Scala Execution Time Monitor

The SETM project is a small wrapper for the JETM project with the goal to provide an easier and more Scala-like API.

Usage

Simply wrap the timing function around the piece of code that you would like to monitor:

import de.tuilmenau.setm.SETM._

object Main {
  def greeting(s: String) = timing("greeting func") {
    // complex operations, e.g.
    (0 until 100).foreach(i => println(s"Hello $s"))
   }

  def main(args: Array[String]) {

    start()

    timing("program total") {

      val names = timing("create names") { Array("Tick","Trick","Track") }

      for(name <- names)
        greeting(name)

    }

    collect()
  }
}

In the main method, we start the ETM monitor by called start() from the class SETM. We then start doing our work and at the end, we collect the runtime information. The call to collect will print the results on the screen.

We define our function of which we want to monitor execution time, called greeting. The method body is wrapped in the call to timing. This will create a measurement called greeting func. We can also measure whole blocks of code and value or variable assignments.

You will see an output similar to this:

|-------------------|---|---------|-------|-------|-------|
| Measurement Point | # | Average |  Min  |  Max  | Total |
|-------------------|---|---------|-------|-------|-------|
| program total     | 1 |   7.381 | 7.381 | 7.381 | 7.381 |
|   create names    | 1 |   0.028 | 0.028 | 0.028 | 0.028 |
|   greeting func   | 3 |   2.218 | 1.440 | 3.610 | 6.655 |
|-------------------|---|---------|-------|-------|-------|

We see that the measurement point greeting func was executed three times with an average runtime of around 2.2 seconds as well as the min/max and total execution times.

Installation

You can use the sbt assembly plugin that is included in the project to build a fat jar file that you can add to your project. Alternatively, just add this project as a Git submodule and/or sbt sub project.

Limitations

Currently, there is no way to configure the underlying JETM monitor. SETM will use nesting to represent the call stack and print the results to STDOUT using JETM's SimpleTextRenderer. This will be addressed soon.