ActiveSpace uses the Space abstraction which is a simple JavaSpaces like abstraction which sits neatly on a high performance message bus.
The Space API is specifically designed for high performance streaming and for lightweight and high throughput SEDA style architectures.
The Space API simplifies the JavaSpaces API to create a simple yet powerful API which is capable of reusing all of the features of modern message bus technologies like hierarchial wildcards, headers, routing, rules and so forth along with both queues & publish/subscribe and pluggable Quality of Service features
When using ActiveSpace in a distributed or clustered environment, the creation of a space with or without a predicate will cause the infrastructure to begin buffering up a number of objects in RAM of the client (assuming there are objects availble to be delivered) so that take or listen operations are super fast and not require any synchronous networking latency, locking or contention.
To get a better feel for ActiveSpace try the javadoc or browse the source code
The basic idea is to make a Space more capable of its features and configuration. A space can use a JMS destination, which supports wildcards for creating hierarchies and grouping spaces together. e.g. this allows some consumers to look at a sub-space while other consumers look at larger groups or hierarchies.
A space can work in two primary modes
Also rather than the template mechanism in JavaSpaces, where you use an empty object with public fields as a way of querying a space, we use an SQL 92 query string to filter spaces. So we can create a child space with some predicate applied.
In addition the space can have additional configuration properties like
Finally we can easily support a variety of transaction modes such as
The core put and take operations are quite similar and things work at the POJO level so things are nice and simple. There are a few differences however.