Zhenchang Xing

Feature Location in a Collection of Product Variants

Companies often develop and maintain a collection of product variants that share some common features but also support different, customer-specific features. To reengineering such legacy product variants for systematic reuse, one must identify features and their implementing code units (e.g. functions, files) in different product variants. Information retrieval (IR) techniques may be applied for that purpose. In this paper, we discuss problems that hinder direct application of IR techniques to a collection of product variants. To counter these problems, we present an approach to support effective feature location in product variants. The novelty of our approach is that we exploit commonalities and differences of product variants by software differencing and FCA techniques so that IR technique can achieve satisfactory results for feature location in product variants. We have implemented our approach and conducted evaluation with a collection of nine Linux kernel product variants. Our evaluation shows that our approach always significantly outperforms a direct application of IR technique in the subject product variants.

Concern Localization using Information Retrieval: An Empirical Study on Linux Kernel

Many software maintenance activities need to find code units (functions, files, etc.) that implement a certain concern (features, bugs, etc.). To facilitate such activities, many approaches have been proposed to automatically link code units with concerns described in natural languages, which are termed as concern localization and often employ Information Retrieval (IR) techniques. There has not been a study that evaluates and compares the effectiveness of latest IR techniques on a large dataset. This study fills this gap by investigating ten IR techniques, some of which are new and have not been used for concern localization, on a Linux kernel dataset. The Linux kernel dataset contains more than 1,500 concerns that are linked to over 85,000 C functions. We have evaluated the effectiveness of the ten techniques on recovering the links between the concerns and the implementing functions and ranked the IR techniques based on their precisions on concern localization. Keywords-concern localization; information retrieval; Linux kernel; mean average precision;

An exploratory study of feature location process: Distinct phases, recurring patterns, and elementary actions

Developers often have to locate the parts of the source code that contribute to a specific feature during software maintenance tasks. This activity, referred to as feature location in software engineering, is a human- and knowledge-intensive process. Researchers have investigated information retrieval, static/dynamic analysis based techniques to assist developers in such feature location activities. However, little work has been done on better understanding how developers perform feature location tasks. In this paper, we report an exploratory study of feature location process, consisting of two experiments in which developers were given unfamiliar systems and asked to complete six feature location tasks in two hours. Our study suggests that feature location process can be understood hierarchically at three levels of granularities: phase, pattern, and action. Furthermore, our study suggests that these feature-location phases, patterns and actions can be effectively imparted to junior developers and consequently improve their performance on feature location tasks. Our results open up new opportunities to feature location research, which could lead to better tool support and more rigorous feature location process.

Understanding Feature Evolution in a Family of Product Variants

Existing software product variants, developed by ad hoc reuse such as copy-paste-modify, are often a starting point for building Software Product Line (SPL). Understanding of how features evolved in product variants is a prerequisite to transition from ad hoc to systematic SPL reuse. We propose a method that assists analysts in detecting changes to product features during evolution. We first entail that features and their inter-dependencies for each product variant are documented as product feature model. We then apply model differencing algorithm to identify evolutionary changes that occurred to features of different product variants. We evaluate the effectiveness of our approach on a family of medium-size financial systems. We also investigate the scalability of our approach with synthetic data. The evaluation demonstrates that our approach yields good results and scales to large systems. Our approach enables the subsequent variability analysis and consolidation of product variants in the task of reengineering product variants into SPL.

How developers perform feature location tasks: a human‐centric and process‐oriented exploratory study

Developers often have to locate the parts of source code that contribute to a specific feature during software maintenance tasks. This activity, referred to as feature location in software engineering, is a human‐intensive and knowledge‐intensive process. Researchers have investigated (semi‐)automatic analysis‐based techniques to assist developers in such feature location activities. However, little work has been carried out on better understanding how developers perform feature location tasks. In this paper, we report an exploratory study of feature location process, consisting of three experiments in which developers were given unfamiliar systems and asked to complete six feature location tasks. Our study suggests that feature location process can be understood hierarchically at three levels of granularity: phase, pattern, and action. Furthermore, our statistical analysis shows that these feature location phases, patterns, and actions can be effectively imparted to junior developers and consequently improve their performance on feature location tasks. Our qualitative observations and interviews also suggest that external factors, for example, human factors, task properties, and in‐process feedbacks, affect the choices and usage of different feature location patterns and actions. Our results open up new opportunities to feature location research, which could lead to better tool support and more rigorous feature location process. Copyright

Incremental and iterative reengineering towards Software Product Line: An industrial case study

It is common in practice that a Software Product Line (SPL) is constructed by reengineering a set of existing variant products. To alleviate the problems of high risks of failures and the limitations of resources and cost, incremental reengineering towards a SPL is a natural choice in many cases. However, several problems remain unaddressed properly, such as how to define increments, how to satisfy regular product delivery in parallel with reengineering, and how to achieve early successes. In this paper, we report an industrial case study on a successful SPL-targeted reengineering project conducted in Alcatel-Lucent. In this project, the project team applied the principles of agile development in the process of SPL reengineering. The key practices of the project include value-based increment definition, domain-driven responsibility alignment, iterative component refactoring and integration. We analyze the reengineering process of a major component qualitatively and quantitatively, with the focus on initial investment required, trend of investment, returns on investment and quality improvement. Our case study shows that incremental and iterative approach with stakeholder-value considerations can help to achieve steady and successful SPL reengineering in a cost-effective manner. We also find that SPL adoption can be regarded as an emergent result of the reconstruction and improvement of existing product assets.

Iterative context-aware feature location (NIER track)

Locating the program element(s) relevant to a particular feature is an important step in efficient maintenance of a software system. The existing feature location techniques analyze each feature independently and perform a one-time analysis after being provided an initial input. As a result, these techniques are sensitive to the quality of the input, and they tend to miss the nonlocal interactions among features. In this paper, we propose to address the proceeding two issues in feature location using an iterative context-aware approach. The underlying intuition is that the features are not independent of each other, and the structure of source code resembles the structure of features. The distinguishing characteristics of the proposed approach are: 1) it takes into account the structural similarity between a feature and a program element to determine their relevance; 2) it employs an iterative process to propagate the relevance of the established mappings between a feature and a program element to the neighboring features and program elements. Our initial evaluation suggests the proposed approach is more robust and can significantly increase the recall of feature location with a slight decrease in precision.

Monitoring Software Quality Evolution by Analyzing Deviation Trends of Modularity Views

In the long-term evolution of software systems, various maintenance activities such as functionality extension, bug fixing, refactoring may positively or negatively affect the quality of design and implementation. The trend of quality degradation caused by negative affections may accumulate and cause serious difficulties for future maintenance of the software if they were not addressed properly in time. In this paper, we propose an approach for monitoring the degradation trends of software design in evolution and providing useful feedbacks for evolution decisions. The approach is based on the assumption that the deviations between different modularity views and their trends in evolution can be used to monitor the degradation trends of design. Currently, our approach considers three modularity views, namely package view, structural cluster view and semantic cluster view. Package view denotes the package structure reflecting the desired modularity view, Structural cluster view and semantic cluster view are the modularity views extracted from implementation by software clustering based on formal information and non-formal information, respectively. Then based on the three modularity views extracted from each version, our approach calculates the similarity between different views as the measurement of modularity deviations, and analyzes the deviation trends over a series of versions. We conduct an empirical study on three open-source systems, which confirms that continuous monitoring of deviation trends of modularity views can provide useful feedbacks for future evolution decisions.

CloneDiff: semantic differencing of clones

Clone detection provides a scalable and efficient way to detect similar codes, while program differencing is a powerful and effective way to analyze similar codes. CloneDiff, a Program Dependence Graphs (PDGs) differencing tool, complements clone detection with program differencing for the purpose of characterizing clones. It captures semantic information of clones from PDGs, and uses graph matching techniques to compute a precise characterization of clones in terms of a category of semantic differences.

Improving product line architecture design and customization by raising the level of variability modeling

Product Line Architecture (PLA) plays a central role in software product line development. In order to support architecture-level variability modeling, most architecture description languages (ADLs) introduce architectural variation elements, such as optional component, connector and interface, which must be customized during product derivation. Variation elements are many, and design and customization of PLA at the level of individual variation elements are difficult and error-prone. We observed that developers usually perceive architecture variability from the perspective of variant features or variant design decisions that are mapped into groups of architecture variation elements. In the paper, we describe heuristics to identify configurations of variation elements that typically form such groups. We call them variation constructs. We developed an architecture variability management method and a tool that allow developers to work at the variation construct level rather than at the level of individual variation elements. We have applied and evaluated the proposed method in the development and maintenance of a medium-size financial product line. Our experience indicates that by raising variability modeling from variation element to construct level, architecture design and customizations become more intuitive. Not only does our method reduce the design and customization effort, but also better ensures consistent configuration of architectural variation elements, avoiding errors.