Rubinius parser confusion

gistfile1.rb

# Here's the code being evaluated
# rubinius-2.0.0dev

html_class = build_class("") {|c| c << "required" if required }
#  => "required" 

### FIRST EXAMPLE
# Symbolic expression tree with no arguments to build_class

'html_class = build_class() {|c| c << "required" if required }'.to_sexp
#  => [:lasgn, :html_class, 
#       [:iter, 
#         [:call, nil, :build_class, [:arglist]], 
#         [:lasgn, :c], [:if, [:call, nil, :required, nil], [:call, [:lvar, :c], :<<, [:arglist, [:str, "required"]]], nil]]]

# The block becomes the second argument of `iter` like expected

### SECOND EXAMPLE
# Symbolic expression tree with 1 argument to build_class

'html_class = build_class("") {|c| c << "required" if required }'.to_sexp
#  => [:lasgn, :html_class, 
#       [:iter, 
#         [:call, nil, :build_class, [:arglist, [:str, ""], [:iter, [:lasgn, :c], [:if, [:call, nil, :required, nil], [:call, [:lvar, :c], :<<, [:arglist, [:str, "required"]]], nil]]]], 
#         [:lasgn, :c], [:if, [:call, nil, :required, nil], [:call, [:lvar, :c], :<<, [:arglist, [:str, "required"]]], nil]]] 

# Why is the block repeated in the second example? It becomes a parameter to `arglist` and `iter` with the presence of a method argument

No idea.

From #rubinius:

14:49 < brixen> he should be looking at the AST output
14:49 < brixen> if there is confusion
14:50 < brixen> the sexp output is a hack on top of the AST
14:50 < brixen> that aside, I don't understand his question

Thanks Steve. I'll hop on

I found the issue in the #to_sexp output. I'll look into using the AST nodes instead