Call script through subproce.Popen without blocking explanation needed - subprocess

Question

Basically this code reads from a pipe and constantly prints output without blocking ... Here is the whole code:
1)First script:
if __name__ == '__main__':
for i in range(5):
print str(i)+'\n',
sys.stdout.flush()
time.sleep(1)
2) Second script:
def log_worker(stdout):
while True:
output = non_block_read(stdout).strip()
if output:
print output
def non_block_read(output):
''' even in a thread, a normal read with block until the buffer is full '''
fd = output.fileno()
fl = fcntl.fcntl(fd, fcntl.F_GETFL)
fcntl.fcntl(fd, fcntl.F_SETFL, fl | os.O_NONBLOCK)
try:
return output.read()
except:
return ''
if __name__ == '__main__':
mysql_process = subprocess.Popen(['python','-u', 'flush.py'], stdin=sys.stdin,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE)
thread = Thread(target=log_worker, args=[mysql_process.stdout])
thread.daemon = True
thread.start()
mysql_process.wait()
thread.join(timeout=1)
I would like to know why it works that way :
1) If i take Thread absolutely away , and just call log_worker in the main at also prints everything one by one , but the problem is it hangs after completion without finishing. And i read somewhere that here thread is exactly used for it to finish or more correctly thread dies when it finishes printing something , So why does it work that way ? What thread exactly does here and how ?
2) If i keep the thread but remove mysql_process.wait() and thread.join it prints nothing .... Why ? I read that Popen.wait is meant for its child process to terminate. Set and return returncode attribute. What is the child process here and why/how it is child O_O ?
3) If i only remove thread.join(timeout=1) then it finishes but with error Exception in thread Thread-1 (most likely raised during interpreter shutdown):. Why ? What role .join plays here.
4) I read the documentation of functions used in non_block_read function , but still am confused. Okay it is obvious they take file descriptor and set it to non blocking . The thing i am confused is , on what can i use all those functions , i mean i understand that on files , but how come they use it on stdout O_O ? It's not a file , it's a stream ~~ ?
All this i do to execute a script with subprocess.Popen in tornado script , and constantly send output to client/myself without blocking, so if anyone can kind of help me do that i would really appreciate it , because i can't imagine how to kind of get this output from thread in a way that i can constanly insert it in self.send in tornadio2 ...
def on_message(self, message):
# list = subprocess.Popen([r"ls", "-l"], stdout=subprocess.PIPE)
# list_stdout = list.communicate()[0]
for i in range(1,10):
time.sleep(1)
self.send(i)

Answer 1

With the dev version (3.2), that's realy easy
from tornado.web import RequestHandler,asynchronous
from tornado.process import Subprocess
class home(RequestHandler):
#asynchrounous
def get(self):
seld.sp=Subprocess('date && sleep 5 && date',shell=True,stdout=Subprocess.STREAM)
self.sp.set_exit_callback(self.next)
def next(self,s):
self.sp.stdout.read_until_close(self.finish)
application = ...