A Compositional Approach to Support Developers in Developing Ultra-large-scale Systems

我们的社会依赖于软件来处理我们日常生活的每一部分。软件工程支持这样的软件片段的定义，例如，通过回答诸如“如何以正确的方式构建正确的产品？”这样的问题。但是，当今社会所涉及的软件的复杂性在不断增长，在这种情况下，规模改变了软件开发人员的一切。超大规模系统（ulss）的概念是由Northop等人在2006年提出的，用于识别一类新的系统，其特点是根据任何可能的维度（例如源代码、需求、用户数量、存储数据），它们具有内在的复杂性和大规模方面，并且不断发展。关注点分离对于将这种大规模系统划分为开发人员可以贡献的可管理单元至关重要。学术界和工业界投入了许多努力来解决与这些系统相关的研究挑战。然而，他们各自应对不同的挑战。此外，这些工作针对的是不同的涉众（例如，产品所有者、架构师），而不是软件开发人员。这种隔离和缺乏重点导致了这样一种情况，即软件开发人员参与了ulss的紧急设计，但在日常任务中却得不到支持。我的研究计划的长期目标是通过确定支持开发人员设计和实现ulss的核心元素来调查和衡量软件开发人员如何设计ulss。我定义了三个短期目标：(I)从开发商的角度为ulss建模，（ii）定义组合模型以支持开发商，以及（iii）对ulss进行实证分析以收集需求并验证所获得的结果。为了实现这些目标，我从软件开发人员的角度考虑了与ulss相关的三个研究任务系列：(I)超大规模源代码的开发，（ii）不断发展的需求的管理，以及（iii）软件开发人员在ulss的设计、实现和维护中的角色。我的项目的结果将作为开源数据和软件提供，我们也依靠开源系统来衡量应用于现实生活中的ulss的贡献的好处。

Our society relies on software to deal with every part of our daily lives. Software engineering supports the definition of such pieces of software, e.g., by answering to questions like “how to build the right product in the right way?”. But the complexity of the pieces of software involved nowadays on our society is constantly growing and, in this context, scale changes everything for software developers. The notion of Ultra-large-scale Systems (ULSSs) was coined by Northop et al. in 2006 to identify a new class of systems, characterized by their intrinsic complexity and large-scale aspect according to any possible dimensions (e.g., source code, requirements, number of users, stored data), and constant evolution. Separation of concerns is essential to divide such large-scale systems into manageable units to wich developers can contribute. Academia and industry invested many efforts to address the research challenges associated with these systems. However, they addressed the different challenges in isolation. Furthermore, these efforts targeted various stakeholders (e.g., product owners, architects) and were not focused on the software developers. This isolation and lack of focus led to a situation in which software developers are involved in the emergent design of ULSSs but are not supported in their daily tasks. The long-term objective of my research program is then to investigate and to measure how software developers design ULSSs by identifying the core elements that support developers designing and implementing ULSSs. I define three short-term objectives: (i) the modelling of ULSSs from the developer's point of view, (ii) the definition of compositional models to support developers and (iii) the empirical analysis of ULSSs to gather requirements and validate the obtained results. To address these objectives, I consider three families of research tasks relevant to ULSSs from the software developer's point of view: (i) the development of ultra-large-scale source-code, (ii) the management of evolving requirements, and (iii) the role of the software developer in the design, implementation and maintenance of ULSSs. The results of my program will be made available as open-source data and software, and we also rely on open-source systems to measure the benefits of the contribution applied to real-life ULSSs.