Need ideas and insights on a project...

[true911m]

I would like to solicit your thoughts and ideas (code can come later) about how you would approach a project idea I just tossed together. I chose this as a testbed for me to play with objects, specifically multiple instances of a few specific objects, and as I thought about it, it raised a lot of questions about how to manage, contain, signal all of those objects that do almost the same thing at the same time.

I want to build an OO bingo simulator. While it is not necessarily the most efficient approach, let's academically assume it needs to be object-oriented.

We have 75 ball objects, which can be picked randomly by a machine, and which therefore need an awareness of being picked(), since no ball can come up twice in a game.

We have a scorecard grid with 25 square objects on it, and appropriate values for each column of the grid (B: 1-15,i:16-30, etc.), which are chosen randomly on each card. If we work with all 75 possible numbers (maybe not the best approach), each number needs to know if it's on the bingo card for this game, whether it's matched a number that's already been picked(), and whether it's a Freebie (middle square on the card - 'FREE' instead of a number)

Initial questions:

Can like objects be grouped by some mechanism such that changing an external parameter has them respond to that change, or do they each need to be iterated/manipulated each time to become "aware" of global changes? For example, a number is picked() - can the squares "know" if one matches the new number, or do I need to trigger each one to check?

Maybe I'm thinking in more of a parallel-thread mode here, one thread per object, which remains "active" during the game. Is that way out there? It's probably over my head at this point anyway.

Reel me in; tell me what you think.

[bartonc]

We have 75 ball objects, which can be picked randomly by a machine, and which therefore need an awareness of being picked(), since no ball can come up twice in a game.

Actually, I believe that the machine is the object in question and that balls are data that it dispences. The machine (object) has a better chance of encapsulating awareness of which numbers have been dispensed than a ball does.

We have a scorecard grid with 25 square objects on it, and appropriate values for each column of the grid (B: 1-15,i:16-30, etc.), which are chosen randomly on each card. If we work with all 75 possible numbers (maybe not the best approach), each number needs to know if it's on the bingo card for this game, whether it's matched a number that's already been picked(), and whether it's a Freebie (middle square on the card - 'FREE' instead of a number)

Initial questions:

Can like objects be grouped by some mechanism such that changing an external parameter has them respond to that change, or do they each need to be iterated/manipulated each time to become "aware" of global changes? For example, a number is picked() - can the squares "know" if one matches the new number, or do I need to trigger each one to check?

I've got similar thoughts here. Going to look for some examples

Maybe I'm thinking in more of a parallel-thread mode here, one thread per object, which remains "active" during the game. Is that way out there? It's probably over my head at this point anyway.

Reel me in; tell me what you think.

Do you have Robin Dunn's book: wxPython in Action?

[true911m]

Actually, I believe that the machine is the object in question and that balls are data that it dispences. The machine (object) has a better chance of encapsulating awareness of which numbers have been dispensed than a ball does.



I've got similar thoughts here. Going to look for some examples



Do you have Robin Dunn's book: wxPython in Action?

I haven't read it all, just looked up a few things.

I'm not really interested in the GUI in particular, more the logic. If the GUI objects are easier to build/attach to than abstract code items, that's fine.

[true911m]

The reason for the balls, instead of the machine, as objects is that I want to exercise multiple identical objects in action. JFYI.

[bartonc]

The reason for the balls, instead of the machine, as objects is that I want to exercise multiple identical objects in action. JFYI.

Ok, I'll buy that. So if each ball instance needs to know which number was just picked() here's how I'd (almost) lay out the structure:

# I do this a bit differently, but Mr. Dunn says something like this:
 
class NumberDispenser(object):
    def __init__(self):
        self.clients = []
 
    def RegisterClient(self, client):
        self.clients.append(client)
 
    def UpdateClients(self, data):
        for client in self.clients:
            client.Update(data)     # This part I don't like:
                                    # having to know the name of the client function here
 
class NumberClient(object):
    def __init__(self, server, name):
        self.name = name
        server.RegisterClient(self)
 
    def Update(self, data):
        print self.name, data
 
 
if __name__ == "__main__":
    server = NumberDispenser()
    client1 = NumberClient(server, "client #1")
    client2 = NumberClient(server, "client #2")
 
    # call this in response to some event or whatever
    server.UpdateClients(6545)

[bartonc]

I haven't read it all, just looked up a few things.

I'm not really interested in the GUI in particular, more the logic. If the GUI objects are easier to build/attach to than abstract code items, that's fine.

Actually, I don't use the GUI stuff from the book. I really like (and use similar design) the Model-View-Controler idea. There is also threading discussion toward the end.

[true911m]

Actually, I don't use the GUI stuff from the book. I really like (and use similar design) the Model-View-Controler idea. There is also threading discussion toward the end.

OK, that's not ringing any bells. Can you elaborate on this - what, where, refs?

[true911m]

Ok, I'll buy that. So if each ball instance needs to know which number was just picked() here's how I'd (almost) lay out the structure:

    def UpdateClients(self, data):
        for client in self.clients:
            client.Update(data)     # This part I don't like:
                                    # having to know the name of the client function here
 
...
if __name__ == "__main__":
    server = NumberDispenser()
    client1 = NumberClient(server, "client #1")  # <===
    client2 = NumberClient(server, "client #2")  # <===
 

Regarding that, my original thought, which is admittedly BASIC-esque, was to define an array where each element would become an instance. I don't know how to translate that here, but such a construct would alleviate much of the naming overhead. I guess it would also be redundant, since if I did that I wouldn't need the objects at all. :) But like I said, it's an academic exercise...

Can I use some form of eval() to create something like (this is just off the top of my head, trying to get the idea across):

varhdr = 'client'
 
for i in range(75):
    eval(varhdr+str(i)+' = NumberClient(server, "client #'+str(i)+'")')
 

Don't know if that's close to a proper use of eval; in SB we had a means of constructing dynamic lines of code like that so they could be reused, as in a loop.

This problem is what I was referring to when I mentioned a curiousity about how people would "group"cont rol of these objects.

Actually, the use of a class variable, as discussed in that other thread, may make sense here. Such as, when a number is picked,

NumberClient.LastPicked=32
 
for square in grid:
    # execute instance.CheckPicked code here
    # each instance of NumberClient class now has 
    # a self.LastPicked value of 32 to process against
 
 

[bartonc]

Regarding that, my original thought, which is admittedly BASIC-esque, was to define an array where each element would become an instance. I don't know how to translate that here, but such a construct would alleviate much of the naming overhead. I guess it would also be redundant, since if I did that I wouldn't need the objects at all. :) But like I said, it's an academic exercise...

Can I use some form of eval() to create something like (this is just off the top of my head, trying to get the idea across):

varhdr = 'client'
 
for i in range(75):
    eval(varhdr+str(i)+' = NumberClient(server, "client #'+str(i)+'")')
 

Don't know if that's close to a proper use of eval; in SB we had a means of constructing dynamic lines of code like that so they could be reused, as in a loop.

This problem is what I was referring to when I mentioned a curiousity about how people would "group"cont rol of these objects.

Actually, the use of a class variable, as discussed in that other thread, may make sense here. Such as, when a number is picked,

NumberClient.LastPicked=32
 
for square in grid:
    # execute instance.CheckPicked code here
    # each instance of NumberClient class now has 
    # a self.LastPicked value of 32 to process against
 
 

A list comprehension is what you are looking for. A comprehension is a very handy tool when you don't need much stuff going on in a loop, but you do need elements appended to a list. This technique creates the list object and fills it in all at once. Notice, also, the C-line string formatting. I use it all the time.

class NumberDispenser(object):
    def __init__(self):
        self.clients = []
 
    def RegisterClient(self, client):
        self.clients.append(client)
 
    def UpdateClients(self, data):
        for client in self.clients:
            client.Update(data)     # This part I don't like:
                                    # having to know the name of the client function here
 
class NumberClient(object):
    def __init__(self, server, name):
        self.name = name
        server.RegisterClient(self)
 
    def Update(self, data):
        print self.name, data
 
 
if __name__ == "__main__":
    server = NumberDispenser()
    # list comprehension syntax uses [] on around the expession, generators use ()
    instanceList = [NumberClient(server, "client%d" %i) for i in range(5)]
    for inst in instanceList:
        print inst.name

[bartonc]

The list comprehension here

instanceList = [NumberClient(server, "client%d" %i) for i in range(5)]

is equivalent to

instanceList = []
for i in range(5):
    instanceList.append(NumberClient(server, "client%d" %i))

which may be easier to read. I learned comprehensions so long ago that I have forgotten if the benefits actually outweigh the mashed syntax. If I get a break, I'll look it up and get back to you.