- From: <bugzilla@wiggum.w3.org>
- Date: Tue, 25 Apr 2006 00:01:59 +0000
- To: public-qt-comments@w3.org
- CC:
http://www.w3.org/Bugs/Public/show_bug.cgi?id=1804 ------- Comment #17 from simeon@us.ibm.com 2006-04-25 00:01 ------- Michael, Again, trying to find some common ground on this thread. Here are a few additional thoughts. (a) I want to confirm that indeed the Formal Semantics specification has been written with the intent that non-terminals do not participate in the 'syntactic object' they described. (b) We do not have any concrete evidence that the current interpretation is broken. (c) I believe I understand your concern that some inference rules may be triggered by being matched with a syntactic object that may not be relevant for that rule. That said, I am not quite as concerned as you may be. The formal semantics processes essentially 3 kinds of objects: values, types, and expressions. I believe judgments are never such that they may bind two of those kinds of objects for the same pattern so there should not be any case where an object of one kind can be matched instead of an item of another kind (e.g., expression 1,2,3 will never be confused with value 1,2,3). Then for one such kind of objects, let say values, the syntactic objects are independant of the grammar production and the grammar can be seen as a constraint on which subset of the objects can be matched. Obviously this reasoning does not represent a formal proof, just the basic intuition. (d) Whatever interpretation is choosen, addressing fully your concern will require some significant work (a formal proof and a detailed review in either case). Now let me try and propose a plan to move us forward, in two steps: First step: * For now and considering (a) and (b) above I would suggest to keep the intended interpretation since this is the intended semantics and the spec is based on that interpretation. * We should clarify the intended semantics by explaining the role/semantics of non-terminals in the corresponding notation section. We can then close bugs 1804 and 1803. Second step, assuming someone has the interest and resources to do this: * Review the current interpretation to see if anything is broken with the current interpretation, and if so, post a technical item about this on bugzilla. Let me know how that sounds, - Jerome
Received on Tuesday, 25 April 2006 00:02:03 UTC