From: dblack@...
Date: 2007-08-17T07:28:58+09:00
Subject: Re: Lazy function definition pattern in Ruby?

--1926193751-1334731984-1187303337=:28099
Content-Type: MULTIPART/MIXED; BOUNDARY="1926193751-1334731984-1187303337=:28099"

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

--1926193751-1334731984-1187303337=:28099
Content-Type: TEXT/PLAIN; charset=X-UNKNOWN; format=flowed
Content-Transfer-Encoding: QUOTED-PRINTABLE

Hi --

On Fri, 17 Aug 2007, Wolfgang N=C3=A1dasi-Donner wrote:

> Sam Kong wrote:
>> Hi,
>>
>> Yesterday, I read a blog about lazy function definition pattern in
>> JavaScript at http://peter.michaux.ca/article/3556 .
>> It was interesting and insightful.
>>
>> <snip>
>> Write a function foo that returns a Date object that holds the time
>> that foo was first called.
>>
>> var foo =3D function() {
>>     var t =3D new Date();
>>     foo =3D function() {
>>         return t;
>>     };
>>     return foo();
>> };
>> </snip>
>>
>> In ruby, one would write the following way or something like that if
>> he wants to cache the first value.
>>
>> def foo
>>   @t or (@t =3D Time.new)
>> end
>>
>> But the writer wants to remove the conditional part because it's run
>> every time the function is called.
>> JavaScript allows functions to be redefined very easily.
>> I think ruby allows it but not very easily.
>>
>> I came up with this idea.
>>
>> class Lazy
>>   def method_missing *args
>>     if args[0] =3D=3D :foo
>>       @t =3D Time.new
>>       class << self
>>         def foo
>>           @t
>>         end
>>       end
>>       return foo
>>     end
>>   end
>> end
>>
>> But I believe that ruby gurus will have better ideas.
>> What would be the lazy function definition pattern in ruby?
>> And do you think it's useful?
>>
>> Thanks in advance.
>>
>> Sam
>
> You can define a method inside a method directly.
>
> class Bar
>  def foo
>    @t =3D Time.new
>    def foo
>      @t
>    end
>    @t
>  end
> end
>
> x=3DBar.new
> p x.foo # =3D> Thu Aug 16 22:17:17 +0200 2007
> sleep 5
> p x.foo # =3D> Thu Aug 16 22:17:17 +0200 2007

A possible problem with that code is that it only works for one
instance of Bar:

   Bar.new.foo   # Thu Aug 16 18:23:02 -0400 2007
   Bar.new.foo   # nil

My code is per-object:

   Bar.new.foo   # Thu Aug 16 18:25:02 -0400 2007
   Bar.new.foo   # Thu Aug 16 18:25:03 -0400 2007

and each object keeps its own. Another possibility is:

   class Bar
     def foo
       t =3D Time.now
       self.class.class_eval do
         define_method(:foo) { t }
       end
       t
     end
   end

That would preserve the time from the very first call to the method,
across all instances. It depends how one wants to fine-tune the
behavior, I guess.


David

--=20
* Books:
   RAILS ROUTING (new! http://www.awprofessional.com/title/0321509242)
   RUBY FOR RAILS (http://www.manning.com/black)
* Ruby/Rails training
     & consulting:  Ruby Power and Light, LLC (http://www.rubypal.com)
--1926193751-1334731984-1187303337=:28099--
--1926193751-1334731984-1187303337=:28099--