Return 0 for confidence when there's only results for one "good" or one "bad" rev.

tools/bisect-perf-regression_test.py

 # Copyright 2014 The Chromium Authors. All rights reserved.
 # Use of this source code is governed by a BSD-style license that can be
 # found in the LICENSE file.
 
 import unittest
 
 from auto_bisect import source_control as source_control_module
 
 # Special import necessary because filename contains dash characters.
 bisect_perf_module = __import__('bisect-perf-regression')
 
 
 class BisectPerfRegressionTest(unittest.TestCase):
   """Test case for other functions and classes in bisect-perf-regression.py."""
 
   def _AssertConfidence(self, score, bad_values, good_values):
     """Checks whether the given sets of values have a given confidence score.
 
     The score represents our confidence that the two sets of values wouldn't
     be as different as they are just by chance; that is, that some real change
     occurred between the two sets of values.
 
     Args:
       score: Expected confidence score.
       bad_values: First list of numbers.
       good_values: Second list of numbers.
     """
-    # ConfidenceScore takes a list of lists but these lists are flattened.
-    confidence = bisect_perf_module.ConfidenceScore([bad_values], [good_values])
+    # ConfidenceScore takes a list of lists but these lists are flattened
+    # inside the function.
+    confidence = bisect_perf_module.ConfidenceScore(
+        [[v] for v in bad_values],
+        [[v] for v in good_values])
     self.assertEqual(score, confidence)
 
   def testConfidenceScore_ZeroConfidence(self):
     # The good and bad sets contain the same values, so the confidence that
     # they're different should be zero.
     self._AssertConfidence(0.0, [4, 5, 7, 6, 8, 7], [8, 7, 6, 7, 5, 4])
 
   def testConfidenceScore_MediumConfidence(self):
     self._AssertConfidence(80.0, [0, 1, 1, 1, 2, 2], [1, 1, 1, 3, 3, 4])
 
   def testConfidenceScore_HighConfidence(self):
     self._AssertConfidence(95.0, [0, 1, 1, 1, 2, 2], [1, 2, 2, 3, 3, 4])
 
   def testConfidenceScore_VeryHighConfidence(self):
     # Confidence is high if the two sets of values have no internal variance.
     self._AssertConfidence(99.9, [1, 1, 1, 1], [1.2, 1.2, 1.2, 1.2])
     self._AssertConfidence(99.9, [1, 1, 1, 1], [1.01, 1.01, 1.01, 1.01])
 
-  def testConfidenceScore_ImbalancedSampleSize(self):
-    # The second set of numbers only contains one number, so confidence is low.
-    self._AssertConfidence(
-        80.0, [1.1, 1.2, 1.1, 1.2, 1.0, 1.3, 1.2, 1.3],[1.4])
+  def testConfidenceScore_UnbalancedSampleSize(self):
+    # The second set of numbers only contains one number, so confidence is 0.
+    self._AssertConfidence(0.0, [1.1, 1.2, 1.1, 1.2, 1.0, 1.3, 1.2], [1.4])
 
   def testConfidenceScore_EmptySample(self):
     # Confidence is zero if either or both samples are empty.
     self._AssertConfidence(0.0, [], [])
     self._AssertConfidence(0.0, [], [1.1, 1.2, 1.1, 1.2, 1.0, 1.3, 1.2, 1.3])
     self._AssertConfidence(0.0, [1.1, 1.2, 1.1, 1.2, 1.0, 1.3, 1.2, 1.3], [])
 
   def testConfidenceScore_FunctionalTestResults(self):
     self._AssertConfidence(80.0, [1, 1, 0, 1, 1, 1, 0, 1], [0, 0, 1, 0, 1, 0])
     self._AssertConfidence(99.9, [1, 1, 1, 1, 1, 1, 1, 1], [0, 0, 0, 0, 0, 0])
@@ skipping 159 matching lines @@
         source_control, bisect_options)
     results = bisect_instance.Run(bisect_options.command,
                                   bisect_options.bad_revision,
                                   bisect_options.good_revision,
                                   bisect_options.metric)
     bisect_instance.FormatAndPrintResults(results)
 
 
 if __name__ == '__main__':
   unittest.main()