From: Dave Thomas <Dave@...>
Date: 2001-02-23T01:31:36+09:00
Subject: [ruby-talk:11327] Re: musings about Hash#each_with_index

jmichel@schur.institut.math.jussieu.fr (Jean Michel) writes:

> >Arrays and hashes have similar properties, but are different beasts,
> >In the same way that the old OO argument of Squares and Rectangles
> >can't be resolved by answering a Square is a kind of Rectangle, or a
> >Rectangle is a kind of Square, so too is it impossible to say that an
> >Array is a kind of Hash or a Hash a kind of Array. It seems to me that
> >they just share some properties.
> 
> Watch out  here! The fact  that you can consider  an array as  a special
> case of a hash, of for that matter  a string or a file or a directory as
> a special case of an array is much more than just an analogy. The set of
> properties  which is  shared in  these cases  is an  extremely important
> programming  paradigm, and  a language  which can  express clearly  this
> paradigm in a uniform way in all cases will have a definite advantage. I
> think the  mixins that matju is  trying to make can  potentially achieve
> this for Ruby.

I agree it's an important paradigm: interfaces are the best thing
about Java. I'm just not convinced it's a necessary paradigm. A couple 
of years ago, I'd have been on your side. I was a strong typing bigot, 
and wanted everything tied down. Now I'm learning that it's OK to rely 
on the fact that an interface is defined by the messages that an object 
happens to accept at runtime, and that (for example) using '<<' as a
"write to the end of" operator works just fine without having to
define a "write to the end of" interface.

I see matju's work differently. I don't see it constraining existing
behavior. I see it as

1. A quick way to generate new classes which ape the behavior of
   built-in classes, and more importantly

2. A step towards Ruby-in-Ruby/microRuby or whatever it's called this
   week.



Regards


Dave