| |
|
We present three models for a data-oriented coordination language and show several relationships between the models. We consider a distributed model and two centralized shared data space models, one preserving the order in which data is produced and a second one not necessarily preserving this order. For example, we show that if dataspaces are multisets them the distributed and the centralized systems are observationally equivalent for any program using read, write and delete operations. The same result does not hold if dataspaces are sets or databases.