cx_freeze using sys.stdout.flush() and multiprocessing - python-3.x

Question

I am using python 3.4.2 with cx_freeze 4.3.4 (all 64 bit)
the program I have created works fine under python but when frozen, it starts giving me problems with
sys.stdout.flush()
AttributeError:'TypeNone' object has not attribute'flush'
using methods recommended on here, I have managed to reduce the problem to a Traceback message which flashes on the screen for a few seconds before disappearing. how can I resolve this issue.
Windows Error Screen shot
The stdout.flush is only called when it meets the BREAK command in the multiprocessing section of the code.
Any suggestions to either suppress/redirect the error to my log file or help resolve the source of the problem would be greatly appreciated.
Karl
class vDiags(tk.Tk):
def __init__(self, *args, **kwargs):
tk.Tk.__init__(self, *args, **kwargs)
tk.Tk.wm_title(self, "Diagnostics")
do stuff ............
start_job()
def pinger(job_q, mdic, ping, result_q):
devnull = open(os.devnull, 'w')
logger.info("Starting Pinger")
while True:
ip = job_q.get()
if ip is None:
logger.info("No IP address, finishing early")
break
test_result = {}
try:
if is_valid_ipv4_address(ip) is True:
do more stuff.........
def start_job():
logger.info("Starting start_Job")
pool_size = Variables.poll_size
logger.info("Pool size %s" % pool_size)
jobs = multiprocessing.Queue()
logger.info("Jobs %s" % jobs)
results = multiprocessing.Queue()
logger.info("results %s" % results)
manager = multiprocessing.Manager()
logger.info("manager %s" % manager)
manager_test_dict = manager.dict()
logger.info("manager_test_dict %s" % manager_test_dict)
for key, val in Variables.test_dic.items():
manager_test_dict[key] = val
pool = [multiprocessing.Process(target=pinger, args=(jobs, manager_test_dict, Variables.ping, results))
for i in range(pool_size)
]
for p in pool:
logger.info("p in pool %s" % p)
p.start()
for i in Variables.source_ip:
logger.info("Source IP:> %s" % i)
jobs.put(i)
for p in pool:
logger.info("p in pool (jobs) %s" % p)
jobs.put(None)
for p in pool:
logger.info("p in pool (join) %s" % p)
p.join()
logger.info("Move Results to new Variable")
logger.info(results.qsize())
while not results.empty():
Variables.test_result_raw = updatetree(Variables.test_result_raw, results.get())
logger.info("Finished start_Job")
class fakestd(object):
def write(self, string):
logger.info("write %s" %string)
pass
def flush(self):
logger.info("Flush %s " % self )
pass
if __name__ == '__main__':
# ********** Main App **********
sys.stdout = fakestd()
sys.stderr = fakestd()
multiprocessing.freeze_support()
logger.info("tkinter Installed Version %s" % tk.TkVersion)
app = vDiags()
app.geometry("1280x720")
app.mainloop()
# ********** Main App **********

Answer 1

I am hitting this issue right now (just migrated to multiprocessing from threading)
It appears it is a questionable bug in the multiprocessing module & the multiprocessing.freeze_support() call.
https://bugs.python.org/issue20607
There are reports it is still present in py35 but that stands a chance to be fixed at the py source level.
From a py34 point of view the ONE file I needed to update was:
C:\Python34\Lib\multiprocessing\process.py
diff -u process_orig.py process.py
--- process_orig.py 2016-12-12 12:42:01.568186800 +0000
+++ process.py 2016-12-12 12:37:28.971929900 +0000
## -270,8 +270,14 ##
traceback.print_exc()
finally:
util.info('process exiting with exitcode %d' % exitcode)
- sys.stdout.flush()
- sys.stderr.flush()
+# ---------- JRB modify
+ #sys.stdout.flush() # jrb edit to fix cx_freeze
+ #sys.stderr.flush() # jrb edit to fix cx_freeze
+ if sys.stdout is not None:
+ sys.stdout.flush()
+ if sys.stderr is not None:
+ sys.stderr.flush()
+# ---------- JRB modify
return exitcode