/**
* This example shows how easy it is to use the profiler tool
* in Epsilon.
*
* The fibonacci number can be calculated by using a simple recursive
* algorithm which uses an exponentially growing number of executions.
*
* This example is taken from
* http://www.eclipse.org/epsilon/doc/EpsilonProfilingTools.pdf
* See also
* http://epsilonblog.wordpress.com/2007/12/16/cached-operations-in-eol/
*
* HOW TO RUN THIS EXAMPLE:
*
* 1. Open the Epsilon Profiling window:
* In Eclipse, open Window->Show View->Other.. and search for
* "Profiling" in order to open the Epsilon Profiling view.
*
* 2. Launch the program:
* Select fibonacci.launch
* Open the context-menu (right-click) and select Run As-> fibonacci
*
* 3. Refresh the data in the profiler window:
* In the top right corner of the profiling window, click
* "Refresh view". You should now see that target "Program" has
* been called once and "fib" has been called 1973 times.
*
* 4. Repeat with @cached annotation
* Reset the profiler (top right corner of profiling window) and
* uncomment the @cached annotation located just before the
* fib operation.
*
* Repeat step 1 - 3. You should now see that "fib" has
* been called only 16 times.
**/
// Instantiate the profiler through its native type
var profiler : new Native('org.eclipse.epsilon.eol.tools.ProfilerTool');
// Start profiling for a target called "Program"
profiler.start('Program');
// Calculate the 15th fibonacci number
15.fib();
// Stop profiling for the current target
profiler.stop();
/*
* With caching, the fib operation is executed 1973 times, but when
* Epsilon is instructed to cache the return value for each combination
* of object and input parameters, it is only executed 16 times.
*
**/
// Uncomment the annotation @cached to stop Epsilon
// from recalculating known values
// @cached
operation Integer fib() : Integer {
// Start profiling for target "fib" with this integer as context
profiler.start('fib', self);
var fib : Integer;
// Base cases for the fibonacci number
if (self = 1 or self = 0) {
fib = 1;
} else {
// Recursively calculate the fibonacci number
// (i.e. f(n) = f(n-1) + f(n-2) )
fib = (self-1).fib() + (self-2).fib();
}
// Stop profiling for the current target
profiler.stop();
return fib;
}
Check out the code from the SVN:
Once you have checked out/imported the code, to run the example you need to go through the following steps:
This example demonstrates the caching capabilities and the profiling tools provided by Epsilon.
.emf files are Ecore metamodels expressed using the Emfatic textual syntax.
More examples are available in the examples folder of the SVN repository.