From: Phrogz <phrogz@...>
Date: 2007-11-26T03:40:03+09:00
Subject: Supplying multiple procs to initialization

One of my libraries previously required code like this:

  # Convert markup to HTML, doing something
  # useless with links and commands
  owl = OWLScribble.new( markup )
  owl.wiki_links.each{ |link|
    # do something useful to the link
  }
  owl.wiki_commands.each{ |command|
    # do something useful with the command
  }
  html = owl.to_html

I have changed it so that the user can instead describe how to handle
links and commands in general, so that during initialization the
library doesn't waste time or code lines doing something useless:

  owl = OWLScribble.new( markup ) do |owl|
    owl.handle_wiki_link do |tag, page, link_text|
      # do something useful to the link
    end
    owl.handle_wiki_command do |tag, command, params|
      # do something useful to the command
    end
  end
  puts owl.to_html

The syntax you see above is my current favorite choice for specifying
these two procs. Alternatives that I have rejected as less elegant:

  # Deferred initialization
  owl = OWLScribble.new( markup )
  owl.handle_wiki_link{ ... }
  owl.handle_wiki_command{ ... }
  owl.run
  puts owl.to_html

  # proc params
  do_link = proc{ ... }
  do_command = proc{ ... }
  owl = OWLScribble.new(
    markup,
    :handle_wiki_link=>do_link,
    :handle_wiki_command=>do_command
  )
  puts owl.to_html

Can anyone suggest a more elegant way of supplying multiple processing
procs that should be used during initialization, on a per-instance
basis?