Move test value generation to benchmarked expression

kgryte · kgryte · commit 99acc6bf55a5 · 2017-05-01T23:45:48.000-07:00
Previously, the test value was generated during the "setup" phase.
This resulted in benchmarks which were not oranges to oranges
comparable. For other languages, we generate new test values for
each iteration. Hence, to ensure similar tests, we needed to move
test value generation to be part of the benchmark.
diff --git a/tools/snippets/benchmark/r/benchmark.R b/tools/snippets/benchmark/r/benchmark.R
@@ -59,10 +59,8 @@ print_results <- function( elapsed ) {
 #' @examples
 #' elapsed <- benchmark();
 benchmark <- function() {
-	x <- runif( 1 ); # TODO
-
 	# Run the benchmarks:
-	results <- microbenchmark::microbenchmark( TODO( x ), times = iterations );
+	results <- microbenchmark::microbenchmark( TODO( TODO ), times = iterations );
 
 	# Sum all the raw timing results to get a total "elapsed" time:
 	elapsed <- sum( results$time );
