From: Chris Pine <nemo@...>
Date: 2003-04-17T13:38:37+09:00
Subject: Re: Newbie proc question

----- Original Message -----
From: "James Toomey" <jamesvtoomey@yahoo.com>


Can anyone help reduce my confusion?

<snip>
----------------------------


Well, as I'm sure you have guessed, your examples weren't really great uses
of blocks/procs.

So what's the point of a proc?  Well, it's a little function you can pass
around.  There are many times when you might want to do this.  Basically,
it's another level of abstraction.  Let's look at some low-level
abstractions and work our way up.

Let's say you were writing a program where you wanted to say 'hello' three
times in a row.  You might do something like:

  puts 'hello'
  puts 'hello'
  puts 'hello'

(We could have done a loop or something, but this is good enough for now.)
But what if I wanted to do the same thing from another spot in the code?  I
could just copy and paste, but copy and paste is BAD.  (*Cut* and paste is
fine... just no copying.)  No, the thing to do is to group this code
together in a method:

  def threeHellos
    puts 'hello'
    puts 'hello'
    puts 'hello'
  end

Now I can call this from any number of places in my code, no problem.  I
don't want duplicate code running around.

Now what if I wanted to say 'goodbye' three times in other places in my
code?  I could define a new method `threeGoodbyes', but it would share a
great deal in common with `threeHellos'.  I should refactor:

  def sayItThreeTimes phrase
    puts phrase
    puts phrase
    puts phrase
  end

  def threeHellos
    sayItThreeTimes 'hello'
  end

  def threeGoodbyes
    sayItThreeTimes 'goodbye'
  end


Excellent.  But what if, later on, I want to do something else three times?
Like get a string from user input and echo it?

  def threeEchos
    puts gets
    puts gets
    puts gets
  end

It certainly has something in common with `sayItThreeTimes', but how do we
factor it out?  We use procs.  After some more refactoring:

  def doItThreeTimes aProc
    aProc.call
    aProc.call
    aProc.call
  end

  def threeEchos
    echoProc = proc {puts gets}
    doItThreeTimes echoProc
  end

  def sayItThreeTimes phrase
    sayItProc = proc {puts phrase}
    doItThreeTimes sayItProc
  end

See?  Now we are passing in the proc as a normal parameter.  Ruby has a
special way to pass in one proc per method call, though, using `&':

  def doItThreeTimes (&aProc)
    aProc.call
    aProc.call
    aProc.call
  end

  def threeEchos
    echoProc = proc {puts gets}
    doItThreeTimes (&echoProc)
  end

  def sayItThreeTimes phrase
    sayItProc = proc {puts phrase}
    doItThreeTimes (&sayItProc)
  end

Ruby also has a way to pass that proc in using a block, like this:

  def threeEchos
    doItThreeTimes {puts gets}
  end

  def sayItThreeTimes phrase
    doItThreeTimes {puts phrase}
  end

It means the same thing; it's just that we don't have to give it a name.
Similarly, if we want, we don't have to give it a name in `doItThreeTimes',
either:

  def doItThreeTimes
    yield
    yield
    yield
  end

This also means the same thing:  just call the special proc we passed in.

Of course, this was a silly example; we don't need `doItThreeTimes' when we
have this:

  3.times { ... }

`n.times' takes a block and calls it `n' times.

So that's one common use of blocks:  iterators.  `each' is another popular
method which takes a block.  I assume you are familiar with it?

Blocks are also useful when you have something you want done before and
after some code is called.  Let's say you were writing a profiler.  To find
out how long a method call takes, you must find out the time before you call
the method, then the time after it is done, and subtract them:

  startTime = Time.now
  #  Put your code to profile here.
  callTime = Time.now - startTime

Maybe you do this often, and you find yourself copying and pasting...
nonono!  You can use blocks:

  def profile
    startTime = Time.now
    yield  #  Call the code to be profiled.
    Time.now - startTime
  end

  profile {12345**12345}

You can use this for logging, database transactions... This is often used
for file IO, since you want to open a file, do some stuff to it, then close
it at the end.  If you use a block, you don't have to bother with closing
it -- Ruby will do it for you.

So does that make a little more sense?

Chris