Python global variable not updating in threads

Question

I am writing a code to use scale to update a global value VOL. The initial value of VOL can be passed to function "check_keys", however, when the scale changes its value, VOL remains unchanged in the thread using function "check_keys". Where was I doing wrong?

#!/usr/bin/python
from Xlib.display import Display
from multiprocessing import Process
import threading
import os
import time
import gtk
VOL=1
RUNNING=0
class StatusIcon:
 def __init__(self):
 global RUNNING
 global VOL
 self.statusicon = gtk.StatusIcon()
 self.statusicon.set_from_stock(gtk.STOCK_HOME)
 #self.statusicon.connect("popup-menu", self.right_click_event)
 self.win_build()
 RUNNING=1
 pp = Process(target=self.check_keys)
 pp.start() 
 def quit_program(self, widget):
 global RUNNING
 RUNNING=0
 gtk.main_quit
 def scale_moved(self, widget):
 global VOL
 #print self
 #print VOL
 VOL = self.scale.get_value()
 #print VOL
 def check_keys(self):
 global VOL
 global RUNNING
 disp = Display()
 hold = 0
 samp_inv=0.02
 while RUNNING:
 print "RUNNING"
 print RUNNING
 print "VOL"
 print VOL
 time.sleep(0.2)
 def win_build(self):
 global VOL
 self.window = gtk.Window()
 self.window.set_default_size(400, -1)
 adjustment = gtk.Adjustment(VOL, 0, 150, 5, 10, 0)
 self.scale = gtk.HScale(adjustment)
 self.scale.set_digits(0)
 self.scale.set_update_policy(gtk.UPDATE_DELAYED)
 #VOL=1.
 self.visible=1
 self.window.connect("destroy", lambda w: w.hide_all())
 self.scale.connect("value-changed", self.scale_moved)
 self.window.add(self.scale)
 self.window.set_title("Typewriter Sound Maker")
 self.window.set_position(gtk.WIN_POS_CENTER_ALWAYS)
 self.window.show_all()
StatusIcon()
gtk.main()

• isset: Could you add the missing function: self.right_click_event(), so one can test the code?
• Sean Schon: Thanks isset, right_click_event() is irrelevant, I have removed it so the code can run.

Answer 1

I think the main problem here is: You are using multiprocessing and not threading.

When you use multiprocessing, python will actually fork a new Process, when that happens it can't access the memory of the other process anymore.

You have 2 Options here:

• Select some mechanism to share memory between process, you can find those here: http://docs.python.org/2/library/multiprocessing.html

• or just swap out multiprocessing with threading. http://docs.python.org/2/library/threading.html#module-threading

The down side is that it's not really parallel, because Python is doing the context switch, so one part of your app will be blocked. Anyway, I don't think that concerns you here, because you aren't looking for performance benefits.