仿射会话
Affine Sessions
AI总结:
本研究将会话类型的线性性放宽为仿射性,允许通道最多执行规定行为,从而引入优雅的错误处理机制,并保持会话不卡住的进展属性。
摘要:
会话类型描述了通道所实现的通信结构。具体而言,它们规定了通信的顺序,无论是输入还是输出动作,以及所交换值的类型。对于任何具有会话类型的语言而言,线性性概念至关重要,它确保通道在并发存在下表现出其类型所规定的行为而不受干扰。在本工作中,我们将线性性条件放宽为仿射性,由此通道最多表现出其类型所规定的行为。这种更宽松的设置使我们能够纳入一种优雅的错误处理机制,该机制简化并改进了关于异常的现有相关工作。此外,我们的处理不影响语言的进展属性:会话永远不会卡住。
英文摘要:
Session types describe the structure of communications implemented by channels. In particular, they prescribe the sequence of communications, whether they are input or output actions, and the type of value exchanged. Crucial to any language with session types is the notion of linearity, which is essential to ensure that channels exhibit the behaviour prescribed by their type without interference in the presence of concurrency. In this work we relax the condition of linearity to that of affinity, by which channels exhibit at most the behaviour prescribed by their types. This more liberal setting allows us to incorporate an elegant error handling mechanism which simplifies and improves related works on exceptions. Moreover, our treatment does not affect the progress properties of the language: sessions never get stuck.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。