David Alan Black wrote:


> I think there's more to some of the relationship between strings and
> arrays, though -- especially something like join/split, where what an
> array is join'ing itself into is a string.


just to show that

irb(main):007:0> res = '123'.split(''); p res; puts res.class
["1", "2", "3"]
Array
nil
irb(main):008:0> res = [1,2,3].join; p res; puts res.class
"123"
String
nil

> That makes me think there
> must be a hyper/meta/ur notion of stringhood, logically simultaneous
> with (not subsequent to) the notion of arrayhood.


Returning an object of the parent or child class is not contradicting 
that relationship. If a string is a special case of an array and is 
modeled as a subclass, then an array can still return a string and vice 
versa. (or did I misunderstand you)

But really, this [word-class]#each thing bugs me more :)

 > Yes -- I was mainly thinking of optimizations or sugarizations of
 > syntax (such as the "" constructor)

"" as short for String.new; I agree, that's nice...

Tobi

-- 
http://www.pinkjuice.com/