replacing the NullCulprit class by None

appengine/findit/crash/findit.py

Index: appengine/findit/crash/findit.py
diff --git a/appengine/findit/crash/findit.py b/appengine/findit/crash/findit.py
index e15ab25c08b2cc56128b468fbbb432acfcd925f6..9624ba03a9f8232f4afe7f53eb1235dd7fe0c6dc 100644
--- a/appengine/findit/crash/findit.py
+++ b/appengine/findit/crash/findit.py
@@ -12,7 +12,6 @@ from common import chrome_dependency_fetcher
 from common import constants
 from common import time_util
 from crash.crash_report import CrashReport
-from crash.culprit import NullCulprit
 from model import analysis_status
 from model.crash.crash_config import CrashConfig
 
@@ -277,9 +276,7 @@ class Findit(object):
     """Given a CrashAnalysis ndb.Model, return a Culprit."""
     stacktrace = self.ParseStacktrace(model)
     if stacktrace is None:
-      # TODO(http://crbug.com/659359): refactor things so we don't need
-      # the NullCulprit class.
-      return NullCulprit()
+      return None

[Sharu Jiang, 2016/10/28 20:34:12]

It's ok to return None here, but you need to update the related part in
CrashAnalysisPipeline in the crash_pipeline.py.

And in this case, you still need to push
{'found': False} to clients and update CrashAnalysis model in data store using 
{'found_suspects': False,
 'found_project': False,
 'found_components': False,
 'has_regression_range': False,
 'solution': None}

[wrengr, 2016/10/28 21:11:19]

Indeed, the goal of this CL is to allow returning None here. Shouldn't the
'solution' be the algorithm that was tried and failed to find a culprit, as you
said in the other CL?

[Sharu Jiang, 2016/10/28 23:40:58]

Yes, the algorithm is a better name, however when you update the CrashAnalysis
model, it should be 'solution', we can change that field name, however there
would be a legacy problem... though it won't have much impact right now.

 
     return self._azalea.FindCulprit(CrashReport(
         crashed_version = model.crashed_version,