performance issues with groovy getProperty() method - groovy

Question

suppose that i have a method which has the signature like
def myAction(objet,String propertyName);
i just want to retrieve the property value from objet, i wrote two versions
version 1: I construct the getter method and invoke it on objet
original = propertyName
prefix = original.substring(0,1).toUpperCase()
suffix = original.substring(1,original.length())
methodName = new StringBuilder('get').append(prefix).append(suffix).toString()
value = domain.metaClass.invokeMethod(objet,methodName,null)
version 2: I use directly getProperty method supplied by GroovyObject
value = objet.getProperty(propertyName)
Witch one has better performance?
thinks : )

Answer 1

We can test it to find out using GBench...
I came up with 4 different methods:
#Grab( 'org.gperfutils:gbench:0.4.2-groovy-2.1' )
// Your 'version 1' method
def method1( object, String propertyName ) {
original = propertyName
prefix = original.substring(0,1).toUpperCase()
suffix = original.substring(1,original.length())
methodName = new StringBuilder('get').append(prefix).append(suffix).toString()
assert 'tim' == object.metaClass.invokeMethod(object,methodName,null)
}
// Your 'version 2' method
def method2( object, String propertyName ) {
assert 'tim' == object.getProperty( propertyName )
}
// The same as method 1, but more Groovy
def method3( object, String propertyName ) {
assert 'tim' == object.metaClass.invokeMethod( object, "get${propertyName.capitalize()}",null)
}
// And get the property with the Groovy String templating
def method4( object, String propertyName ) {
assert 'tim' == object."$propertyName"
}
We can then define a class we're going to test:
class Test {
String name = 'tim'
}
def o = new Test()
And then we can run a benchmark over all 4 different versions:
benchmark {
'method1' {
method1( o, 'name' )
}
'method2' {
method2( o, 'name' )
}
'method3' {
method3( o, 'name' )
}
'method4' {
method4( o, 'name' )
}
}.prettyPrint()
On my machine, this prints out:
Environment
===========
* Groovy: 2.1.6
* JVM: Java HotSpot(TM) 64-Bit Server VM (23.25-b01, Oracle Corporation)
* JRE: 1.7.0_25
* Total Memory: 255.125 MB
* Maximum Memory: 1095.125 MB
* OS: Mac OS X (10.8.4, x86_64)
Options
=======
* Warm Up: Auto (- 60 sec)
* CPU Time Measurement: On
user system cpu real
method1 1630 7 1637 1648
method2 429 1 430 435
method3 1368 1 1369 1378
method4 629 1 630 637
So, the fastest is the getProperty one, followed by the object."$propertyName" one, followed by the shorter version of your version 1 method, followed by your original (and slowest) version 1 code
However, I'd argue that method4 is easier to read, and so you have to ask whether it is worth the slight dip in performance