Threading, inheritance

Question

My code is composed of a queue thread class (called MyThread), a class inheriting MyThread (Device) and many class inheriting Device (such as A, B, ...)

Device instantiate a FSM (state machine). So each Device (A, B,..) have the same FSM table.

Now I am trying to implement callbacks of the FSM for/in each device A, B, ...

Here is a sketch of my code:

1.Device module:

MAP = {
'initial' : 'none',
'events': [
    {'name': 'load', 'src': ['OFF', 'ON', 'none'], 'dst': 'LOADED'},
    ],
 'callbacks': {}
}
class Device(MyThread):
    def __init__(self, actor=None):
        MyThread.__init__(self, actor, self.__class__)
        self.MAP =MAP
        self.MAP['callbacks']['onload'] = lambda e: self.initialise()
        self.start()
    def startFSM(self):
        self.started = True
        self.fsm = Fysom(self.MAP)
        self.fsm.startup()

2.MyThread module:

class MyThread(threading.Thread):
    def __init__(self, actor, name):
        threading.Thread.__init__(self, name=name)
        self.actor = actor
    def run(self):
        pass

3.A:

class A(Device):    
    def __init__(self, actor=None):
        Device.__init__(self, actor)
        self.status = None
    def initialise(self):
        print "Here is A"

4.B:

class B(Device):    
    def __init__(self, actor=None):
        Device.__init__(self, actor)
        self.status = None
    def initialise(self):
        print "Here is B"

Current code simplified:

a = A()
b = B()
a.fsm.load()
b.fsm.load()

which will returns :

Here is B
Here is B

instead of :

Here is A
Here is B

or again: a.MAP['callbacks']['onload'] has same memory location as b.MAP['callbacks']['onload']

The point here is to focus on MAP attributes. The issue is that the function saved in self.MAP['callbacks']['onload'] override the last one. But basically they should have one MAP per Device (A, B, ...). And each device should save is own initialise method into his own MAP attribute.

dano · Accepted Answer · 2014-05-19 15:14:28Z

4

The problem is here:

MAP = {
'initial' : 'none',
'events': [
    {'name': 'load', 'src': ['OFF', 'ON', 'none'], 'dst': 'LOADED'},
    ],
 'callbacks': {}
}
class Device(MyThread):
    def __init__(self, actor=None):
        MyThread.__init__(self, actor, self.__class__)
        self.MAP =MAP  # Problem!

You're makingself.MAP a reference to MAP. So every instance of a class that inherits from Device ends up with a self.MAP property that's a reference to the exact same dict (the global MAP dict). What you really want is this, so that each instance gets its own dict:

from copy import deepcopy

MAP = {
'initial' : 'none',
'events': [
    {'name': 'load', 'src': ['OFF', 'ON', 'none'], 'dst': 'LOADED'},
    ],
 'callbacks': {}
}
class Device(MyThread):
    def __init__(self, actor=None):
        MyThread.__init__(self, actor, self.__class__)
        self.MAP = deepcopy(MAP)  # Make sure each MAP is unique.

edited May 19, 2014 at 15:14

answered May 19, 2014 at 14:52

dano

95.5k21 gold badges234 silver badges231 bronze badges

Sign up to request clarification or add additional context in comments.

5 Comments

martineau Over a year ago

I think you would need to make a deep copy of the MAP dictionary otherwise the copies made will share elements -- many of which are containers themselves. Another possibility might to make a Map class and create new instances of it for each Device.

dano Over a year ago

@mertineau Yes, you're right. I'll update the answer.

Katsu Over a year ago

ty for help. I tried it but nothing changed. if for example I do a.MAP['callbacks']['onload'] = 0 it will also overrides b.MAP['callbacks']['onload']. I am wondering if it's not due to polymorphisme.

dano Over a year ago

@Katsu I just updated the answer, I overlooked that MAP.copy() was only doing a shallow copy of the MAP. It changed the example to use deepcopy() which should fix the issue.

Katsu Over a year ago

@dano&mertineau: Nice you are genious

Collectives™ on Stack Overflow

Threading, inheritance

1 Answer 1

5 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

5 Comments

Your Answer

Sign up or log in

Post as a guest

Related