How to program networked portable computers

Abstract

What is the correct programming model for networked portable computers? Such computers come in many shapes and sizes, so it is unreasonable to expect that there is one programming model appropriate for every single variety. However, it is likely that there is some model that will work well for a wide range of cases. We propose the intelligently autonomous model for this role. Our goal for a widely applicable model is that it should allow applications to operate during disconnections, and that during disconnections the user shouldn't be surprised, that is, applications should behave predictably. The intelligently autonomous model fulfills these goals by having a remote file system with caching, and allowing user control over the cache. In more detail, we are interested in portable networked computers that: frequently access external data, have a reasonably powerful CPU, have a wireless network that suffers from intermittent, unpredictable disruptions. We survey existing models, and then explain why we think they fall short of user requirements. In particular, we feel that existing models have wrongly assumed that it is not necessary to customize applications for an intermittently connected environment. Then we present our intelligently autonomous model, and give a very brief report of our experiences with a prototype of it.<>