is composed of a standard Java heap on DRAM and another heap on PM. The core insight is to separate heap objects and store them on DRAM or PM, allowing objects to be accessed much more efficiently. Our evaluation shows that JPDheap outperforms state-of-the-art heap designs by up to 115.96% in increasing applications’ throughput and by up to 87.03% in decreasing the average latency.

There is an increasing demand for evolving software systems to deliver continuous services of no restart. Dynamic software update (DSU) aims to achieve this goal by patching the system state on the fly but is currently hindered from practice due to non-trivial cross-version object state transformations. This paper revisits this problem through an in-depth empirical study of over 190 class changes from Tomcat 8. The study produced an important finding that most non-trivial object state transformers can be constructed by reassembling existing old/new version code snippets. This paper presents a domain-specific language and an efficient algorithm for synthesizing non-trivial object transformers over code reuse. We experimentally evaluated our tool implementation PASTA with real-world software systems, reporting PASTA's effectiveness in succeeding in 7.5X non-trivial object transformation tasks compared with the best existing DSU techniques.

Java virtual machine (JVM) has the well-known slow startup and warmup issues. This is because the JVM needs to dynamically create many runtime data before reaching peak performance, including class metadata, method profile data, and just-in-time (JIT) compiled native code, for each run of even the same application. Many techniques are then proposed to reuse and share these runtime data across different runs. For example, Class Data Sharing (CDS) and Ahead-of-time (AOT) compilation aim to save and share class metadata and compiled native code, respectively. Unfortunately, these techniques are developed independently and cannot leverage the ability of each other well. This paper presents an approach that systematically reuses JVM runtime data to accelerate application startup and warmup. We first propose and implement JWarmup, a technique that can record and reuse JIT compilation data (e.g., compiled methods and their profile data). Then, we feed JIT compilation data to the AOT compiler to perform profile-guided optimization (PGO). We also integrate existing CDS and AOT techniques to further optimize application startup. Evaluation on real-world applications shows that our approach can bring a 41.35% improvement to the application startup. Moreover, our approach can trigger JIT compilation in advance and reduce CPU load at peak time.

Recommending suitable tags for online textual content is a key building block for better content organization and consumption. In this paper, we identify three pillars that impact the accuracy of tag recommendation: (1) sequential text modeling meaning that the intrinsic sequential ordering as well as different areas of text might have an important implication on the corresponding tag(s) , (2) tag correlation meaning that the tags for a certain piece of textual content are often semantically correlated with each other, and (3) content-tag overlapping meaning that the vocabularies of content and tags are overlapped. However, none of the existing methods consider all these three aspects, leading to a suboptimal tag recommendation. In this paper, we propose an integral model to encode all the three aspects in a coherent encoder-decoder framework. In particular, (1) the encoder models the semantics of the textual content via Recurrent Neural Networks with the attention mechanism, (2) the decoder tackles the tag correlation with a prediction path, and (3) a shared embedding layer and an indicator function across encoder-decoder address the content-tag overlapping. Experimental results on three realworld datasets demonstrate that the proposed method significantly outperforms the existing methods in terms of recommendation accuracy.

This paper introduces a new, fully automated modelbased approach for effective testing of Android apps. Different from existing model-based approaches that guide testing with a static GUI model (i.e., the model does not evolve its abstraction during testing, and is thus often imprecise), our approach dynamically optimizes the model by leveraging the runtime information during testing. This capability of model evolution significantly improves model precision, and thus dramatically enhances the testing effectiveness compared to existing approaches, which our evaluation confirms.We have realized our technique in a practical tool, APE. On 15 large, widely-used apps from the Google Play Store, APE outperforms the state-of-the-art Android GUI testing tools in terms of both testing coverage and the number of detected unique crashes. To further demonstrate APE's effectiveness and usability, we conduct another evaluation of APE on 1,316 popular apps, where it found 537 unique crashes. Out of the 38 reported crashes, 13 have been fixed and 5 have been confirmed.

Commit messages, which summarize the source code changes in natural language, are essential for program comprehension and software evolution understanding. Unfortunately, due to the lack of direct motivation, commit messages are sometimes neglected by developers, making it necessary to automatically generate such messages. State-of-the-art adopts learning based approaches such as neural machine translation models for the commit message generation problem. However, they tend to ignore the code structure information and suffer from the out-of-vocabulary issue. In this paper, we propose CoDiSum to address the above two limitations. In particular, we first extract both code structure and code semantics from the source code changes, and then jointly model these two sources of information so as to better learn the representations of the code changes. Moreover, we augment the model with copying mechanism to further mitigate the out-of-vocabulary issue. Experimental evaluations on real data demonstrate that the proposed approach significantly outperforms the state-of-the-art in terms of accurately generating the commit messages.

A typical generate-and-validate automatic program repair (APR) tool needs to repeatedly run the same test suite to validate each generated patch. This procedure is expensive when the number of patches is huge. Additionally, to scale to large programs, a program repair tool has to consider a small patch space in practice and thus may sacrifice the capability to find potential correct repairs. In this work, we propose to speed up automatic program repair to mitigate the above issues. One the one hand, we found that restarting processes to load patched code consumes the majority of total validation time. This problem is even severe when the program is running in a managed runtime such as Java virtual machine (JVM). On the other hand, dynamic software updating (DSU) can load and execute new code without restarting. To this end, we propose to use DSU techniques to speed up automatic program repair and present an empirical study in this paper. Within our study, DSU can bring up to 66.3 times speedup in comparison with the traditional restart approach. However, DSU may not be able to handle all patches and can also incur unknown side effects that lead to inconsistent validation results. We then further study the feasibility and consistency of applying DSU to speed up APR. Our results show that 1) less than 1% patches cannot be dynamically updated using the builtin DSU ability of JVM, and 2) DSU based validation leads to potentially harmful inconsistency in only 16 of 1,897,518 patches.

Ensuring the quality of mobile applications (apps) needs to explore the GUI thoroughly. In practice, exhaustively exploring every GUI widget is unscalable on large real-world apps since it usually suffers from the problem of widgets explosion. To mitigate the problem, many existing testing tools usually detect and group homogeneous widgets heuristicly with different level of model abstraction since these widgets behave the same. However, no heuristic always works well. Heterogeneous widgets with divergent behaviors can be mistakenly grouped, which largely limits the testing effectiveness. This paper proposes a technique to effective GUI testing of Android apps with dynamic feedback-directed widgets grouping. Initially, we group the widgets according to the structure of the GUI. During testing, we observe behaviors of widgets in a group and regroup improperly-grouped widgets dynamically. Then, we apply a feedback-directed strategy to effectively accelerate the GUI exploration. The proposed technique is implemented as a practical tool for Android apps, named WGDroid. We evaluated WGDroid on 17 widely-used Android apps and compared it with the state-of-the-art GUI testing tools, i.e., AimDroid, SAPIENZ, and Monkey on both emulators and real devices. WGDroid outperformed the three tools in all testing coverages and also detected the most unique crashes. In particular, WGDroid discovered 208 more activities on 12 large benchmark apps on real devices and 11 more activities on another 5 benchmark apps on emulators, than the best of the other tools. These results show that WGDroid can significantly accelerate the GUI exploration.

Dynamic software updating (DSU) is a technique to upgrade a running software system on the fly without stopping the system. During updating, the runtime state of the modified components of the system needs to be properly transformed into a new state, so that the modified components can still correctly interact with the rest of the system. However, the transformation is non-trivial to realize due to the gap between the low-level implementations of two versions of a program. This paper presents AOTES, a novel approach to automating object transformations for dynamic updating of Java programs. AOTES bridges the gap by abstracting the old state of an object to a history of method invocations, and re-invoking the new version of all methods in the history to get the desired new state. AOTES requires no instrumentation to record any data and thus has no overhead during normal execution. We propose and implement a novel technique that can synthesize an equivalent history of method invocations based on the current object state only. We evaluated AOTES on software updates taken from Apache Commons Collections, Tomcat, FTP Server and SSHD Server. Experimental results show that AOTES successfully handled 51 of 61 object transformations of 21 updated classes, while two state-of-the-art approaches only handled 11 and 6 of 61, respectively.

Given a program P that exhibits a certain property Ψ (e.g., a C program that crashes GCC when it is being compiled), the goal of program reduction is to minimize P to a smaller variant P′ that still exhibits the same property, i.e., Ψ(P′). Program reduction is important and widely demanded for testing and debugging. For example, all compiler/interpreter development projects need effective program reduction to minimize failure-inducing test programs to ease debugging. However, state-of-the-art program reduction techniques --- notably Delta Debugging (DD), Hierarchical Delta Debugging (HDD), and C-Reduce --- do not perform well in terms of speed (reduction time) and quality (size of reduced programs), or are highly customized for certain languages and thus lack generality.

Activities are the fundamental components of Android applications (apps). However, existing approaches to automated testing for Android apps cannot effectively manage the transitions between activities, e.g., too rarely or too often. Besides, some techniques need to repeatedly restart from scratch and revisit every intermediate activity to reach a specific one, which leads to unnecessarily long transitions and wasted time. To address these problems, we propose AimDroid, a practical model-based approach to automated testing for Android apps that aims to manage the exploration of activities and meantime minimize unnecessary transitions between them. Specifically, AimDroid applies an activity-insulated multi-level strategy during testing and replaying. It systematically discovers unexplored activities and then intensively exploits every discovered individual with a reinforcement learning guided random algorithm. We conduct comprehensive experiments on 50 popular closed-source commercial apps that in total have billions of daily usages in China. The results demonstrate that AimDroid outperforms both Sapienz and Monkey in activity, method and instruction coverage, respectively. In addition, AimDroid also reports more crashes than the other two.

Dynamic software updating (DSU) is a technique that can update running software systems without stopping them. Most existing approaches require programmer participation to guarantee the correctness of dynamic updating. However, manually preparing dynamic updating is error-prone and time-consuming. Therefore, other approaches prefer to aggressively perform updating without programmer intervention, which may definitely lead to unanticipated runtime errors. To reduce human effort and enhance the reliability for dynamic updating, we leverage automatic runtime recovery (ARR) techniques to recover runtime errors caused by improper dynamic updating. This paper presents ADSU, a fully automatic DSU system using ARR. We evaluate ADSU with real updates from widely used open source software systems, i.e., Apache Tomcat, Apache FTP Server and jEdit. The preliminary results have shown that ADSU succeeds in automatically applying 11 of 16 real-world updates that existing counterparts cannot.

Smartphones are indispensable in people's daily lives. As smartphone apps are being increasingly concurrent, developers are increasingly unable to tackle the complexity and to avoid subtle concurrency bugs. To better address this issue, we propose a novel approach to manifesting concurrency bugs in Android apps based on the fact that one can simultaneously generate input events and their schedules for an app. We conduct static-dynamic hybrid analysis to find potentially conflicting resource accesses in an app. The app is then automatically pressure-tested by guided event and schedule generation. We implemented the prototype tool AATT and evaluated it over thirteen popular real-world open-source apps. AATT successfully found 9 concurrency bugs out of which 7 were previously unknown.

Dynamic software updating (DSU) aims to patch software for fixing bugs or adding functions while it is running. Before update, developers need to make a dynamic patch ready, which includes update points, state transformers and a corresponding code patch. Existing practice mostly assumes manual preparation of dynamic patches, but this process can be both time-consuming and error-prone. Some pioneer work attempts to automate this process, but cannot guarantee the generation of safe dynamic patches for most updates. This paper presents a novel approach CURE to automatically generating safe dynamic patches. CURE takes two versions of software and their test cases as input, and automatically synthesizes state transformers and selects update points. We applied CURE to 28 updates for three real-world server software. The experimental results show that CURE generated safe dynamic patches automatically and their corresponding updates achieved an 88.7% success rate, as compared to 74.3% for TOS and 61.2% for default patches.

Software systems are often subject to unexpected runtime errors. Automatic runtime recovery (ARR) techniques aim at recovering them from erroneous states and maintaining them functional in the field. This paper proposes Ares , a novel, practical approach to performing ARR. Our key insight is to leverage a system's already built-in error handling support to recover from unexpected errors. To this end, we synthesize error handlers via two methods: error transformation and early return. We also equip Ares with a lightweight in-vivo testing infrastructure to select the right synthesis methods and avoid potentially dangerous error handlers. Unlike existing ARR techniques based on heavyweight mechanisms (e.g., checkpoint-restart and runtime monitoring), our approach expands the intrinsic capability of runtime error resilience already existing in software systems to handle unexpected errors. Ares's lightweight mechanism makes it practical and easy to be integrated into production environments. We have implemented Ares on top of both the Java HotSpot VM and Android ART, and applied it to 52 real-world bugs. The results are promising — Ares successfully recovers from 39 of them and incurs low overhead.

Deterministic replay tools help programmers debug concurrent programs. However, for long-running programs, a replay tool may generate huge log of shared memory access dependences. In this paper, we present CARE, an application-level deterministic record and replay technique to reduce the log size. The key idea of CARE is logging read-write dependences only at per-thread value prediction cache misses. This strategy records only a subset of all exact read-write dependences, and reduces synchronizations protecting memory reads in the instrumented code. Realizing that such record strategy provides only value-deterministic replay, CARE also adopts variable grouping and action prioritization heuristics to synthesize sequentially consistent executions at replay in linear time. We implemented CARE in Java and experimentally evaluated it with recognized benchmarks. Results showed that CARE successfully resolved all missing read-write dependences, producing sequentially consistent replay for all benchmarks. CARE exhibited 1.7--40X (median 3.4X) smaller runtime overhead, and 1.1--309X (median 7.0X) smaller log size against state-of-the-art technique LEAP.

Practical software systems are subject to frequent updates for fixing their bugs or addressing new requirements. Updating a software system without stopping and restarting it is desired, as this helps reduce the redeployment cost as well as achieving the high availability. Existing techniques for dynamically updating Java programs may introduce noticeable pauses during which these programs are unable to function. We in this paper present Javelus, a dynamic Java update system with greatly reduced pausing time but without sacrificing update flexibility and system efficiency. Different from previous approaches, Javelus uses a lazy update mechanism with which an object-to-update will not be updated until it is really used. We implemented Javelus on top of an industry-strength OpenJDK HotSpot VM. We evaluated Javelus with real updates to Tomcat 7 and the same micro array benchmark used in evaluating Jvolve and DCE VM. The experiments report promising results that Javelus only incurred a pausing time two orders of magnitude smaller than those of Jvolve and DCE VM.

Self-adaptive systems are attractive due to their ability of adapting to changeable environments automatically. However, such systems may be subject to runtime failures when all environmental dynamics cannot be adequately considered at design time. When such failures occur at runtime, a system's internal adaptation logic usually has become inconsistent with its environment, according to our observation. We call this inconsistency sync-loss error. From our project experiences, we empirically identified a strong correlation between sync-loss error and system failure. This motivated us to fix sync-loss error in order to reduce failure for self-adaptive systems. In this paper, we formulate the problem of detecting sync-loss error, and present a framework ReSync to automatically fix sync-loss errors by desynchronizing a system with its environment. We experimentally evaluated ReSync on real robot cars with 20 different system versions. The evaluation reported promising results that ReSync can automatically recover our robot car systems from sync-loss errors, and significantly reduce the failure rate from 90.9% to 11.7-28.8%.