Threading and serial port access

Diez B. Roggisch deets at web.de
Sat Jun 18 09:24:18 EDT 2005


willie at macleod-group.com wrote:
> Hi,
> 
> I'm writing a program which requires the use of three serial ports and
> one parallel port.  My application has a scanning devices on each port,
> which I can access fine with pyserial.  However, I'm unsure of how
> exactly I should be designing the program, I thought I could use
> threading to start class:
> 
> class scanner(Thread):
>         def __init__(self,port):
>                 Thread.__init__(self)
>                 self.port = port
>         def scan(self):
>                 ser = serial.Serial(port)
>                 print ser.portstr
>                 id = ser.read(12)
>                 ser.close
> 
> But this doesn't work as I thought when I call it like:
> 
> for port in range(0,totalserialports):  # loop through all serial ports
>         print "starting thread for port %d" %(port)
>         NewThread = scanner(port)
>         NewThread.scan()
>         NewThread.start()
> 
> I get:
> 
> starting thread for port 0
> /dev/ttyS0
> 
> Now, I know that I haven't specified any port timeouts, but I don't
> want it to timeout, I want to open each port and keep it open
> indefinately.  Threading seems to block waiting for the read from the
> serial port.  How can I open every serial port at the same time, read
> from it, do an action and then go back to it?  Anyone got any good
> documentation sources for threading that explain things clearly and
> gives examples? (I'm running python 2.3.4)

The problem is not your code (as far as I can see that w/o running it), 
but that you didn't understand the threading API. You have to overload 
the run-method of a Thread object and then start your thread - which 
will result in executing the run method in a separate thread.

Like this:

class Poll(threading.Thread):
      def __init__(self):
          threading.Thread.__init__(self)
          self.setDaemon(True)
          self.running = True

       def run(self):
           while self.running:
               .... # do whatever you want here


for t in [Poll() for i in xrange(3)]:
      t.start()


Apart from that the approach you use is wasting resources - if you are 
concerned about that (or better style...) use e.g. twisted with the 
serial and parallel support and its so-called select reactor. The idea 
behind that concept is that the OS is responsible for scannig IO-Ports. 
It notifies an application about newly arrived data by the system 
function select - which means that your program sits and waits not 
cosuming any resources until actual data arrives. See the module select 
for an overview, and google for "python twisted serial".

regards,
Diez



More information about the Python-list mailing list