From: Trans <transfire@...>
Date: 2007-11-12T03:39:21+09:00
Subject: Re: equivalent injecting implementations?



On Nov 11, 9:50 am, "Rick DeNatale" <rick.denat...@gmail.com> wrote:
> Yes, I suspected that this was what Trans was trying to do, but since
> the second proposed 'maybe equivalent' implementation didn't seem to
> be doing the same thing, I was probing for what he really wanted.
>
> Having said that, and played a bit with the concept. I'm not sure how
> useful an injecting method (at least with that name is).
>
> As Robert Klemme points out, it breaks depending on what you are
> injecting. And it's subject to "cargo culting."
>
> module Enumerable
>   def injecting(obj, &blk)
>     inject(obj) { |mem, var| yield(mem, var); mem }
>   end
> end
>
> As Ezra points out:
>
> [1,2,3].inject({}) { |memo,n| memo[n] = n ;  memo} # => {1=>1, 2=>2, 3=>3}
> [1,2,3].injecting({}) { |memo,n| memo[n] = n}      # => {1=>1, 2=>2, 3=>3}
>
> But this only works because the block has a side-effect on the memo
> object.  And yes, in cases like this each instead of inject will have
> the same side effects.
>
> [1,2,3].inject(0) { |memo,n| memo + n}             # => 6
> [1,2,3].injecting(0) { |memo,n| memo + n}          # => 0
>
> Here's where the cargo cult danger comes in. Someone who comes across
> 'injecting' without understanding it might run into things like this.

Hmmm... does this mean that the same effect can be achieved with:

  module Enumerable
    def inject!(obj)
      each { |var| yield(obj, var) }
      obj
    end
  end

I think the common usecase is simply not having to create the initial
object outside the loop statement, eg.

  obj = 0
  each { |i| ... obj ... }

so that would make sense.

T.