[Findit] Fix 1500 byte cap on failed_steps_and_tests and output_nodes

appengine/findit/model/wf_failure_group.py

 # Copyright 2016 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.
 
 from google.appengine.ext import ndb
 
 from model.base_build_model import BaseBuildModel
 
 
 class WfFailureGroup(BaseBuildModel):
@@ skipping 13 matching lines @@
 
   @staticmethod
   def Get(master_name, builder_name, build_number):  # pragma: no cover
     return WfFailureGroup._CreateKey(
         master_name, builder_name, build_number).get()
 
   # Integer representation for build failure type.
   # Refer to common/waterfall/failure_type.py for all the failure types.
   build_failure_type = ndb.IntegerProperty(indexed=True)
 
+  # When the group was created.
+  created_time = ndb.DateTimeProperty(indexed=True)
+
   # The blame list of CLs that make up the regression range for this group.
   blame_list = ndb.JsonProperty(indexed=False, compressed=True)
 
   # The list of compile failure output nodes (from signals).
   # Only not None if this group represents a compile failure.
-  output_nodes = ndb.JsonProperty(indexed=True)
+  output_nodes = ndb.JsonProperty(indexed=False)
 
   # A sorted list of lists of the failed steps and tests of a test failure.
   # Only not None if this group represents a test failure.
   # Example:
   # [
   #     ['step_a', 'test1'],
   #     ['step_a', 'test2'],
   #     ['step_b', None]
   # ]
   # ndb.JsonProperty uses json.dumps() without sort_keys=True, which causes
   # dicts that are identical except for their internal key order to have
   # different JSON representations. So, for the failed_steps_and_tests JSON
   # property, a list is used instead of a dict (json.dumps() preserves the order
   # of list elements). This enables a WfFailureGroup query based on equivalent
   # failed_steps_and_tests to return all of the matching results, instead of
   # missing some results. Missing results is a possibility if
   # failed_steps_and_tests used a dict, and the keys of the original dict (that
   # went to the database) were JSONified to string in a different order than the
   # keys of the dict used in the query. For example:
   # '{"step_a": [], "step_y": []}' versus '{"step_y": [], "step_a": []}'.
-  failed_steps_and_tests = ndb.JsonProperty(indexed=True)
+  failed_steps_and_tests = ndb.JsonProperty(indexed=False)
 
   # The sorted list of suspected tuples, if available, from heuristic analysis.
   suspected_tuples = ndb.JsonProperty(indexed=False, compressed=True)
+

[lijeffrey, 2016/08/09 23:00:00]

nit: remove empty line

[josiahk, 2016/08/09 23:10:36]

Done.