Re: Pretty Printing more widely

On Friday, August 27, 2004, 1:04:21 AM, Hugh wrote:

>> I agree in spirit, but let me offer an alternative.  #pp is
>> conceptually similar to #p.  #p doesn't exist in IO-like classes
>> because it's simply a shortcut for 'puts X.inspect'.  To keep the

> I don't agree with that reason.  puts is just a shortcut for adding
> "\n" into print statements, but we still have it.

I'm not arguing for reducing everything as far as possible.  The point
is:

  x.inspect   returns a string and 'p x'  prints that string.
  x.something returns a string and 'pp x' prints that string.

If there were some value for "something", then 'p' and 'pp' would have
a certain consistency which I think they deserve.

For the record, I would't mind #p and #pp to be defined in IO-like
objects, but I'm not clamouring for it.

>> analogy between #p and #pp, why can't #pp be a shortcut for "puts
>> X.something"?

> It is -- it is x.prettyprint.  The problem is that it goes to $>,
> which you may not want to meddle with.

The following gives me a NoMethodError.

  require 'pp'
  [1,2,3].prettyprint

And you're right, I have no interest in meddling with $> :)

>> In 'extensions', I define Object#pp_s.  It's not a great name, but

> define it as what?   When IO inherits it the meaning will surely
> neet to be different from when Array inherits it?

  class Object
    def pp_s
      pps = StringIO.new
      PP.pp(self, pps)
      pps.string
    end
  end

There's no need for specialisation, is there?

  [1,2,3].pp_s              # -> "[1, 2, 3]\n"

Cheers,
Gavin