From: Dave Thomas <Dave@...>
Date: 2002-02-13T03:09:58+09:00
Subject: Re: [newbie] make first letter lowercase

matz@ruby-lang.org (Yukihiro Matsumoto) writes:

> |It isn't very orthogonal.
> 
> I guess I understand what you mean, but could you tell me how it's not
> orthogonal just to clarify?

Sorry - I was feeling minimalist this morning :)

It seems to me that the intent here is expose some part of a string
as a String it its own right, allow operations on that string, and
then have those operations affect the original. One way of doing that
would be to have every single string mutator to accept start and end
positions (in fact to be consistent they should also accept a range,
a regexp, and a string). So we'd have

  str.upcase!(2,4)
  str.tr!('a', 'b', 5, 6)

and so on. This seems bad: we have a single concept, the exposed
substring, and put its implementation into many different methods.

So, perhaps a better way might be to make the concept of exposing a
substring for manipulation orthogonal to that manipulation. Something
like:

   str.expose!(1,2).upcase
   str.expose!(5,6).tr('a', 'b')

We could then make it more generic:

   ary.expose(2,-2)!.sort    # sort all bug first and last elements

In general, this could probably be done as a mixin on all classes that
implement slices.


Dave