User interruption should still create results.html for run-webkit-tests.

Tools/Scripts/webkitpy/layout_tests/controllers/manager.py

Index: Tools/Scripts/webkitpy/layout_tests/controllers/manager.py
diff --git a/Tools/Scripts/webkitpy/layout_tests/controllers/manager.py b/Tools/Scripts/webkitpy/layout_tests/controllers/manager.py
index 5fef70088e0fe95f1629917a32412f5e1d4b2f50..3e9b7bfd12de83378a4a2730074ec6d7fa3cf6d1 100644
--- a/Tools/Scripts/webkitpy/layout_tests/controllers/manager.py
+++ b/Tools/Scripts/webkitpy/layout_tests/controllers/manager.py
@@ -38,6 +38,7 @@ import datetime
 import json
 import logging
 import random
+import signal
 import sys
 import time
 
@@ -51,6 +52,9 @@ from webkitpy.layout_tests.models import test_failures
 from webkitpy.layout_tests.models import test_run_results
 from webkitpy.layout_tests.models.test_input import TestInput
 
+# This mirrors what the shell normally does.
+INTERRUPTED_EXIT_STATUS = signal.SIGINT + 128
+
 _log = logging.getLogger(__name__)
 
 # Builder base URL where we have the archived test results.
@@ -249,10 +253,12 @@ class Manager(object):
 
             results_path = self._filesystem.join(self._results_directory, "results.html")
             self._copy_results_html_file(results_path)
-            if self._options.show_results and (exit_code or (self._options.full_results_html and initial_results.total_failures)):
-                self._port.show_results_html_file(results_path)
-
-        self._printer.print_results(time.time() - start_time, initial_results, summarized_failing_results)
+            if initial_results.interrupted:
+                exit_code = INTERRUPTED_EXIT_STATUS
+            else:
+                if self._options.show_results and (exit_code or (self._options.full_results_html and initial_results.total_failures)):
+                    self._port.show_results_html_file(results_path)
+                self._printer.print_results(time.time() - start_time, initial_results, summarized_failing_results)

[Dirk Pranke, 2013/09/03 20:55:27]

I see that you've changed this so that printer.print_results() is only called if
we weren't interrupted (as a result of my prior comment).

I think this will be fine. However, we also print out  results to stdout for the
buildbot to parse in run_webkit_tests.py:77-79 , and so we'll still execute that
on a ctrl-C, I think.

The catch here is that the run_webkit_tests.py lines *have* to be printed on the
bots if we exit early because of too many errors, but locally, we probably don't
want this printed out if we do a ctrl-C. That would require you to be able to
distinguish the two cases. 

I'm not sure if we really need to fix this or not; mostly it depends on how
annoying the extra logging will be. Can you take a quick look and see what you
think (and whether it's relatively easy to fix)? If not, I'm okay w/ landing
this patch as-is and revisiting separately if we need to fix that.

         return test_run_results.RunDetails(exit_code, summarized_full_results, summarized_failing_results, initial_results, retry_results, enabled_pixel_tests_in_retry)
 
     def _run_tests(self, tests_to_run, tests_to_skip, repeat_each, iterations, num_workers, retrying):