Generative patterns - higher level programming

生成模式 - 高级编程

• 批准号: 8191-2006
• 负责人: Szafron, Duane
• 金额: $ 2.09万
• 依托单位: University of Alberta
• 依托单位国家: 加拿大
• 项目类别: Discovery Grants Program - Individual
• 财政年份: 2008
• 资助国家: 加拿大
• 起止时间: 2008-01-01 至 2009-12-31
• 项目状态: 已结题
• 来源: https://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=389689
• 关键词: Generative; patterns; higher; level; programming

The goal of my research is to provide non-programmers with the ability to generate computer programs in specific application domains, without writing a program in the traditional sense. I am not proposing to solve the difficult problem of automatic program generation from specifications in arbitrary application domains. Instead, I want to solve a similar problem in a more limited scope.  In this research, I will try to apply automatic program generation to the creation of computer game adventures. Programmers write a single game engine that can be used to "play" many game adventures. However, custom scripting code must be attached to game objects for each game adventure that uses the game engine. The research goal is that a story writer with no programming skills can generate this scripting code, without learning how to program or finding a programmer to write the scripting code.The scientific technique that I will use to try to solve this problem is generative design patterns, based on the well-known idea of design patterns. Traditional design patterns are descriptive. Each design pattern describes a set of known solutions to a recurring design problem, by providing a design lexicon, solution structures and the reasoning behind the solutions. Since each pattern is a family of solutions, it must be adapted to a specific context during program construction. An adapted design pattern is a detailed specification that can be translated to code quickly by experienced programmers who have implemented the same design pattern in other contexts. I have successfully generalized descriptive design patterns to generative design patterns in the domain of parallel programming, by replacing the final step of manually translating the solution to code by an automatic code generation step. However, it can be argued that giving sequential programmers the ability to generate parallel code is easier than giving non-programmers the ability to generate scripting code. Therefore, this research will establish whether non-programmers can use generative design patterns to generate scripting code in a specific application domain - computer games. If this research is successful, future research could try to use generative design patterns in other specific application domains.

我研究的目标是为非程序员提供在特定应用领域生成计算机程序的能力，而不是编写传统意义上的程序。我并不打算解决从任意应用领域的规范自动生成程序的难题。相反，我想在一个更有限的范围内解决一个类似的问题。在这个研究中，我将尝试将自动程序生成应用到计算机游戏冒险的创作中。程序员编写一个单一的游戏引擎，可以用来“玩”许多游戏冒险。但是，对于使用游戏引擎的每个游戏冒险，必须将自定义脚本代码附加到游戏对象。研究的目标是让一个没有编程技能的故事作者可以生成脚本代码，而不需要学习如何编程或找一个程序员来编写脚本代码。我将使用的科学技术来解决这个问题是基于众所周知的设计模式思想的生成设计模式。传统的设计模式是描述性的。每个设计模式通过提供设计词典、解决方案结构和解决方案背后的推理，描述了一组已知的解决方案。由于每种模式都是一系列解决方案，因此在程序构建期间必须使其适应特定的上下文。适应的设计模式是一个详细的规范，可以由经验丰富的程序员快速翻译成代码，这些程序员已经在其他上下文中实现了相同的设计模式。我已经成功地将描述性设计模式推广到并行编程领域的生成性设计模式，用自动代码生成步骤取代了手动将解决方案翻译为代码的最后一步。然而，可以说，让顺序程序员生成并行代码的能力比让非程序员生成脚本代码的能力更容易。因此，本研究将确定非程序员是否可以使用生成式设计模式来生成脚本代码在一个特定的应用领域-计算机游戏。如果这项研究是成功的，未来的研究可以尝试在其他特定的应用领域使用生成式设计模式。