Gary Oberbrunner committed 95fc746

Merged pull request #38, from eyan: add target name to debug=time

Fixes #2873.

Comments (0)

Files changed (3)

+  From eyan on Bitbucket:
+    - Print target name with command execution time with --debug=time
   From Thomas Berg and Evgeny Podjachev:
     - Fix subprocess spawning on Windows.  Work around a Windows
       bug that can crash python occasionally when using -jN. (#2449)


             finish_time = time.time()
             last_command_end = finish_time
             cumulative_command_time = cumulative_command_time+finish_time-start_time
-            sys.stdout.write("Command execution time:%s:%f seconds\n"%(str(self.node), finish_time-start_time))
+            sys.stdout.write("Command execution time: %s: %f seconds\n"%(str(self.node), finish_time-start_time))
     def do_failed(self, status=2):


 expected_total_time = complete_time - overhead
-pattern = r'Command execution time:(.*):(\d+\.\d+) seconds'
+pattern = r'Command execution time: (.*): (\d+\.\d+) seconds'
 targets = []
 times = []
 for target,time in re.findall(pattern, test.stdout()):
 outside of the 1%% tolerance.
 """ % locals())
-if not within_tolerance(total_time, expected_total_time, 0.15):
+if not within_tolerance(total_time, expected_total_time, 0.20):
     # This tolerance check seems empirically to work fine if there's
     # a light load on the system, but on a heavily loaded system the
     # timings get screwy and it can fail frequently.  Some obvious