Tkinter understanding mainloop
I'm using an MVC / MVA design pattern, with multiple types of "views". One type is a "GuiView", which is a Tk window. I pass a view reference to my window object which does things like link buttons back to view functions (which the adapter / controller class also calls).
In order to do that, the view object constructor needed to be completed prior to creating the window object. After creating and displaying the window, I wanted to do some initial tasks with the view automatically. At first I tried doing them post mainloop(), but that didn't work because mainloop() blocked!
As such, I created the window object and used tk.update() to draw it. Then, I kicked off my initial tasks, and finally started the mainloop.
import Tkinter as tk
class Window(tk.Frame):
def __init__(self, master=None, view=None ):
tk.Frame.__init__( self, master )
self.view_ = view
""" Setup window linking it to the view... """
class GuiView( MyViewSuperClass ):
def open( self ):
self.tkRoot_ = tk.Tk()
self.window_ = Window( master=None, view=self )
self.window_.pack()
self.refresh()
self.onOpen()
self.tkRoot_.mainloop()
def onOpen( self ):
""" Do some initial tasks... """
def refresh( self ):
self.tkRoot_.update()
while 1:
root.update()
... is (very!) roughly similar to:
root.mainloop()
The difference is, mainloop is the correct way to code and the infinite loop is subtly incorrect. I suspect, though, that the vast majority of the time, either will work. It's just that mainloop is a much cleaner solution. After all, calling mainloop is essentially this under the covers:
while the_window_has_not_been_destroyed():
wait_until_the_event_queue_is_not_empty()
event = event_queue.pop()
event.handle()
... which, as you can see, isn't much different than your own while loop. So, why create your own infinite loop when tkinter already has one you can use?
Put in the simplest terms possible: always call mainloop as the last logical line of code in your program. That's how Tkinter was designed to be used.
tk.mainloop() blocks. It means that execution of your Python commands halts there. You can see that by writing:
while 1:
ball.draw()
tk.mainloop()
print("hello") #NEW CODE
time.sleep(0.01)
You will never see the output from the print statement. Because there is no loop, the ball doesn't move.
On the other hand, the methods update_idletasks() and update() here:
while True:
ball.draw()
tk.update_idletasks()
tk.update()
...do not block; after those methods finish, execution will continue, so the while loop will execute over and over, which makes the ball move.
An infinite loop containing the method calls update_idletasks() and update() can act as a substitute for calling tk.mainloop(). Note that the whole while loop can be said to block just like tk.mainloop() because nothing after the while loop will execute.
However, tk.mainloop() is not a substitute for just the lines:
tk.update_idletasks()
tk.update()
Rather, tk.mainloop() is a substitute for the whole while loop:
while True:
tk.update_idletasks()
tk.update()
Response to comment:
Here is what the tcl docs say:
Update idletasks
This subcommand of update flushes all currently-scheduled idle events from Tcl's event queue. Idle events are used to postpone processing until “there is nothing else to do”, with the typical use case for them being Tk's redrawing and geometry recalculations. By postponing these until Tk is idle, expensive redraw operations are not done until everything from a cluster of events (e.g., button release, change of current window, etc.) are processed at the script level. This makes Tk seem much faster, but if you're in the middle of doing some long running processing, it can also mean that no idle events are processed for a long time. By calling update idletasks, redraws due to internal changes of state are processed immediately. (Redraws due to system events, e.g., being deiconified by the user, need a full update to be processed.)
APN As described in Update considered harmful, use of update to handle redraws not handled by update idletasks has many issues. Joe English in a comp.lang.tcl posting describes an alternative:
So update_idletasks() causes some subset of events to be processed that update() causes to be processed.
From the update docs:
update ?idletasks?
The update command is used to bring the application “up to date” by entering the Tcl event loop repeatedly until all pending events (including idle callbacks) have been processed.
If the idletasks keyword is specified as an argument to the command, then no new events or errors are processed; only idle callbacks are invoked. This causes operations that are normally deferred, such as display updates and window layout calculations, to be performed immediately.
KBK (12 February 2000) -- My personal opinion is that the [update] command is not one of the best practices, and a programmer is well advised to avoid it. I have seldom if ever seen a use of [update] that could not be more effectively programmed by another means, generally appropriate use of event callbacks. By the way, this caution applies to all the Tcl commands (vwait and tkwait are the other common culprits) that enter the event loop recursively, with the exception of using a single [vwait] at global level to launch the event loop inside a shell that doesn't launch it automatically.
The commonest purposes for which I've seen [update] recommended are:
- Keeping the GUI alive while some long-running calculation is executing. See Countdown program for an alternative. 2) Waiting for a window to be configured before doing things like geometry management on it. The alternative is to bind on events such as that notify the process of a window's geometry. See Centering a window for an alternative.
What's wrong with update? There are several answers. First, it tends to complicate the code of the surrounding GUI. If you work the exercises in the Countdown program, you'll get a feel for how much easier it can be when each event is processed on its own callback. Second, it's a source of insidious bugs. The general problem is that executing [update] has nearly unconstrained side effects; on return from [update], a script can easily discover that the rug has been pulled out from under it. There's further discussion of this phenomenon over at Update considered harmful.
.....
Is there any chance I can make my program work without the while loop?
Yes, but things get a little tricky. You might think something like the following would work:
class Ball:
def __init__(self, canvas, color):
self.canvas = canvas
self.id = canvas.create_oval(10, 10, 25, 25, fill=color)
self.canvas.move(self.id, 245, 100)
def draw(self):
while True:
self.canvas.move(self.id, 0, -1)
ball = Ball(canvas, "red")
ball.draw()
tk.mainloop()
The problem is that ball.draw() will cause execution to enter an infinite loop in the draw() method, so tk.mainloop() will never execute, and your widgets will never display. In gui programming, infinite loops have to be avoided at all costs in order to keep the widgets responsive to user input, e.g. mouse clicks.
So, the question is: how do you execute something over and over again without actually creating an infinite loop? Tkinter has an answer for that problem: a widget's after() method:
from Tkinter import *
import random
import time
tk = Tk()
tk.title = "Game"
tk.resizable(0,0)
tk.wm_attributes("-topmost", 1)
canvas = Canvas(tk, width=500, height=400, bd=0, highlightthickness=0)
canvas.pack()
class Ball:
def __init__(self, canvas, color):
self.canvas = canvas
self.id = canvas.create_oval(10, 10, 25, 25, fill=color)
self.canvas.move(self.id, 245, 100)
def draw(self):
self.canvas.move(self.id, 0, -1)
self.canvas.after(1, self.draw) #(time_delay, method_to_execute)
ball = Ball(canvas, "red")
ball.draw() #Changed per Bryan Oakley's comment
tk.mainloop()
The after() method doesn't block (it actually creates another thread of execution), so execution continues on in your python program after after() is called, which means tk.mainloop() executes next, so your widgets get configured and displayed. The after() method also allows your widgets to remain responsive to other user input. Try running the following program, and then click your mouse on different spots on the canvas:
from Tkinter import *
import random
import time
root = Tk()
root.title = "Game"
root.resizable(0,0)
root.wm_attributes("-topmost", 1)
canvas = Canvas(root, width=500, height=400, bd=0, highlightthickness=0)
canvas.pack()
class Ball:
def __init__(self, canvas, color):
self.canvas = canvas
self.id = canvas.create_oval(10, 10, 25, 25, fill=color)
self.canvas.move(self.id, 245, 100)
self.canvas.bind("<Button-1>", self.canvas_onclick)
self.text_id = self.canvas.create_text(300, 200, anchor='se')
self.canvas.itemconfig(self.text_id, text='hello')
def canvas_onclick(self, event):
self.canvas.itemconfig(
self.text_id,
text="You clicked at ({}, {})".format(event.x, event.y)
)
def draw(self):
self.canvas.move(self.id, 0, -1)
self.canvas.after(50, self.draw)
ball = Ball(canvas, "red")
ball.draw() #Changed per Bryan Oakley's comment.
root.mainloop()