News for hackers. Stuff that matters.
Re:Ruby Koans. Best way to learn Ruby. by ximeng on Sunday August 22, 2010 @09:44AM (#1624601)

  def test_changing_hashes
    hash = { :one => "uno", :two => "dos" }
    hash[:one] = "eins"

    expected = { :one => "eins", :two => "dos" }
    assert_equal expected, hash

    # Bonus Question: Why was "expected" broken out into a variable
    # rather than used as a literal?
  end
Can anyone help with the bonus question? From searching Google, I read that the reason:

    assert_equal { :one => "eins", :two => "dos" }, hash
gives an error is that Ruby believes this to be a block. But then why doesn't it consider the hash to be a block below:

    expected = { :one => "eins", :two => "dos" }
?

Edit: thanks jashkenas for the explanation! Edit2: for anybody else trying these, I think the version in github is more up-to-date, with some corrections:

http://github.com/edgecase/ruby_koans

Re:Ruby Koans. Best way to learn Ruby. by jashkenas on Sunday August 22, 2010 @10:02AM (#1624621)
This is one of those gotchas with Ruby's syntax for blocks clashing with Ruby's syntax for hash literals. I'm not familiar with the details of precisely why Ruby's parser handles some cases, and not others, but I bet there's a special case for assignment -- you can't ever assign a block literal, so it has to be a hash...

Here are some of the variants:

    method({ ... })  # hash

    method { ... }   # block, even though a hash would be possible.

    method = { ... } # hash

    method a, b, :c => d  # implicit hash

    method a, b { ... } # SyntaxError

    method(a, b) { ... } # block

Only #2 in that list is really a problem, because there are situations where you'd legitimately like to use both.
Re:Ruby Koans. Best way to learn Ruby. by andrewvc on Sunday August 22, 2010 @12:11PM (#1624795)
This is one reason why I always use parens on my method calls. Some rubyists find it cluttering, yeah, but I think it makes reading code easier.

The other one to watch out for is precedence of {} vs do; end. Example:

  > x = [1,2].map do |i| i*2; end
   => [2, 4] 
  > puts [1,2].map do |i| i*2; end
  1
  2
   => nil
  > puts [1,2].map { |i| i*2}
  2
  4
   => nil
Re:Ruby Koans. Best way to learn Ruby. by jashkenas on Sunday August 22, 2010 @12:23PM (#1624811)
Agreed. It's why we've tried to be careful with lambda/hash/call syntax in CoffeeScript, especially where some of it is optional, and can be left out. For example:

    x = [1, 2].map (i) -> i * 2
... is without optional parens.

    x = [1, 2].map((i) -> i * 2)
... is with them.

    configure start: 10, end: 25
... is the syntax for calling a method and passing an implicit hash -- and with the optional braces and parens added, it's closer to JavaScript...

    configure({start: 10, end: 25})
Re:Ruby Koans. Best way to learn Ruby. by jimweirich on Sunday August 22, 2010 @05:35PM (#1625324)
#5 is also not a syntax error. Rather b is interpreted as a method and the block is associated with it.