Supporting the evolution of software ecosystems

如今，越来越多的现代软件系统是通过将各种可重用的软件组件(也称为库或包)粘合在一起来构建的。这些为开发人员提供多种功能的软件组件通常发布在所谓的软件生态系统上。最近的行业报告显示，从第三方重用的代码是内部开发代码量的四倍。然而，开发如此丰富的软件生态系统并依赖它们绝非轻而易举，因为这些生态系统中的各个组件之间存在着严重的相互依赖。例如，在生态系统中的单个组件中引入的一个简单问题可能会导致生态系统中数千个其他组件的崩溃，并导致依赖这些组件的软件系统崩溃。另一方面，开发人员需要定期使用新功能和错误或安全修复来更新他们使用的组件；但他们应该避免过于频繁地引入破坏性更改，因为这会给他们带来额外的负担。这些观察表明，迫切需要新的技术和方法来支持软件生态系统的演变和维护。该研究计划的长期目标是创建创新的方法和技术，为软件生态系统的发展提供自动化支持。该研究计划的短期目标将沿着两个目标进行：(1)研究对整个或部分生态系统产生负面影响的组件级别的风险变化(例如，单个组件的变化导致生态系统中其他依赖组件崩溃)和(2)调查和开发方法，以自动更新软件系统的依赖关系，并在需要时提供修复程序。在本研究方案中，将采用定量和定性相结合的研究方法，系统地分析软件生态系统的大规模数据集，提出支持软件生态系统整体演化的实用解决方案，并了解软件生态系统开发人员对所提解决方案的看法。这项拟议研究的结果将为降低日益增长的软件维护成本和提高开发人员的生产率提供必要的见解和工具支持，从而为改善我们的软件行业做出贡献。此外，拟议的研究将暴露、培训和使高素质人员(HQP)能够为软件研究的最新水平做出贡献。

Modern software systems today are increasingly being built by gluing together various reusable software components (aka libraries or packages). These software components that provide developers with multiple functionalities are often published on what is known as software ecosystems. Recent industry reports show that code reused from third parties is four times the amount of code being developed in-house. However, exploiting such rich software ecosystems and depending on them is far from trivial due to the drastic interdependence among individual components in these ecosystems. For example, one simple issue introduced in a single component in an ecosystem may lead to the crash of thousands of other components in the ecosystem and result in the crashing of the software systems that depend on these components. On the flip side, developers need to regularly update their used components with new functionalities and bug or security fixes; but they should avoid introducing breaking changes too frequently as this imposes an extra burden on them. These observations show that new techniques and approaches are urgently needed to support the evolution and maintenance of software ecosystems. The long-term goal of this research program is to create innovative methods and techniques that provide automated support for the evolution of software ecosystems. The short-term goal of this research program will be carried out along two objectives: (1) studying the component-level risky changes that negatively impact the whole ecosystem or part of it (e.g., a change in an individual component that leads to the crashing of other dependent components in the ecosystem) and (2) investigating and developing approaches to automatically update the dependencies of software systems and provide fixes if needed. In this research proposal, both quantitative and qualitative research methods will be used to systematically analyze the large-scale datasets of software ecosystems, propose pragmatic solutions to support the overall evolution of software ecosystems, and understand the software ecosystem developers' perspectives about the proposed solutions. The outcomes of this proposed research will provide essential insights and tool support to reduce the growing software maintenance costs and increase developers' productivity, thereby contributing to improving our software industry. Furthermore, the proposed research will expose, train and enable highly qualified personnel (HQP) to contribute to the state-of-the-art in software research.