From: Rich Kilmer <rich@...>
Date: 2002-02-14T04:59:02+09:00
Subject: RE: qualms about respond_to? idiom



> -----Original Message-----
> From: dave@thomases.com [mailto:dave@thomases.com]On Behalf Of Dave
> Thomas
> Sent: Wednesday, February 13, 2002 12:23 PM
> To: ruby-talk ML
> Subject: Re: qualms about respond_to? idiom
>
> <snip>
>
> If this were a big deal, I'd probably arrange to wrap the stream
> objects in objects that handled things like the stack manipulation,
> delegating the normal stream object calls.
>
> However, there's also a bit of a smell to the original:
>
>   if e.responds_to? :tagStart
>     do_somthing if e.tagStart
>   end
>
> What are we doing here? All tags have a method that says "yes I'm a
> start tag" or "no I'm not a start tag", and we're using that method to
> determine if something is a tag. We then _call_ that method to see it
> it's a start tag.
>
> So the next step would be to say "only start tags respond to
> :tagStart", and the code becomes
>
>   do_somthing if e.responds_to? :tagStart
>
> But then that smells: we have a method that we never call, simply so
> we can use it as a flag. Hmm, let's have a mixin called StartTag that
> anything that is a start tag includes
>
>   module StartTag; end
>
> Then we have
>
>   do_somthing if e.kind_of? StartTag
>
>
> Does that communicate the intent?
>
>

Regarding Mixins vs. individual methods (respond_to?), I think using a Mixin
gives you a larger namespace to manage identification of behaviors (whereas
methods are in a single "flat" namespace).

<silly_example>
module USAF
 module DEFCON_1
  def do_your_worst
   @missles.launch
  end
 end
end

can be semantically differentiated with kind_of? from:

module StateFarmInsurance
 def do_your_worst
   @policy.cancel
 end
end
</silly_example>

whereas (respond_to? :do_your_worst) cannot.

-rich