Large Language Models (LLMs) demonstrate capabilities in code generation, potentially boosting developer productivity. However, their adoption remains limited by high computational costs, among other factors. Small Language Models (SLMs) present a lightweight alternative. While LLMs have been evaluated on competitive programming tasks, prior work often emphasizes metrics like Elo or pass rates, neglecting failure analysis. The potential of SLMs in this space remains underexplored. In this study, we benchmark three open SLMs—Llama-3.2-3B, Gemma-3-12B, and Phi-4-14B—across 280 Codeforces problems spanning Elo ratings from 800 to 2100 and covering 36 distinct topics. All models were tasked with generating Python solutions. Phi-4-14B achieved the best SLM performance with a pass@3 of 63.6%, nearing o3-mini-high (86.8%). Combining Python and C++ outputs increased Phi-4-14B’s pass@6 to 73.6%. A qualitative analysis revealed some failures stemmed from minor implementation issues rather than reasoning flaws.

Software Engineering Agents (SWE agents) can autonomously perform development tasks on benchmarks like SWE Bench, but still face challenges when tackling complex and ambiguous real-world tasks. Consequently, SWE agents are often designed to allow interactivity with developers, enabling collaborative problem-solving. To understand how developers collaborate with SWE agents and the barriers they face in such interactions, we observed 19 developers using an in-IDE agent to resolve 33 open issues in repositories to which they had previously contributed. Participants successfully resolved about half of these issues, with those solving issues incrementally having greater success than those using a one-shot approach. Participants who actively collaborated with the agent and iterated on its outputs were also more successful, though they faced challenges in trusting the agent’s responses and collaborating on debugging and testing. Our findings suggest that to facilitate successful collaborations, both SWE agents and developers should actively contribute to tasks throughout all stages of the software development process. SWE agents can enable this by challenging and engaging in discussions with developers, rather than being conclusive or sycophantic.

Generative AI has the potential to enhance education by providing personalized feedback to students at scale. Recent work has proposed techniques to improve AI-generated programming hints and has evaluated their performance based on expert-designed rubrics or student ratings. However, it remains unclear how the rubrics used to design these techniques align with students' perceived helpfulness of hints. In this paper, we systematically study the mismatches in perceived hint quality from students' and experts' perspectives based on the deployment of AI-generated hints in a Python programming course. We analyze scenarios with discrepancies between student and expert evaluations, in particular, where experts rated a hint as high-quality while the student found it unhelpful. We identify key reasons for these discrepancies and classify them into categories, such as hints not accounting for the student's main concern or not considering previous help requests. Finally, we propose and discuss preliminary results on potential methods to bridge these gaps, first by extending the expert-designed quality rubric and then by adapting the hint generation process, e.g., incorporating the student's comments or history. These efforts contribute toward scalable, personalized, and pedagogically sound AI-assisted feedback systems, which are particularly important for high-enrollment educational settings.

A race is underway to embed advanced AI capabil-ities into products. These product “copilots” enable users to ask questions in natural language and receive relevant responses that are specific to the user's context. In fact, virtually every large technology company is looking to add these capabilities to their software products. However, for most software engineers, this is often their first encounter with integrating AI-powered technol-ogy. Furthermore, software engineering processes and tools have not caught up with the challenges and scale involved with building AI-powered applications. In this work, we present the findings of an interview study with 26 professional software engineers responsible for building product copilots at various companies. From our interviews, we found pain points at every step of the engineering process and the challenges that strained existing development practices. We then conducted group brainstorming sessions to collaborative on opportunities and tool designs for the broader software engineering community.

Spreadsheet programming is challenging. Programmers use spreadsheet programming knowledge (e.g., formulas) and problem-solving skills to combine actions into complex tasks. Advancements in large language models have introduced language agents that observe, plan, and perform tasks, showing promise for spreadsheet creation. We present TableTalk, a spreadsheet programming agent embodying three design principles—scaffolding, flexibility, and incrementality—derived from studies with seven spreadsheet programmers and 85 Excel templates. TableTalk guides programmers through structured plans based on professional workflows, generating three potential next steps to adapt plans to programmer needs. It uses pre-defined tools to generate spreadsheet components and incrementally build spreadsheets. In a study with 20 programmers, TableTalk produced higher-quality spreadsheets 2.3 times more likely to be preferred than the baseline. It reduced cognitive load and thinking time by 12.6%. From this, we derive design guidelines for agentic spreadsheet programming tools and discuss implications on spreadsheet programming, end-user programming, AI-assisted programming, and human-agent collaboration.

Priyanshu Gupta, Shashank Kirtania, Ananya Singha, Sumit Gulwani, Arjun Radhakrishna, Gustavo Soares, Sherry Shi. Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing. 2024.

Generative AI and large language models hold great promise in enhancing programming education by automatically generating individualized feedback for students. We investigate the role of generative AI models in providing human tutor-style programming hints to help students resolve errors in their buggy programs. Recent works have benchmarked state-of-the-art models for various feedback generation scenarios; however, their overall quality is still inferior to human tutors and not yet ready for real-world deployment. In this paper, we seek to push the limits of generative AI models toward providing high-quality programming hints and develop a novel technique, GPT4HINTS-GPT3.5VAL. As a first step, our technique leverages GPT-4 as a “tutor” model to generate hints – it boosts the generative quality by using symbolic information of failing test cases and fixes in prompts. As a next step, our technique leverages GPT-3.5, a weaker model, as a “student” model to further validate the hint quality – it performs an automatic quality validation by simulating the potential utility of providing this feedback. We show the efficacy of our technique via extensive evaluation using three real-world datasets of Python programs covering a variety of concepts ranging from basic algorithms to regular expressions and data analysis using pandas library.

Data Science (DS) has emerged as a new academic discipline where students are introduced to data-centric thinking and generating data-driven insights through programming. Unlike traditional introductory Computer Science (CS) education, which focuses on program syntax and core CS topics (e.g., algorithms and data structures), introductory DS education emphasizes skills such as analyzing data to gain insights by making effective use of programming libraries (e.g., re, NumPy, pandas, scikit-learn). To better understand learners' needs and pain points when they are introduced to DS programming, we investigated a large online course on data manipulation designed for graduate students who do not have a CS or Statistics undergraduate degree. We qualitatively analyzed students' incorrect code submissions for computational notebook-based assignments in Python. We identified common mistakes and grouped them into the following themes: (1) programming language and environment misconceptions, (2) logical mistakes due to data or problem-statement misunderstanding or incorrectly dealing with missing values, (3) semantic mistakes due to incorrect use of DS libraries, and (4) suboptimal coding. Our work provides instructors insights to understand student needs in introductory DS courses and improve course pedagogy, and recommendations for developing assessment and feedback tools to support students in large courses.

Despite advancements in IDE tooling, code understanding, generation, and automated repair, debugging continues to present significant challenges. Existing debugging strategies available to developers in literature are often too mechanical and rigid for day-to-day issues. Recent advances in Large Language Models (LLMs) promise practical solutions that allow for more free-form debugging strategies. While LLMs offer satisfactory assistance in some cases, they often leap to action without sufficient context, making implicit assumptions and providing inaccurate responses. Moreover, the dialogue between developers and LLMs predominantly takes the form of question-answer pairs, placing the burden of formulating the correct questions and sustaining multi-turn conversations on the developer. We introduce Robin, a novel multi-agent conversational AI-assistant within GitHub Copilot Chat, specifically designed for debugging. Robin moves beyond the question-answer pairs by introducing the investigate & respond pattern, that focuses on using information gathered automatically from the IDE or gathered interactively from the developer before responding. Robin incorporates a general debugging strategy to systematically analyze bugs to sustain collaborative interactions while ensuring that the conversation does not deviate from the debugging task at hand. Through a within-subjects user study with 16 industry professionals, we find that equipping Robin to-(1) leverage the insert expansion interaction pattern, (2) facilitate turn-taking, and (3) utilize debugging strategies-leads to lowered conversation barriers, a 2.5 x improvement in bug localization and a substantial 3.5x improvement in bug resolution compared to AI-assisted debugging in Visual Studio prior to Robin.

Students often make mistakes in their introductory programming assignments as part of their learning process. Unfortunately, providing custom repairs for these mistakes can require a substantial amount of time and effort from class instructors. Automated program repair (APR) techniques can be used to synthesize such fixes. Prior work has explored the use of symbolic and neural techniques for APR in the education domain. Both types of approaches require either substantial engineering efforts or large amounts of data and training. We propose to use a large language model trained on code, such as Codex (a version of GPT), to build an APR system -- PyDex -- for introductory Python programming assignments. Our system can fix both syntactic and semantic mistakes by combining multi-modal prompts, iterative querying, test-case-based selection of few-shots, and program chunking. We evaluate PyDex on 286 real student programs and compare to three baselines, including one that combines a state-of-the-art Python syntax repair engine, BIFI, and a state-of-the-art Python semantic repair engine for student assignments, Refactory. We find that PyDex can fix more programs and produce smaller patches on average.

Generative AI and large language models hold great promise in enhancing computing education by powering next-generation educational technologies. State-of-the-art models like OpenAI’s ChatGPT [8] and GPT-4 [9] could enhance programming education in various roles, e.g., by acting as a personalized digital tutor for a student, a digital assistant for an educator, and a digital peer for collaborative learning [1, 2, 7]. In our work, we seek to comprehensively evaluate and benchmark state-of-the-art large language models for various scenarios in programming education.

Developers spend a significant amount of time in editing code for a variety of reasons such as bug fixing or adding new features. Designing effective methods to predict code edits has been an active yet challenging area of research due to the diversity of code edits and the difficulty of capturing the developer intent. In this work, we address these challenges by endowing pre-trained large language models (LLMs) with the knowledge of relevant prior associated edits, which we call the Grace (Generation conditioned on Associated Code Edits) method. The generative capability of the LLMs helps address the diversity in code changes and conditioning code generation on prior edits helps capture the latent developer intent. We evaluate two well-known LLMs, codex and CodeT5, in zero-shot and fine-tuning settings respectively. In our experiments with two datasets, Grace boosts the performance of the LLMs significantly, enabling them to generate 29% and 54% more correctly edited code in top-1 suggestions relative to the current state-of-the-art symbolic and neural approaches, respectively.

Recent development in NL2CODE (Natural Language to Code) research allows end-users, especially novice programmers to create a concrete implementation of their ideas such as data visualization by providing natural language (NL) instructions. An NL2CODE system often fails to achieve its goal due to three major challenges: the user's words have contextual semantics, the user may not include all details needed for code generation, and the system results are imperfect and require further refinement. To address the aforementioned three challenges for NL to Visualization, we propose a new approach and its supporting tool named NL2VIZ with three salient features: (1) leveraging not only the user's NL input but also the data and program context that the NL query is upon, (2) using hard/soft constraints to reflect different confidence levels in the constraints retrieved from the user input and data/program context, and (3) providing support for result refinement and reuse.

Integrated Development Environments (IDEs) provide tool support to automate many source code editing tasks. Traditionally, IDEs use only the spatial context, i.e., the location where the developer is editing, to generate candidate edit recommendations. However, spatial context alone is often not sufficient to confidently predict the developer’s next edit, and thus IDEs generate many suggestions at a location. Therefore, IDEs generally do not actively offer suggestions and instead, the developer is usually required to click on a specific icon or menu and then select from a large list of potential suggestions. As a consequence, developers often miss the opportunity to use the tool support because they are not aware it exists or forget to use it. To better understand common patterns in developer behavior and produce better edit recommendations, we can additionally use the temporal context, i.e., the edits that a developer was recently performing. To enable edit recommendations based on temporal context, we present Overwatch, a novel technique for learning edit sequence patterns from traces of developers’ edits performed in an IDE. Our experiments show that Overwatch has 78% precision and that Overwatch not only completed edits when developers missed the opportunity to use the IDE tool support but also predicted new edits that have no tool support in the IDE.

Use of third-party libraries is extremely common in application software. The libraries evolve to accommodate new features or mitigate security vulnerabilities, thereby breaking the Application Programming Interface(API) used by the software. Such breaking changes in the libraries may discourage client code from using the new library versions thereby keeping the application vulnerable and not up-to-date. We propose a novel output-oriented program synthesis algorithm to automate API usage adaptations via program transformation. Our aim is not only to rely on the few example human adaptations of the clients from the old library version to the new library version, since this can lead to over-fitting transformation rules. Instead, we also rely on example usages of the new updated library in clients, which provide valuable context for synthesizing and applying the transformation rules. Our tool APIFix provides an automated mechanism to transform application code using the old library versions to code using the new library versions - thereby achieving automated API usage adaptation to fix the effect of breaking changes. Our evaluation shows that the transformation rules inferred by APIFix achieve 98.7% precision and 91.5% recall. By comparing our approach to state-of-the-art program synthesis approaches, we show that our approach significantly reduces over-fitting while synthesizing transformation rules for API usage adaptations.

Multi-modal program synthesis refers to the task of synthesizing programs (code) from their specification given in different forms, such as a combination of natural language and examples. Examples provide a precise but incomplete specification, and natural language provides an ambiguous but more "complete" task description. Machine-learned pre-trained models (PTMs) are adept at handling ambiguous natural language, but struggle with generating syntactically and semantically precise code. Program synthesis techniques can generate correct code, often even from incomplete but precise specifications, such as examples, but they are unable to work with the ambiguity of natural languages. We present an approach that combines PTMs with component-based synthesis (CBS): PTMs are used to generate candidates programs from the natural language description of the task, which are then used to guide the CBS procedure to find the program that matches the precise examples-based specification. We use our combination approach to instantiate multi-modal synthesis systems for two programming domains: the domain of regular expressions and the domain of CSS selectors. Our evaluation demonstrates the effectiveness of our domain-agnostic approach in comparison to a state-of-the-art specialized system, and the generality of our approach in providing multi-modal program synthesis from natural language and examples in different programming domains.

While editing code, it is common for developers to make multiple related repeated edits that are all instances of a more general program transformation. Since this process can be tedious and error-prone, we study the problem of automatically learning program transformations from past edits, which can then be used to predict future edits. We take a novel view of the problem as a semi-supervised learning problem: apart from the concrete edits that are instances of the general transformation, the learning procedure also exploits access to additional inputs (program subtrees) that are marked as positive or negative depending on whether the transformation applies on those inputs. We present a procedure to solve the semi-supervised transformation learning problem using anti-unification and programming-by-example synthesis technology. To eliminate reliance on access to marked additional inputs, we generalize the semi-supervised learning procedure to a feedback-driven procedure that also generates the marked additional inputs in an iterative loop. We apply these ideas to build and evaluate three applications that use different mechanisms for generating feedback. Compared to existing tools that learn program transformations from edits, our feedback-driven semi-supervised approach is vastly more effective in successfully predicting edits with significantly lesser amounts of past edit data.

This Research Full Paper presents a study on the evaluation of feedback tools in introductory programming classes. Recently, several tools have been proposed in order to provide guidance and help students overcome conceptual difficulties in programming education. Some tools leverage clustering algorithms and program repair techniques to automatically generate personalized hints for students' incorrect programs. In contrast, some teachers choose to present students with program visualization tools to help them understand the dynamic execution of a source code. These tools are used to help students get correct solutions for programming assignments. However, due to limitations in assessments, it is still unclear how effective the feedback provided by these tools is. In this study, we analyzed the effectiveness of a tool for generating personalized hints and a tool for visualizing programs. To do so, we conducted a user study in which students, assisted by these tools, implemented solutions for three programming problems. Our results show that personalized hints can significantly reduce student's effort to get correct solutions. In addition, personalized hints can provide students with an understanding of problem solving similar to when using test cases. However, students who used the program visualization tool got lower post-test performance than using other tools.

When working with a document, users often perform context-specific repetitive edits – changes to the document that are similar but specific to the contexts at their locations. Programming by demonstration/examples (PBD/PBE) systems automate these tasks by learning programs to perform the repetitive edits from demonstration or examples. However, PBD/PBE systems are not widely adopted, mainly because they require modal UIs – users must enter a special mode to give the demonstration/examples. This paper presents Blue-Pencil, a modeless system for synthesizing edit suggestions on the fly. Blue-Pencil observes users as they make changes to the document, silently identifies repetitive changes, and automatically suggests transformations that can apply at other locations. Blue-Pencil is parameterized – it allows the ”plug-and-play” of different PBE engines to support different document types and different kinds of transformations. We demonstrate this parameterization by instantiating Blue-Pencil to several domains – C# and SQL code, markdown documents, and spreadsheets – using various existing PBE engines. Our evaluation on 37 code editing sessions shows that Blue-Pencil synthesized edit suggestions with a precision of 0.89 and a recall of 1.0, and took 199 ms to return suggestions on average. Finally, we report on several improvements based on feedback gleaned from a field study with professional programmers to investigate the use of Blue-Pencil during long code editing sessions. Blue-Pencil has been integrated with Visual Studio IntelliCode to power the IntelliCode refactorings feature.

Refactoring engines may have overly strong preconditions preventing developers from applying useful transformations. We find that 32 percent of the Eclipse and JRRT test suites are concerned with detecting overly strong preconditions. In general, developers manually write test cases, which is costly and error prone. Our previous technique detects overly strong preconditions using differential testing. However, it needs at least two refactoring engines. In this work, we propose a technique to detect overly strong preconditions in refactoring engines without needing reference implementations. We automatically generate programs and attempt to refactor them. For each rejected transformation, we attempt to apply it again after disabling the preconditions that lead the refactoring engine to reject the transformation. If it applies a behavior preserving transformation, we consider the disabled preconditions overly strong. We evaluate 10 refactorings of Eclipse and JRRT by generating 154,040 programs. We find 15 overly strong preconditions in Eclipse and 15 in JRRT. Our technique detects 11 bugs that our previous technique cannot detect while missing 5 bugs. We evaluate the technique by replacing the programs generated by JDolly with the input programs of Eclipse and JRRT test suites. Our technique detects 14 overly strong preconditions in Eclipse and 4 in JRRT.

Automatic program transformation tools can be valuable for programmers to help them with refactoring tasks, and for Computer Science students in the form of tutoring systems that suggest repairs to programming assignments. However, manually creating catalogs of transformations is complex and time-consuming. In this paper, we present REFAZER, a technique for automatically learning program transformations. REFAZER builds on the observation that code edits performed by developers can be used as input-output examples for learning program transformations. Example edits may share the same structure but involve different variables and subexpressions, which must be generalized in a transformation at the right level of abstraction. To learn transformations, REFAZER leverages state-of-the-art programming-by-example methodology using the following key components: (a) a novel domain-specific language (DSL) for describing program transformations, (b) domain-specific deductive algorithms for efficiently synthesizing transformations in the DSL, and (c) functions for ranking the synthesized transformations. We instantiate and evaluate REFAZER in two domains. First, given examples of code edits used by students to fix incorrect programming assignment submissions, we learn program transformations that can fix other students' submissions with similar faults. In our evaluation conducted on 4 programming tasks performed by 720 students, our technique helped to fix incorrect submissions for 87% of the students. In the second domain, we use repetitive code edits applied by developers to the same project to synthesize a program transformation that applies these edits to other locations in the code. In our evaluation conducted on 56 scenarios of repetitive edits taken from three large C# open-source projects, REFAZER learns the intended program transformation in 84% of the cases using only 2.9 examples on average.

In large introductory programming classes, teacher feedback on individual incorrect student submissions is often infeasible. Program synthesis techniques are capable of fixing student bugs and generating hints automatically, but they lack the deep domain knowledge of a teacher and can generate functionally correct but stylistically poor fixes. We introduce a mixed-initiative approach which combines teacher expertise with data-driven program synthesis techniques. We demonstrate our novel approach in two systems that use different interaction mechanisms. Our systems use program synthesis to learn bug-fixing code transformations and then cluster incorrect submissions by the transformations that correct them. The MistakeBrowser system learns transformations from examples of students fixing bugs in their own submissions. The FixPropagator system learns transformations from teachers fixing bugs in incorrect student submissions. Teachers can write feedback about a single submission or a cluster of submissions and propagate the feedback to all other submissions that can be fixed by the same transformation. Two studies suggest this approach helps teachers better understand student bugs and write reusable feedback that scales to a massive introductory programming classroom.

Recent advances in program synthesis offer means to automatically debug student submissions and generate personalized feedback in massive programming classrooms. When automatically generating feedback for programming assignments, a key challenge is designing pedagogically useful hints that are as effective as the manual feedback given by teachers. Through an analysis of teachers' hint-giving practices in 132 online Q&A posts, we establish three design guidelines that an effective feedback design should follow. Based on these guidelines, we develop a feedback system that leverages both program synthesis and visualization techniques. Our system compares the dynamic code execution of both incorrect and fixed code and highlights how the error leads to a difference in behavior and where the incorrect code trace diverges from the expected solution. Results from our study suggest that our system enables students to detect and fix bugs that are not caught by students using another existing visual debugging tool.

Recommendation systems are software tools and techniques that provide customized content to users.The collaborative filtering is one of the most prominent approaches in the recommendation area.Among the collaborative algorithms, one of the most popular is the k-Nearest Neighbors (kNN) which is an instance-based learning method.The kNN generates recommendations based on the ratings of the most similar users (nearest neighbors) to the target one.Despite being quite effective, the algorithm performance drops while running on large datasets.We propose a method, called Restricted Space kNN that is based on the restriction of the neighbors search space through a fast and efficient heuristic.The heuristic builds the new search space from the most active users.As a result, we found that using only 15% of the original search space the proposed method generated recommendations almost as accurate as the standard kNN, but with almost 58% less running time.

Recommendation systems have been used in several application domains, most recently for TV (Digital TV, Smart TV, etc.).Several approaches can be used to recommend items, tags, etc., mainly based on user feedback.However, in the Digital TV domain, user feedback has to be done generally by using the remote control, which should be avoided to improve user experience, since assigning explicit feedback to items is restricted by the characteristics of this domain (difficulties when typing with the remote control, etc.).Moreover, in the Smart TV environment several types of items can be recommended (movies, musics, books, etc.).Thus, the recommendation should be generic enough to suit to different content.To solve the problem of acquiring explicit feedback and still generate personalized recommendations to be used by different Smart TV applications, this work proposes a recommendation architecture based on the extraction and classification of terms by analyzing the textual descriptions of TV programs present on electronic programming guides.In order to validate the proposed solution, a prototype using a real dataset has been developed, showing that using the recommended terms it is possible to generate final recommendations for different Smart TV applications.

Programming by Examples (PBE) has the potential to revolutionize end-user programming by enabling end users, most of whom are non-programmers, to create small scripts for automating repetitive tasks. However, examples, though often easy to provide, are an ambiguous specification of the user's intent. Because of that, a key impedance in adoption of PBE systems is the lack of user confidence in the correctness of the program that was synthesized by the system. We present two novel user interaction models that communicate actionable information to the user to help resolve ambiguity in the examples. One of these models allows the user to effectively navigate between the huge set of programs that are consistent with the examples provided by the user. The other model uses active learning to ask directed example-based questions to the user on the test input data over which the user intends to run the synthesized program. Our user studies show that each of these models significantly reduces the number of errors in the performed task without any difference in completion time. Moreover, both models are perceived as useful, and the proactive active-learning based model has a slightly higher preference regarding the users' confidence in the result.

Proving refactoring sound with respect to a formal semantics is considered a challenge. In practice, developers write test cases to check their refactoring implementations. However, it is difficult and time consuming to have a good test suite since it requires complex inputs (programs) and an oracle to check whether it is possible to apply the transformation. If it is possible, the resulting program must preserve the observable behavior. There are some automated techniques for testing refactoring engines. Nevertheless, they may have limitations related to the program generator (exhaustiveness, setup, expressiveness), automation (types of oracles, bug categorization), time consumption or kinds of refactorings that can be tested. In this paper, we extend our previous technique to test refactoring engines. We improve expressiveness of the program generator for testing more kinds of refactorings, such as Extract Function. Moreover, developers just need to specify the input's structure in a declarative language. They may also set the technique to skip some consecutive test inputs to improve performance. We evaluate our technique in 18 refactoring implementations of Java (Eclipse and JRRT) and C (Eclipse). We identify 76 bugs (53 new bugs) related to compilation errors, behavioral changes, and overly strong conditions. We also compare the impact of the skip on the time consumption and bug detection in our technique. By using a skip of 25 in the program generator, it reduces in 96% the time to test the refactoring implementations while missing only 3.9% of the bugs. In a few seconds, it finds the first failure related to compilation error or behavioral change.

Refactoring is a transformation that preserves the external behavior of a program and improves its internal quality. Usually, compilation errors and behavioral changes are avoided by preconditions determined for each refactoring transformation. However, to formally define these preconditions and transfer them to program checks is a rather complex task. In practice, refactoring engine developers commonly implement refactorings in an ad hoc manner since no guidelines are available for evaluating the correctness of refactoring implementations. As a result, even mainstream refactoring engines contain critical bugs. We present a technique to test Java refactoring engines. It automates test input generation by using a Java program generator that exhaustively generates programs for a given scope of Java declarations. The refactoring under test is applied to each generated program. The technique uses SafeRefactor, a tool for detecting behavioral changes, as an oracle to evaluate the correctness of these transformations. Finally, the technique classifies the failing transformations by the kind of behavioral change or compilation error introduced by them. We have evaluated this technique by testing 29 refactorings in Eclipse JDT, NetBeans, and the JastAdd Refactoring Tools. We analyzed 153,444 transformations, and identified 57 bugs related to compilation errors, and 63 bugs related to behavioral changes.

Each refactoring implementation must check a number of conditions to guarantee behavior preservation. However, specifying and checking them are difficult. Sometimes refactoring tool developers may define overly strong conditions that prevent useful behavior-preserving transformations to be performed. We propose an approach for identifying overly strong conditions in refactoring implementations. We automatically generate a number of programs as test inputs for refactoring implementations. Then, we apply the same refactoring to each test input using two different implementations, and compare both results. We use Safe Refactor to evaluate whether a transformation preserves behavior. We evaluated our approach in 10 kinds of refactorings for Java implemented by three tools: Eclipse and Netbeans, and the JastAdd Refactoring Tool (JRRT). In a sample of 42,774 transformations, we identified 17 and 7 kinds of overly strong conditions in Eclipse and JRRT, respectively.