Tkinter Grid Columnspan ignored

Question

Consider the following python script

#!/usr/bin/env python
from Tkinter import Tk, Label

width = SOME_VALUE_HERE

root = Tk()
label1 = Label(root, text='1 columns wide')
label2 = Label(root, text='%i columns wide' % width)

label1.grid()
label2.grid(row=0,column=1,columnspan=width)

root.mainloop()

When I run this, no matter what value is set for 'SOME_VALUE_HERE', both labels take up half the window, regardless of whether or not Grid.columnconfigure is called, or the sticky parameter is used in grid().

Unless I've overlooked something, I would have thought that setting the columnspan would force the second label to be 'SOME_VALUE_HERE' times as wide as the first.

Have I misunderstood how grid works? How would I go about achieving this behavior?

Answer 1

By default, empty grid column are zero width, so you described the following table. Grid geometry manager will by default try to optimize the screen real estate used by your application. It will integrate all the constraint and produce the fittest layout.

+---------------+---------------++++
|       0       |       1       |||| <-- 2,3,4 empty, 0 width
+---------------+---------------++++
| 1 column wide | 4 column wide    |
+---------------+---------------++++

To provide strict proportional column width, you have to use the uniform option of columnconfigure. uniform takes an arbitrary value to designate the group of the column that share these proportions, and the weight argument is used to properly handle widget resizing.

label1.grid(row=0, column=0)
label2.grid(row=0,column=1, columnspan=width)
for i in range(width+1):
    root.grid_columnconfigure(i, weight=1, uniform="foo")

Note that with only these two labels, you could achieve the same layout by adjusting the width of column 1. Differences will occur still while you populate column 2,3,4...

label2.grid(row=0,column=1) #no columnspan
root.grid_columnconfigure(0, weight=1, uniform="foo")
root.grid_columnconfigure(1, weight=width, uniform="foo")