getNode interprets the node path, and traverses the messages based on the type. (Because of this, I would expect HL7 and XML node paths to be relatively slower than CSV node paths, based on complexity)
I would also expect the messageCache to be faster, since it is stored as a hash map of strings.
to test this hypothesis, I ran and timed a loop with 10,000 getNodes or messageCache.getValues.
The results were:
an average time of 62 milliseconds for the messageCache
and an average of 218 milliseconds for getNode
Therefore, for small values at least, you could say that the messageCache is the more efficient approach. (however, the messageCache trade off is that it uses additional memory to store the value). Worth the trade off, in my opinion.
thanks to Mike Ross for his assistance with this approach!
code below, for those that want to tweak / enhance it:
var visitID; var i;
messageCache.setValue('visitID', source.getNode("PV1-19"));
// Start
var msgCacheStart = new Date();
for (i = 0; i <= 10000; i++) {
visitID = messageCache.getValue('visitID');
}
//Stop
var msgCacheStop = new Date();
var msgCacheTime = Math.abs(msgCacheStart - msgCacheStop);
qie.info("msgCacheTime: " + msgCacheTime);
// Start
var getNodeStart = new Date();
for (i = 0; i <= 10000; i++) {
visitID = source.getNode("PV1-19");
}
//Stop
var getNodeStop = new Date();
var getNodeTime = Math.abs(getNodeStart - getNodeStop);
qie.info("getNodeTime: " + getNodeTime);
qie.info("getNodeTime / msgCacheTime: " + getNodeTime / msgCacheTime);
