Newlines before comments

In the following code:

class C
  # foo
end

I would expect the newline token after the C constant to be emitted before the comment token, but it is the opposite:

irb(main):005> Parser::CurrentRuby.new.tokenize(Parser::Source::Buffer.new("", source: "class C\n# foo\nend\n"))
=> 
[s(:class,
  s(:const, nil, :C), nil, nil),
 [#<Parser::Source::Comment :2:1 "# foo">],
 [[:kCLASS, ["class", #<Parser::Source::Range  0...5>]],
  [:tCONSTANT, ["C", #<Parser::Source::Range  6...7>]],
  [:tCOMMENT, ["# foo", #<Parser::Source::Range  8...13>]],
  [:tNL, [nil, #<Parser::Source::Range  7...8>]],
  [:kEND, ["end", #<Parser::Source::Range  14...17>]],
  [:tNL, [nil, #<Parser::Source::Range  17...18>]]]]

I can work around this, but I just want to clarify what the rule is here?

Start with the Parser::CurrentRuby.tokenize example using the shown class and comment source, and inspect how the comment and newline tokens are emitted. Compare the observed order with the parser's intended rule for comments and newlines; the issue is done when that rule is clarified and the expected behavior is established.