From: "David A. Black" <dblack@...>
Date: 2005-07-12T02:58:47+09:00
Subject: Re: accessing index inside map

--927295978-88016008-1121104697=:17833
Content-Type: MULTIPART/MIXED; BOUNDARY="927295978-88016008-1121104697=:17833"

  This message is in MIME format.  The first part should be readable text,
  while the remaining parts are likely unreadable without MIME-aware tools.

--927295978-88016008-1121104697=:17833
Content-Type: TEXT/PLAIN; charset=X-UNKNOWN; format=flowed
Content-Transfer-Encoding: QUOTED-PRINTABLE

Hi --

On Mon, 11 Jul 2005, [iso-8859-1] "Pe=F1a, Botp" wrote:

> dblack@wobblini wrote:
>
> #> OK, i'll be shooting for the moon here. How about passing a parameter?=
 Is
> it
> #> then possible to make #each or #map same behavior w #each_with_index o=
r
> #> #map_with_index or whatever *_with_index so that the ff works,
> #>
> #> def each(with_index=3Dtrue) do... end
> #>
> #> def map(with_index=3Dtrue) do ... end
> #>
> #> and thus, we can do
> #>
> #> [1,2,3,4,5,6].map(true) {|x,i|[2,5].include?(i) ? x : x*2}
> #
> #I'd hope not to see that.  I'm actually hoping that maybe someday all
> #the boolean flag methods (like instance_methods(true) etc.) will be
> #changed :-)  They've always seemed to me to be the most cryptic thing
> #in all of Ruby.
> #
>
> Yes, you are correct. But there is some sort of give and take here or
> balance if i may say (i think i heard the word balance fr you, sir David)=
=2E.
>
> How about
>
> [1,2,3,4,5,6].map(WITH_INDEX) {|x,i|[2,5].include?(i) ? x : x*2}
> or
> [1,2,3,4,5,6].map(with_index=3D>true) {|x,i|[2,5].include?(i) ? x : x*2}
>
> if we don't pass flags or attribs, wouldn't that make creating additional
> methods like
>
> each
> each_with_index
> map
> map_with_index
> collect
> collect_with_index
> ...
>
> ?
>
> it's with_index es all over..

I don't know why, at least not the "..." part.  Matz has always
considered each case of a possible new method separately, on its own
merits.  (And collect_with_index would presumably be just an alias of
map_with_index :-)  I don't think there has to be some automatic
trigger where lots of other _with_index methods had to come into
existence, unless there were cases where there was a good reason for
them.  My main argument in favor of map_with_index is actually not
symmetry with each_with_index, but the fact that so many of us have
had to write that method.  I don't have an exact formula, but it's
struck me over the years (as with singleton_class) that there's a case
to be made on those grounds for adding it to Array.


David

--=20
David A. Black
dblack@wobblini.net
--927295978-88016008-1121104697=:17833--
--927295978-88016008-1121104697=:17833--