proposed pre-CR edits to Turtle spec

I finally got a good chance to make a detailed review of the Last Call grammar and test out the semantics.

I tweaked the grammar slightly but the language it defines is the same:
  Struck NIL because it wasn't referenced
  Unfactored blank because it wasn't always blank (could be rdf:nil).
  Changed [Pp][Rr][Ee][Ff][Ii][Xx] to "PREFIX" with a note about case.
  Added missing xs before unicode code points in PN_CHARS_BASE and PN_CHARS.

In the semantics, I updated the term constructors to indicate which escaping was relevent where and updated the Triple Constructors after testing the algorithm against all of the Turtle tests that Andy submitted.

Green stuff in <> is stuff I propose to add. Red I propose to remove. I think it's worth getting this right before CR. Most of my feedback comes from my own XMass CR period. I believe all of this is clarifications or editorial adjustments so we could still go to CR immediately.

I also aligned the Turtle grammar intro more with the SPARQL spec 'cause it's pretty well weathered and had a lot of useful points. (We'll probably want to strike "6 In signed numbers, no white space is allowed between the sign and the number." if folks agree that it's not helpful to the Turtle spec, but stuff like identifying tokens and case sensitivity is a good idea.)

BTW, we don't need the At-risk that we discussed for the { <s> <p> <o> ;;; . } decision as it was present in the LC grammar.

Received on Monday, 31 December 2012 04:06:41 UTC