Addressing Andy's comments.

regtest.sh

@@ -171,10 +171,6 @@ _run-one-instance() {
 
     banner "Using gen_counts.R"
 
-    echo tests/gen_counts.R $distr $num_clients $values_per_client $params_file \
-                       $true_map_file "$instance_dir/case"
-
-
     tests/gen_counts.R $distr $num_clients $values_per_client $params_file \
                        $true_map_file "$instance_dir/case"
   else
@@ -232,11 +228,6 @@ _run-one-instance-logged() {
     || log "Test case $test_case_id (instance $test_case_run) failed"
 }
 
-show-help() {
-  tests/gen_sim_input.py || true
-  tests/rappor_sim.py -h || true
-}
-
 make-summary() {
   local dir=$1
   local filename=${2:-results.html}
@@ -273,8 +264,7 @@ _setup-test-instances() {
   local fast_counts=$2
 
   while read line; do
-    for ((i=1; i<=$instances; i++))
-    do 
+    for i in $(seq 1 $instances); do
       read case_name _ <<< $line  # extract the first token
       echo $case_name $i $fast_counts
     done
@@ -304,8 +294,8 @@ _run-tests() {
     func=_run-one-instance  # output to the console
   else
     func=_run-one-instance-logged
-    processors=$(grep -c ^processor /proc/cpuinfo)
-    if test $processors -gt 1; then  # leave one for the OS
+    processors=$(grep -c ^processor /proc/cpuinfo || echo 4)  # POSIX-specific
+    if test $processors -gt 1; then  # leave one CPU for the OS
       processors=$(expr $processors - 1)
     fi
     log "Running $processors parallel processes"

tests/analyze.R

@@ -84,7 +84,7 @@ LoadInputs <- function(prefix_case, prefix_instance, ctx) {
   
   # The line is searched for, and elapsed time is extracted, by make_summary.py.
   # Should the formating or wording change, make_summary must be updated too.
-  Log(c("Inference took", timing["elapsed"], "seconds"))
+  Log("Inference took %.3f seconds", timing[["elapsed"]])
    
   Log("Analysis Results:")
   str(rappor)

tests/gen_counts.R

@@ -134,9 +134,9 @@ ComputePdf <- function(distr, range) {
   } else if (distr == 'zipf1') {
     # Since the distrubition defined over a finite set, we allow the parameter
     # of the Zipf distribution to be 1.
-    pdf <- sapply(1:range, function(x) 1/x)
+    pdf <- sapply(1:range, function(x) 1 / x)
   } else if (distr == 'zipf1.5') {
-    pdf <- sapply(1:range, function(x) 1/x^1.5)
+    pdf <- sapply(1:range, function(x) 1 / x^1.5)
   }  
   else {
     stop(sprintf("Invalid distribution '%s'", distr))

tests/gen_reports.R

@@ -52,11 +52,11 @@ main <- function(argv) {
   
   stopifnot(length(values) == num_clients)
 
-  # Shuffle values randomly (make take a few sec for > 10^8 inputs)
+  # Shuffle values randomly (may take a few sec for > 10^8 inputs)
   values <- sample(values)
 
   # Obtain reports by prefixing values with "v"s. Even slower than shuffling.
-  reports <- paste("v", format(values, trim = TRUE), sep = "")
+  reports <- sprintf("v%d", values)
 
   reports <- cbind(1:num_clients, reports)  # paste together "1 v342"
   reports <- reports[rep(1:nrow(reports), each = reports_per_client), ]

tests/make_summary.py

@@ -243,10 +243,10 @@ def FormatCell1(test_case, test_instance, metrics_file, log_file, plot_file,
     if link_to_plots:
       link = '#{}_{}'.format(test_case, test_instance)  # anchor
     else:
-      link = relpath_report + '/' + 'dist.png'
+      link = os.path.join(relpath_report, 'dist.png')
   else:  # no results likely due to an error, puts a link to the log file
     external_file = log_file
-    link = relpath_report + '/' + 'log.txt'
+    link = os.path.join(relpath_report, 'log.txt')
 
   if os.path.isfile(external_file):
     return '<td><a href="{}">{}</a></td>'.format(link, test_case)

tests/regtest_spec.py

@@ -36,7 +36,7 @@ DISTRIBUTIONS = (
 DISTRIBUTION_PARAMS = (
     # name, num unique values, num clients, values per client
     ('tiny', 100, 1000, 1),  # test for insufficient data
-    ('small', 100, 100000, 1),
+    ('small', 100, 1000000, 1),
     ('medium', 1000, 10000000, 1),
     ('large', 10000, 100000000, 1),
 )