Hacker News Reader

Comment by Tichy | Hacker News Reader

I only looked at the first few slides. Don't understand the problem with the block variables? The current version seems clearer to me than the new proposal. Variables are visible in the block they have been defined in, what could be easier to understand? Probably I am missing something, though...

View on HN

The major problem with block variables in ruby is that if you do something like

    x = 4
    1.upto(5) do |x|
      puts x
    end
    puts "after the block x is #{x}"

It will print:

    1
    2
    3
    4
    5
   after the block x is 5

That is probably not what you would expect!

true, but for this it would be fine(?):

    x = 4
    1.upto(5) do |y|
      x = y
      puts x
    end
    puts "after the block x is #{x}

So the problem is only the parameter to the block?

Yes I think so - your example will still print "after the block x is 5", but in that case that probably is what you would expect.

Maybe my head has just gone blank, but I cannot think of a way to get a local called 'x' inside that block that is in scope only within the block.

In Perl you would have done

    my $x = 5;
    foreach my $i in (1..5) {
      my $x = $i;
      ...
    }

And it would have given you a new 'block local' and left your original x alone, but I cannot think of how you can do the equivalent of 'my' in Ruby ...

The problem is that Ruby's scoping rules aren't consistent. Sometimes it appears lexically scoped, but sometimes it acts dynamically scoped.

Sometimes this leads to funny looking code (if you know to avoid it) or truly inscrutable bugs (if you don't).

That is what I'd expect.

1.upto implies starting at 1... otherwise why would upto be called on 1?

If you want the behavior you expect, type:

  x = 4
  5.times do 
    x += 1
  end

Normally you'd expect the parameter of a block to be a new variable, not to grab a variable from another scope and use that instead. So the "expected" value is 4, since x = 4 on the first line.

You'd expect these two examples to output the same number:

  x = 1
  my_fun do |x|
    x = 5
  end
  puts x # => 5

and

  x = 1
  my_fun do |y|
    y = 5
  end
  puts x # => 1

where my_fun could be defined as:

  def my_fun 
    yield 2
  end

You're right, and that is the behavior in Ruby 1.9.

I posted the example I did so that those unfamiliar with Ruby would see that the parent was a contrived example to use upto in that context. Sure it's awkward b/c it's not the intended semantics of the upto method, hence the ambiguity.

Maybe an ideal language would not introduce that sort of ambiguity... I personally slightly perfer the 1.9 syntax.

Reply on news.ycombinator.com