Report known bisect results at the end of each step.

tools/bisect-perf-regression.py

 #!/usr/bin/env python
 # Copyright (c) 2013 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.
 
 """Performance Test Bisect Tool
 
 This script bisects a series of changelists using binary search. It starts at
 a bad revision where a performance metric has regressed, and asks for a last
 known-good revision. It will then binary search across this revision range by
@@ skipping 2219 matching lines @@
             next_revision_data['passed'] = 'Build Failed'
 
           print run_results[0]
 
           # If the build is broken, remove it and redo search.
           revision_list.pop(next_revision_index)
 
           max_revision -= 1
 
         if self.opts.output_buildbot_annotations:
+          self._PrintPartialResults(results)
           bisect_utils.OutputAnnotationStepClosed()
     else:
       # Weren't able to sync and retrieve the revision range.
       results['error'] = 'An error occurred attempting to retrieve revision '\
                          'range: [%s..%s]' % (good_revision, bad_revision)
 
     return results
 
+  def _PrintPartialResults(self, results_dict):
+    revision_data = results_dict['revision_data']
+    revision_data_sorted = sorted(revision_data.iteritems(),
+                                  key = lambda x: x[1]['sort'])
+    results_dict = self._GetResultsDict(revision_data, revision_data_sorted)
+    first_working_revision = results_dict['first_working_revision']
+    last_broken_revision = results_dict['last_broken_revision']
+
+    self._PrintTestedCommitsTable(revision_data_sorted,
+                                  results_dict['first_working_revision'],
+                                  results_dict['last_broken_revision'],
+                                  100, final_step=False)
+
   def _PrintConfidence(self, results_dict):
     # The perf dashboard specifically looks for the string
     # "Confidence in Bisection Results: 100%" to decide whether or not
     # to cc the author(s). If you change this, please update the perf
     # dashboard as well.
     print 'Confidence in Bisection Results: %d%%' % results_dict['confidence']
 
   def _PrintBanner(self, results_dict):
     print
     print " __o_\___          Aw Snap! We hit a speed bump!"
@@ skipping 30 matching lines @@
       except IndexError:
         print
         print 'Failed to parse svn revision from body:'
         print
         print info['body']
         print
     print 'Commit  : %s' % cl
     print 'Date    : %s' % info['date']
 
   def _PrintTestedCommitsTable(self, revision_data_sorted,
-      first_working_revision, last_broken_revision, confidence):
+      first_working_revision, last_broken_revision, confidence,
+      final_step=True):
     print
-    print 'Tested commits:'
+    if final_step:
+      print 'Tested commits:'
+    else:
+      print 'Partial results:'
     print '  %20s  %40s  %12s %14s %13s' % ('Depot'.center(20, ' '),
         'Commit SHA'.center(40, ' '), 'Mean'.center(12, ' '),
         'Std. Error'.center(14, ' '), 'State'.center(13, ' '))
     state = 0
     for current_id, current_data in revision_data_sorted:
       if current_data['value']:
         if (current_id == last_broken_revision or
             current_id == first_working_revision):
           # If confidence is too low, don't add this empty line since it's
           # used to put focus on a suspected CL.
-          if confidence:
+          if confidence and final_step:
             print
           state += 1
+          if state == 2 and not final_step:
+            # Just want a separation between "bad" and "good" cl's.
+            print
 
         state_str = 'Bad'
-        if state == 1:
+        if state == 1 and final_step:
           state_str = 'Suspected CL'
         elif state == 2:
           state_str = 'Good'
 
         # If confidence is too low, don't bother outputting good/bad.
         if not confidence:
           state_str = ''
         state_str = state_str.center(13, ' ')
 
         std_error = ('+-%.02f' %
@@ skipping 47 matching lines @@
     print 'Average build time : %s' % datetime.timedelta(
         seconds=int(step_build_time_avg))
     print 'Average test time  : %s' % datetime.timedelta(
         seconds=int(step_perf_time_avg))
 
   def _PrintWarnings(self):
     if not self.warnings:
       return
     print
     print 'WARNINGS:'
-    for w in self.warnings:
+    for w in set(self.warnings):
       print '  !!! %s' % w
 
   def _GetResultsDict(self, revision_data, revision_data_sorted):
     # Find range where it possibly broke.
     first_working_revision = None
     first_working_revision_index = -1
     last_broken_revision = None
     last_broken_revision_index = -1
 
     for i in xrange(len(revision_data_sorted)):
@@ skipping 571 matching lines @@
       # The perf dashboard scrapes the "results" step in order to comment on
       # bugs. If you change this, please update the perf dashboard as well.
       bisect_utils.OutputAnnotationStepStart('Results')
     print 'Error: %s' % e.message
     if opts.output_buildbot_annotations:
       bisect_utils.OutputAnnotationStepClosed()
   return 1
 
 if __name__ == '__main__':
   sys.exit(main())