Added support for improvement_direction to relevant values, which is propogated to chartjson.

tools/telemetry/telemetry/value/summary.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.
 
 from collections import defaultdict
 
 from telemetry.value import failure
 from telemetry.value import merge_values
 from telemetry.value import skip
+from telemetry.value import summarizable
 
 
 class Summary(object):
   """Computes summary values from the per-page-run values produced by a test.
 
   Some telemetry benchmark repeat a number of times in order to get a reliable
   measurement. The test does not have to handle merging of these runs:
   summarizer does it for you.
 
   For instance, if two pages run, 3 and 1 time respectively:
@@ skipping 40 matching lines @@
     values, then summary values.
 
     """
     return self._interleaved_computed_per_page_values_and_summaries
 
   def _ComputePerPageValues(self, all_page_specific_values):
     all_successful_page_values = [
         v for v in all_page_specific_values if not (isinstance(
             v, failure.FailureValue) or isinstance(v, skip.SkipValue))]
 
+    # All successful values must be summarizable.
+    for v in all_successful_page_values:
+      assert isinstance(v, summarizable.SummarizableValue), "Not summarizable: %r" % v

[qyearsley, 2015/01/27 00:50:46]

Looks like it's just over 80 chars. Also, I think single quotes are preferred
over double, quotes, so, this could be formatted as:

assert isinstance(v, summarizable.SummarizableValue), ('Not summariz'
                                                       'able: %r' % v)

or

is_summarizable = isinstance(v, summarizable.SummarizableValue)
assert is_summarizable, 'Not summarizable: %r' % v

or, maybe it's appropriate use a different kind of exception here?

+
     # We will later need to determine how many values were originally created
     # for each value name, to apply a workaround meant to clean up the printf
     # output.
     num_successful_pages_for_value_name = defaultdict(int)
     for v in all_successful_page_values:
       num_successful_pages_for_value_name[v.name] += 1
 
     # By here, due to page repeat options, all_values_from_successful_pages
     # contains values of the same name not only from mulitple pages, but also
     # from the same name. So even if, for instance, only one page ran, it may
@@ skipping 74 matching lines @@
       should_print = True
     else:
       should_print = False
 
     if not should_print:
       return
 
     # Actually save the result.
     self._computed_per_page_values.append(value)
     self._interleaved_computed_per_page_values_and_summaries.append(value)