From: Dominik Bathon <dbatml@...>
Date: 2005-06-09T08:24:45+09:00
Subject: Re: A little Quiz (Solution)

Obviously no one really likes to decipher undocumented code... (I should  
have known after the "Code Cleaning"-Quiz ;-)
Well, so let me tell you what it does:

As James Britt already figured out, it lets you call one method on all  
elements of an Array. If no block is given this works like collect.  
Instead of

[-1, -2, -3].collect { |x| x.abs }
=> [1, 2, 3]

you can just write

[-1, -2, -3].abs
=> [1, 2, 3]

Since #abs is not part of the Array interface, method_missing is  
triggered. But it does more than collect, it even works on nested Arrays:

[[-1,-2],[-3,4],-2,5].abs
=> [[1, 2], [3, 4], 2, 5]

If you try a method that is part of Array's interface, method_missing  
obviously won't be called:

%w[abc de fghi].size
=> 3

But there is a trick (maybe hack):

%w[abc de fghi]._x_size
=> [3, 2, 4]

The optional "_x_" is automatically stripped. But this also has a  
downside, it doesn't work with nested arrays, but on the other hand you  
can choose an explicit depth:

[[[1,2,3], "iuhiu"], ["iu", [4,5,6]]]._x__x_size
=> [[3, 5], [2, 3]]

But methods also can get arguments, so how are they handled? Some examples:

%w[abc de fgh ijk].ljust(5)
=> ["abc  ", "de   ", "fgh  ", "ijk  "]

%w[abc de fgh ijk].ljust((4..8).to_a)
=> ["abc ", "de   ", "fgh   ", "ijk    "]

%w[abc de fgh ijk].ljust(5, %w[1 2 3 4])
=> ["abc11", "de222", "fgh33", "ijk44"]

%w[abc de fgh ijk].ljust([5, 3, 5, 6], %w[1 2 3 4])
=> ["abc11", "de2", "fgh33", "ijk444"]

I think you can get the idea.

But all this was just the beginning ;-)

My motivation to write this actually was to be able to say

[3, 3].times { |x, y|
     p [x, y]
}

instead of

3.times { |x|
     3.times { |y|
         p [x, y]
     }
}

Well, now I can. And of course you can combine this with the above:

[1,2].upto(4) { |*a| p a }
output:
[1, 2]
[1, 3]
[1, 4]
[2, 2]
[2, 3]
[2, 4]
[3, 2]
[3, 3]
[3, 4]
[4, 2]
[4, 3]
[4, 4]

[1,2].upto([2, 3]) { |*a| p a }
output:
[1, 2]
[1, 3]
[2, 2]
[2, 3]

[%w[a b c], [1, 2, 3]]._x_each { |*a| p a }
output:
["a", 1]
["a", 2]
["a", 3]
["b", 1]
["b", 2]
["b", 3]
["c", 1]
["c", 2]
["c", 3]

If multiple arguments would be yielded to the blocks, they are grouped by  
arrays:

[%w[a b c], 1..3]._x_each_with_index { |*a| p a }
output:
[["a", 0], [1, 0]]
[["a", 0], [2, 1]]
[["a", 0], [3, 2]]
[["b", 1], [1, 0]]
[["b", 1], [2, 1]]
[["b", 1], [3, 2]]
[["c", 2], [1, 0]]
[["c", 2], [2, 1]]
[["c", 2], [3, 2]]

Because I wanted the multidimensional iteration, it doesn't work nice with  
things like collect:

[%w[a b], 1..2]._x_collect { |*a| a }
=> [[[["a", 1], ["a", 2]], [["b", 1], ["b", 2]]], [["b", 1], ["b", 2]]]

Maybe I should add another special prefix ( ;-) ) so that it works more  
like the non block version. Maybe like this:

[%w[a b], 1..2]._y_collect { |*a| a+a }
=> [["aa", "bb"], [2, 4]]

Or maybe use _i_ for the multidimensional iteration and _x_ for the rest...


I also thougt about adding it to Enumerable, but I think it doesn't make  
to much sense for other Enumerables (like Ranges) and if you really need  
it you can always use #to_a.

I realize there are lots of inconsistences in this concept, but I really  
like it generally. So, what do you think?

Dominik