fixed preparation of inputs for benchmark

README.md

@@ -37,13 +37,18 @@ To run the main methods:
     $ ps -ef | sbt "runMain edu.luc.etl.osdi.processtree.scala.groupby.Main"
     $ ps -ef | sbt "runMain edu.luc.etl.osdi.processtree.scala.fold.Main"
 
+To generate larger data sets for testing:
+
+    $ sbt "test:runMain edu.luc.etl.osdi.processtree.scala.fakeps.Main 100000" > data.txt
+
 To run the benchmarks:
 
+    $ sbt 'test:runMain edu.luc.etl.osdi.processtree.scala.fakeps.BenchmarkFold'
+    $ sbt 'test:runMain edu.luc.etl.osdi.processtree.scala.fakeps.BenchmarkArray'
     $ sbt 'test:runMain edu.luc.etl.osdi.processtree.scala.mutable.Benchmark -silent'
     $ sbt 'test:runMain edu.luc.etl.osdi.processtree.scala.groupby.Benchmark -silent'
     $ sbt 'test:runMain edu.luc.etl.osdi.processtree.scala.fold.Benchmark -silent'
 
 On Windows, if you installed [Git](http://git-scm.com/) with the recommended
-third option,
-*Use Git and optional Unix tools from the Windows Command Prompt*,
+third option, *Use Git and optional Unix tools from the Windows Command Prompt*,
 then you will have a `ps` command available.

src/main/scala/fold/Main.scala

@@ -17,6 +17,6 @@ trait FoldTreeBuilder extends common.TreeBuilder {
     processes.foldLeft(Map.empty: ProcessTree) { (m, p) =>
       val ppid = p._2
       val children = m.getOrElse(ppid, Vector.empty) :+ p
-      m.updated(ppid, children)
+      m + (ppid -> children)
     }
 }

src/test/scala/common/Benchmark.scala

@@ -9,14 +9,16 @@ import org.scalameter.api._
   */
 abstract class Benchmark(label: String) extends Bench.LocalTime with TreeBuilder {
 
-  val sizes: Gen[Int] = Gen.exponential("processes")(100, 1000000, 10)
+  val sizes: Gen[Int] = Gen.exponential("processes")(10, 1000000, 10)
 
-  val inputs: Gen[Iterator[Process]] = sizes map fakeps.fakePs cached
+  /** Inputs cached as (immutable) sequences so we can reuse them. */
+  val inputs: Gen[Seq[Process]] = sizes.map(n => fakeps.fakePs(n).toSeq).cached
 
   performance of label in {
     measure method "buildTree" in {
       using (inputs) in { ps =>
-        buildTree(ps)
+        // need to get a fresh iterator over the cached sequence
+        buildTree(ps.iterator)
       }
     }
   }

src/test/scala/fakeps/Main.scala

@@ -15,7 +15,6 @@ object Main extends App {
     System.exit(1)
   }
 
-  val n = arg.get
   val ps = fakePs(arg.get)
   println("%-10s %-10s %-10s".format("PID", "PPID", "CMD"))
   for ((pid, ppid, cmd) <- ps)

src/test/scala/fakeps/benchmarks.scala

@@ -5,7 +5,7 @@ import org.scalameter.api._
 
 abstract class Benchmark(val f: Int => Iterator[(Int, Int)], val label: String) extends Bench.LocalTime {
 
-  val sizes: Gen[Int] = Gen.exponential("processes")(10, 1000000, 10)
+  val sizes: Gen[Int] = Gen.exponential("processes")(10, 100000, 10)
 
   measure method label in {
     using (sizes) in { n =>

src/test/scala/fakeps/fakeps.scala

@@ -13,20 +13,20 @@ package object fakeps {
     * Generates a barebones process tree (ppid -> pid*) of size n
     * using an immutable implementation.
     */
-  def fakePsFoldSlow(n: Int): Iterator[(Int, Int)] = inverseEdges {
+  def fakePsFoldSlow(n: Int): Iterator[(Int, Int)] = reverseEdges {
     (2 to n).foldLeft {
       Map(0 -> Seq(1), 1 -> Seq.empty)
     } { (ps, nextPid) =>
       val randomPid = 1 + Random.nextInt(nextPid - 1)
-      ps + (nextPid -> ps.getOrElse(nextPid, Seq.empty), randomPid -> (nextPid +: ps(randomPid)))
+      ps + (randomPid -> (nextPid +: ps(randomPid)), nextPid -> Seq.empty)
     }
   }
 
   /**
     * Generates a barebones process tree (ppid -> pid*) of size n
     * using an immutable implementation.
     */
-  def fakePsFold(n: Int): Iterator[(Int, Int)] = inverseEdges {
+  def fakePsFold(n: Int): Iterator[(Int, Int)] = reverseEdges {
     (2 to n).foldLeft {
       Map(0 -> Seq(1), 1 -> Seq.empty)
     } { (ps, nextPid) =>
@@ -39,7 +39,7 @@ package object fakeps {
     * Generates a barebones process tree (ppid -> pid*) of size n
     * using a mutable implementation.
     */
-  def fakePsMutable(n: Int): Iterator[(Int, Int)] = inverseEdges {
+  def fakePsMutable(n: Int): Iterator[(Int, Int)] = reverseEdges {
     val ps = MMap(0 -> ArrayBuffer(1), 1 -> ArrayBuffer.empty[Int])
     (2 to n) foreach { nextPid =>
       val randomPid = 1 + Random.nextInt(nextPid - 1)
@@ -57,14 +57,15 @@ package object fakeps {
     val ps = ArrayBuffer.fill(n + 1)(ArrayBuffer.empty[Int])
     ps(0) += 1
     (2 to n) foreach { nextPid =>
-      ps(1 + Random.nextInt(nextPid - 1)) += nextPid
+      val randomPid = 1 + Random.nextInt(nextPid - 1)
+      ps(randomPid) += nextPid
     }
-    for (ppid <- (0 to n).iterator; pid <- ps(ppid).iterator) yield (pid, ppid)
+    for (ppid <- ps.indices.iterator ; pid <- ps(ppid).iterator) yield (pid, ppid)
   }
 
   /** Converts a tree (ppid -> pid*) into an iterator of pid -> ppid edges. */
-  def inverseEdges(m: Map[Int, Iterable[Int]]): Iterator[(Int, Int)] =
-    for (ppid <- m.keys.iterator; pid <- m(ppid).iterator) yield (pid, ppid)
+  def reverseEdges(m: Map[Int, Iterable[Int]]): Iterator[(Int, Int)] =
+    for (ppid <- m.keys.iterator ; pid <- m(ppid).iterator) yield (pid, ppid)
 
   /** Adds a command string to each pid -> ppid edge. */
   def addCmd(i: Iterator[(Int, Int)], s: String): Iterator[Process] =
@@ -74,5 +75,5 @@ package object fakeps {
   def addCmd(i: Iterator[(Int, Int)]): Iterator[Process] = addCmd(i, "Fake Process")
 
   /** Generates the fake ps command output. */
-  def fakePs(n: Int) = addCmd(fakePsArray(n))
+  def fakePs(n: Int) = addCmd(fakePsFold(n))
 }