D'Caml: Native support for distributed ML programming in heterogeneous environment

Abstract

Distributed Caml (D'Caml) is a distributed implementation of Caml, a dialect of ML. The compiler produces native code for diverse execution platforms. The distributed shared memory allows transmission and sharing of arbitrary ML objects including higher-order functions, exceptions, and mutable objects without affecting the sequential semantics of ML. The distributed garbage collector reclaims unused distributed data-structures. Examples demonstrate expressivity of higher-order distributed programming using Distributed Caml. The paper presents the design, implementation, and preliminary performance results of the system. © Springer-Verlag Berlin Heidelberg 1999.