Copying and pasting of source code is a common activity in software engineering. Often, the code is not copied as it is and it may be modified for various purposes; e.g. refactoring, bug fixing, or even software plagiarism. These code modifications could affect the performance of code similarity analysers including code clone and plagiarism detectors to some certain degree. We are interested in two types of code modification in this study: pervasive modifications, i.e. transformations that may have a global effect, and local modifications, i.e. code changes that are contained in a single method or code block. We evaluate 30 code similarity detection techniques and tools using five experimental scenarios for Java source code. These are (1) pervasively modified code, created with tools for source code and bytecode obfuscation, and boiler-plate code, (2) source code normalisation through compilation and decompilation using different decompilers, (3) reuse of optimal configurations over different data sets, (4) tool evaluation using ranked-based measures, and (5) local + global code modifications. Our experimental results show that in the presence of pervasive modifications, some of the general textual similarity measures can offer similar performance to specialised code similarity tools, whilst in the presence of boiler-plate code, highly specialised source code similarity detection techniques and tools outperform textual similarity measures. Our study strongly validates the use of compilation/decompilation as a normalisation technique. Its use reduced false classifications to zero for three of the tools. Moreover, we demonstrate that optimal configurations are very sensitive to a specific data set. After directly applying optimal configurations derived from one data set to another, the tools perform poorly on the new data set. The code similarity analysers are thoroughly evaluated not only based on several well-known pair-based and query-based error measures but also on each specific type of pervasive code modification. This broad, thorough study is the largest in existence and potentially an invaluable guide for future users of similarity detection in source code.

Online code clones are code fragments that are copied from software projects or online sources to Stack Overflow as examples. Due to an absence of a checking mechanism after the code has been copied to Stack Overflow, they can become toxic code snippets, e.g., they suffer from being outdated or violating the original software license. We present a study of online code clones on Stack Overflow and their toxicity by incorporating two developer surveys and a large-scale code clone detection. A survey of 201 high-reputation Stack Overflow answerers (33% response rate) showed that 131 participants (65%) have ever been notified of outdated code and 26 of them (20%) rarely or never fix the code. 138 answerers (69%) never check for licensing conflicts between their copied code snippets and Stack Overflow's CC BY-SA 3.0. A survey of 87 Stack Overflow visitors shows that they experienced several issues from Stack Overflow answers: mismatched solutions, outdated solutions, incorrect solutions, and buggy code. 85% of them are not aware of CC BY-SA 3.0 license enforced by Stack Overflow, and 66% never check for license conflicts when reusing code snippets. Our clone detection found online clone pairs between 72,365 Java code snippets on Stack Overflow and 111 open source projects in the curated Qualitas corpus. We analysed 2,289 non-trivial online clone candidates. Our investigation revealed strong evidence that 153 clones have been copied from a Qualitas project to Stack Overflow. We found 100 of them (66%) to be outdated, of which 10 were buggy and harmful for reuse. Furthermore, we found 214 code snippets that could potentially violate the license of their original software and appear 7,112 times in 2,427 GitHub projects.We performed an empirical study of online code clones between Stack Overflow and 111 Java open source projects 8. https://ucl-crest.github.io/cloverflow-web This is the author's version of an article that has been published in this journal. Changes were made to this version by the publisher prior to publication.The final version of record is available at http://dx.

This paper presents a novel code clone search technique that is accurate, incremental, and scalable to hundreds of million lines of code. Our technique incorporates multiple code representations (i.e., a technique to transform code into various representations to capture different types of clones), query reduction (i.e., a technique to select clone search keywords based on their uniqueness), and a customised ranking function (i.e., a technique to allow a specific clone type to be ranked on top of the search results) to improve clone search performance. We implemented the technique in a clone search tool, called Siamese, and evaluated its search accuracy and scalability on three established clone data sets. Siamese offers the highest mean average precision of 95% and 99% on two clone benchmarks compared to seven state-of-the-art clone detection tools, and reported the largest number of Type-3 clones compared to three other code search engines. Siamese is scalable and can return cloned code snippets within 8 seconds for a code corpus of 365 million lines of code. Using an index of 130,719 GitHub projects, we demonstrate that Siamese's incremental indexing capability dramatically decreases the index preparation time for large-scale data sets with multiple releases of software projects. The paper discusses the applications of Siamese to facilitate software development and research with two use cases including online code clone detection and clone search with automated license analysis.

Source code analysis to detect code cloning, code plagiarism, and code reuse suffers from the problem of pervasive code modifications, i.e. transformations that may have a global effect. We compare 30 similarity detection techniques and tools against pervasive code modifications. We evaluate the tools using two experimental scenarios for Java source code. These are (1) pervasive modifications created with tools for source code and bytecode obfuscation and (2) source code normalisation through compilation and decompilation using different decompilers. Our experimental results show that highly specialised source code similarity detection techniques and tools can perform better than more general, textual similarity measures. Our study strongly validates the use of compilation/decompilation as a normalisation technique. Its use reduced false classifications to zero for six of the tools. This broad, thorough study is the largest in existence and potentially an invaluable guide for future users of similarity detection in source code.

Although considered one of the most important decisions in the software development lifecycle, empirical evidence on how developers perform and perceive architectural changes remains scarce. Architectural decisions have far-reaching consequences yet, we know relatively little about the level of developers' awareness of their changes' impact on the software's architecture. We also know little about whether architecture-related discussions between developers lead to better architectural changes. To provide a better understanding of these questions, we use the code review data from 7 open source systems to investigate developers' intent and awareness when performing changes alongside the evolution of the changes during the reviewing process. We extracted the code base of 18,400 reviews and 51,889 revisions. 4,171 of the reviews have changes in their computed architectural metrics, and 731 present significant changes to the architecture. We manually inspected all reviews that caused significant changes and found that developers are discussing the impact of their changes on the architectural structure in only 31% of the cases, suggesting a lack of awareness. Moreover, we noticed that in 73% of the cases in which developers provided architectural feedback during code review, the comments were addressed, where the final merged revision tended to exhibit higher architectural improvement than reviews in which the system's structure is not discussed.

Abstract-We study effects of compilation and decompilation to code clone detection in Java. Compilation/decompilation canonicalise syntactic changes made to source code and can be used as source code normalisation. We used NiCad to detect clones before and after decompilation in three open source software systems, JUnit, JFreeChart, and Tomcat. We filtered and compared the clones in the original and decompiled clone set and found that 1,201 clone pairs (78.7%) are common between the two sets while 326 pairs (21.3%) are only in one of the sets. A manual investigation identified 325 out of the 326 pairs as true clones. The 252 original-only clone pairs contain a single false positive while the 74 decompiled-only clone pairs are all true positives. Many clones in the original source code that are detected only after decompilation are type-3 clones that are difficult to detect due to added or deleted statements, keywords, package names; flipped if-else statements; or changed loops. We suggest to use decompilation as normalisation to complement clone detection. By combining clones found before and after decompilation, one can achieve higher recall without losing precision.

Abstract-This paper introduces a new code clone detection technique based on image similarity. The technique captures visual perception of code seen by humans in an IDE by applying syntax highlighting and images conversion on raw source code text. We compared two similarity measures, Jaccard and earth mover's distance (EMD) for our image-based code clone detection technique. Jaccard similarity offered better detection performance than EMD. The F1 score of our technique on detecting Java clones with pervasive code modifications is comparable to five well-known code clone detectors: CCFinderX, Deckard, iClones, NiCad, and Simian. A Gaussian blur filter is chosen as a normalisation technique for type-2 and type-3 clones. We found that blurring code images before similarity computation resulted in higher precision and recall. The detection performance after including the blur filter increased by 1 to 6 percent. The manual investigation of clone pairs in three software systems revealed that our technique, while it missed some of the true clones, could also detect additional true clone pairs missed by NiCad.

No abstract