Evaluation - Autonomics Development: A Domain-Specific Aspect Language Approach - page 103

service Desktop {

protocol :jms{
options {
initialContextFactory =
"org.apache.activemq.jndi.ActiveMQInitialContextFactory" ;
persist = true;
sendQueue = "REMOTEJ.SEND" ;
receiveQueue = "temporary" ;
servers = "tcp://localhost:61616,tcp://bookworm:61616" ;
serverThreads = 5;
receiveTimeout = 5000;

}
pointcut void Server.updateData(Object) {
recovery =abort;

}
pointcut void Server.stopViewer(InetAddress) {
recovery =abort;

}
pointcut void Server.startViewer(InetAddress) {
recovery =abort;

}
pointcut byte[] Server.getScreenCapture(InetAddress) {
recovery =abort;

}
pointcut Rectangle Server.getScreenRect(InetAddress) {
recovery =abort;

}

}

}

Figure 6.8. Java Remote Desktop JMS protocol DDL file.

Table 6.1. Remote desktop LOC comparison.

Original RemoteJ % Difference

LOC - JMS 1556 1431 8.03

LOC - REST 1556 1427 8.29

Number of Classes 12 9 25

As well as successfully being able to convert and run the Java remote desktop
application, we were also, by removing the distribution concern, able to greatly
simplify the application as illustrated in Table 6.1 where, using the JMS protocol,
the number of lines of code was reduced by 8.03% and, using the REST protocol,
by 8.29%. In addition, the number of classes was reduced from 12 to 9, a 25%
reduction.