On Sunday 18 February 2001 20:43, David Alan Black wrote:
> On Mon, 19 Feb 2001, craig duncan wrote:
> > Ben Tilly wrote:
> > > craig duncan <duncan / nycap.rr.com> wrote:
> > > >Yukihiro Matsumoto wrote:
> > > > > 'to_s' is a mean to convert something into a string (if you really
> > > > > want to), OTOH 'to_str' is a mean for something that wants to
> > > > > pretend to be a string. to_a and to_ary are in similar
> > > > > relationship.
> > > >
> > > >The problem with this explanation is that i have no idea what it means
> > > > for something to "pretend to be a string" as opposed to being one.
> > >
> > > Think of it this way.  foo.to_s tells foo to describe itself
> > > as a string.  foo.to_str checks with foo that it is OK to
> > > think of foo as being its description, and gets that
> > > description.
> > >
> > > So to_str is the same as to_s except with an extra assertion
> > > about the correct usage of that object.
> >
> > Wow!  That was a little too subtle for me to guess.  I think because of
> > the minimalism that i (initially) perceive to be lacking in this design. 
> > to_str can always fill in for to_s, yes?
>
> No, because to_s is inherited from Object by every object, whereas
> to_str is only defined for a couple of built-in types (String,
> Exception).
>
> > So what do you need to_s for?  I guess you could say that you use
> > to_s when you _don't_ want the object to be used as a string
> > . . . even though you want it to _appear_ that way.  ?  Oh, well, i
> > guess i can accept that there _might_ be some usage for that
> > distinction (can't think what, though).  Thanks, Ben.
>
> Here's my current understanding of it, for what it's worth:
>
> Every object will respond to to_s, and to_s will always give you
> *something*.  But sometimes you might want a given class to have a
> non-canonical way of showing itself as a string.  In such a case, you
> can define a to_str method, such that your objects have their own idea
> of how to represent themselves as strings.
>
> A little test/demo:
>
>    class Pretender
>
>      attr_reader :to_str
>
>      def initialize(s)
>        @to_str = "I am a Pretender object initialized with #{s}."
>      end
>
>    end
>
>    p = Pretender.new(Math::PI)
>
>    puts "Canonical/inherited conversion to string:"
>    puts p.to_s
>    puts
>    puts "Pretender object's own choice as to how to appear as a string:"
>    puts p.to_str
>
>
> Once an object has its to_str method, it will be able to respond with
> something when it is asked to do so by (for example) String#<<.
>
>

Or, consider that varying methods either assume an object is a string
and rasie an excpetion if it is not (e.g. puts) or make no such assumption 
but does the conversion internally (e.g print) so that for:

list = ['test', 2, false, 2.3]

the following:

list.each { |i| print i, "\n" }
list.each { |i| puts i.to_s }

produce the same result, but:

list.each { |i| puts i }

raises an exception.

Now, class Array (v 1.62) does not have a built-in to_str method, but
it can, as in:

class Array
  def to_str(x='')
    s = ''
    self.each { |i| s += i.to_s + x }
    s =~ /^(.*)#{s}$/
    $1
  end
end

which produces the following:

list.to_str #-> 'test2false2.3'
list.to_str "\s" #->'test 2 false 2.3'
list.to_str "," #->'test,2,false,2.3'
list.to_str '::' #->'test::2::false::2.3'

thus showing one of the main benefts of Ruby, which is if core functionality 
is 'lacking' for a specific user need, the user simply rolls his/her own! :-)

BTW list.to_str "\n" per the above -> nil. Does this violate POLS? No. As 
written, the method thus called should return nil. A non-string argument 
raises an exception.  It should...it is a very poorly written method! :-)

Regards,

Kent Starr
elderburn / mindspring.com
 

raises an excpetion.