From: Ben Tilly <ben_tilly@...>
Date: 2000-12-16T03:55:32+09:00
Subject: [ruby-talk:7340] Re: Why isn't Perl highly orthogonal?

David Alan Black <dblack@candle.superlink.net> wrote:
>
>Hello --
>
>On Fri, 15 Dec 2000, Ben Tilly wrote:
>
> > David Alan Black <dblack@candle.superlink.net> wrote:
> > >I'm not sure what it is you're saying wouldn't work.  For example:
> > >
> > >a = [1,2,3,4,5]
> > >b = a.map do |e| [e,-e] end
> > >c = b.map_with_indices do |e,i| "element #{i} is #{e.inspect}" end
> > >
> > >p c[2]  # => "element 2 is [3, -3]"
> >
> > I was thinking of the following:
> >
> >   a = [1, [], [3,4,[]]]
> >
> > Pass that to flatten and you will get [1,3,4]. which is
> > likely not what someone with that data structure intended
> > to do!
>
>Right.  But who's passing anything to flatten?  Hmmm....  Maybe you're
>still thinking of my original example?  But even there, the flattening
>affects only the result array, not the input array:

I was talking about your hack to have a map return multiple
values, map to anon arrays then flatten.  That hack will
not work on all data structures because flatten is
recursive.  Your original map_with_indices didn't have
that problem.

>[1, [], [3,4,[]]] .map do |e| [e.inspect, "hello"]  end .flatten
>
>  => ["1", "hello", "[]", "hello", "[3, 4, []]", "hello"]
>
>The flattening was just to show that you can get more elements out
>than you put in, which I think was what you originally said you'd had
>trouble with.  If you don't flatten (which is fine with me -- really!
>I don't mind! :-) then you get an array of arrays -- which may be what
>one wants in a given case, but would not have made the point.

I had trouble doing a straight map which returned more.  When
you flatten or push, that is behaviour I understand.  Some
context.  In current production versions of Perl:

  %some_hash = map {($_, 1)} @some_array;

is slow even though push is fast.  I was trying to test that
in Ruby.  But since that idiom is not supported directly, there
was nothing to test.

>And (just to clarify) my map_with_indices method doesn't flatten
>anything.  It just, umm, maps with indices:
>
I knew that.  I was talking about your map then flatten trick
to map an array of one size to one of different size.

[...]
> > I don't have enough of a sense yet to even try to say what
> > Ruby-esque behaviour should be.  Sorry.
>
>Don't apologize :-)  I just meant that it seemed to me that a,b could
>behave in iterator blocks as it does in method definitions -- i.e.,
>that the a,b syntax per se did not have to introduce ambiguities or
>imply unexpected list behavior.  My characterization of it in terms of
>Ruby-esque-ness was really just referring back to your earlier
>statement:
>
>   I don't think that Ruby is list-oriented or that list-oriented ways
>   of thinking are a good fit for it.  Having lists with variable numbers
>   of arguments is not part of the basic design and I don't think should
>   be hacked on top when it doesn't fit conceptually.
>
>Now, come on Ben, admit it -- there's just a *little* bit of
>willingness to say what is and is not Ruby-esque creeping in around
>the edges, isn't there? :-)
>
Willingness does not equal ability.  I was putting out a guess
based on an impression.  The fact that you (with more
experience) didn't think my impression was obvious suggests that
I am missing something.

Cheers,
Ben
_________________________________________________________________
Get your FREE download of MSN Explorer at http://explorer.msn.com