Implementing distributed Linda in Standard ML

We have implemented the Linda model of shared distributed tuple space in a functional programming language, Standard ML. We use ML's flexible type system and pattern matching facilities to provide ML programmers with the basic Linda operations on tuples. No preprocessor is used, and no compiler changes are required. We use separate ML modules to implement the Linda interface, operations on tuple space, communication of tuples over the network, and replication of tuple spaces. Our approach allows different compositions of these modules to be used to configure a system with either local or remote access to tuple space, and with either a centralized or distributed implementation of tuple space. The resulting implementation of Linda in Standard ML offers an attractive way to separate the functional and the imperative portions of a distributed system. Individual processes can be written in ML in a pure functional style and the Linda shared tuple space can be used to interconnect the processes and maintain the state of the system. This research was sponsored by the Defense Advanced Research Projects Agency and monitored by the Air Force Systems Command under Contract F19628-91-C-0128. The views and conclusions contained in this document are those of the authors and should not be interpreted as representing the official policies, either expressed or implied, of the Defense Advanced Research Projects Agency or the U.S. Government.