The JAMSEL framework was conceived and created during the development of trading systems for top-tier banks on Wall Street. These trading systems tyically made heavy use of JMS and TIBCO/Rendezvous .
As the benefits of using JMS selectors became clear, the realization occured that the ability to filter messages on the client-side could be beneficial as well. For example, it might be necessary in some applications to oversubscribe on topics/queues and then filter out unnecessary messages at the client. Clearly, for client-side filtering it would be desirable to allow exactly the same selector syntax as used for the server-side filtering.
A selection/filtering framework that was independent of the transport layer would be of great value. For example, it would be beneficial to have a selector framework that would work seamlessly with JMS, TIBCO/RV, IBM/MQ, Sockets, etc. Taking this one step further, the ability to select and filter data in general - no matter what the source of that data was - using a standard selection syntax would be of value. This realization, coupled with the desire to implement the fastest possible selector, was the driving force behind this framework.
The framework is a complete implementation of the JMS selector sub-system.
The framework is completely transport neutral. Any transport/protocol such as JMS, TIBCO/Rendezvous, IBM/MQ may
may be used by the application. In fact, any class that implements the
The framework provides utility classes that allow JMS messages to be filtered.
The framework provides utility classes that allow TIBCO/Rendezvous messages to be filtered.
The framework uses a JAVACC generated parser to parse the specified selector. During the parse, an in-memory representation of the selector is built. At run-time, this in-memory selector is evaluated without unnecessary overhead. In addition, all selector operations are highly optimized for performance and object creation. In informal benchmarks comparing leading JMS provider implementations, the JAMSEL framework was roughly 4 times faster than its nearest competitor.