Extending the Do! Framework with Dynamic Collections
Abstract
The aim of the Do! project is to ease the task of programming distributed applications using Java. We provide a shared-memory parallel programming model, possibly including additional informations about object locations; we generate distributed programs from parallel programs by using the user's object location specifications, and by automatic transformation of user-define- d classes. Writing a distributed application is a three-step process: in the first step, the programmer writes his application as a shared-memory parallel program, by defining the components of the program; in the second step, the programmer defines the component locations on distinct hosts; in the third step, the Do! preprocessor transforms the shared-memory parallel program into a distributed-memory program. In this paper, we present an extension to our computing framework in order to create and activate tasks at runtime; this extension allows the introduction of load balancing strategies in the framework by taking into account runtime informations in the component location descriptions. This extension has been integrated in the parallel and distributed frameworks, without any modification in the Do! preprocessor.