From: Phil Tomson <ptkwt@...>
Date: 2001-06-04T06:40:15+09:00
Subject: [ruby-talk:16189] RCR: treatment of ranges within array definitions



[I sent this yesterday via the newsgroup comp.lang.ruby, but I don't think
it ever went anywhere, so now I'm sending to the mailing list - sorry if
it's a duplication]

I've been converting a lot of perl code to Ruby lately.

In perl if I put a range into an array that range is expanded so that the
array contains each element described by the range (Perl doesn't have a
range type).  So in perl if I do:

(0..4,8,10..12)
I get an array that contains:
(0,1,2,3,4,8,10,11,12)

If I declare an array in Ruby this way:
array = [0..4,8,10..12]
array contains three elements, the first and third elements are of range
type.

In Ruby in order to get the same array that I declared in Perl above, I
have to do:
array = (0..4).to_a+[8]+(10..12).to_a
which I'm sure most will agree is a bit cumbersome...

So, here is my poposal:  When a range appears within an array definition
(between '[' and ']') expand the range into an array containing each of
the elements defined by the range unless the range is defined within
parenthesis.

For example:
array = [0..4,8,10..12]
array would contain: 0,1,2,3,4,8,10,11,12

array2 = [(0..4),8,10..12]
array2 would contain: (0..4),8,10,11,12  where the first element is a
range type.

I realize that this could break some existing code, someone might have
some current code with [0..4,3..5,20..24] where they expect an array with
three ranges.  They would have to change it to: [(0..4),(3..5),(20..24)]
in order for it to work that way if this RCR were implemented.

Comments?

Phil