Generative communication is the basis of a new distributed programming language called Linda, designed for systems programming in distributed environments. It differs from traditional interprocess communication models by specifying that messages be added as structured tuples to the computation environment, where they exist as independent entities until a process chooses to receive them. Linda is fully distributed in both space and time, allowing for distributed sharing, continuation passing, and structured naming. The language's distinguishing properties include communication orthogonality, space-uncoupling (distributed naming), time-uncoupling, and distributed sharing. These properties enable Linda to support continuation passing, structured naming, and distributed sharing, making it more expressive and simpler than existing distributed programming languages. Linda's design also includes a model of computation with implications beyond distribution and communication. The paper discusses the implementation of generative communication in Linda, focusing on the dynamic global name space required by the model. Examples illustrate the language's features, including basic communication structures, semaphores, distributed naming, and synchronized message exchange. The paper also discusses the use of structured names, continuation passing, and the implementation of resource servers. The examples demonstrate Linda's ability to handle complex distributed programming tasks, including resource management and communication between processes. The paper concludes by comparing Linda to other distributed programming languages and highlighting its unique features and advantages.Generative communication is the basis of a new distributed programming language called Linda, designed for systems programming in distributed environments. It differs from traditional interprocess communication models by specifying that messages be added as structured tuples to the computation environment, where they exist as independent entities until a process chooses to receive them. Linda is fully distributed in both space and time, allowing for distributed sharing, continuation passing, and structured naming. The language's distinguishing properties include communication orthogonality, space-uncoupling (distributed naming), time-uncoupling, and distributed sharing. These properties enable Linda to support continuation passing, structured naming, and distributed sharing, making it more expressive and simpler than existing distributed programming languages. Linda's design also includes a model of computation with implications beyond distribution and communication. The paper discusses the implementation of generative communication in Linda, focusing on the dynamic global name space required by the model. Examples illustrate the language's features, including basic communication structures, semaphores, distributed naming, and synchronized message exchange. The paper also discusses the use of structured names, continuation passing, and the implementation of resource servers. The examples demonstrate Linda's ability to handle complex distributed programming tasks, including resource management and communication between processes. The paper concludes by comparing Linda to other distributed programming languages and highlighting its unique features and advantages.