RE: [w3c sml] Bug 4639 (Action 136) updated cycle proposal document

During the Redmond f2f meeting, we had discussed a proposal where
-- we retain sml:acyclic on SML ref
-- we consider hierarchy of a ref when detecting cycles

For example, consider docs d1 & d2. d1 has element e1 and d2 has element e2. Suppose sml:acyclic=true is defined on complexType A. In this case, a cycle is formed between e1 & e2 if:
-- e1 is or contains SML ref of type A (or a type derived from it) to e2 AND
-- e2 is or contains SML ref of type A (or a type derived from it) to e1

Is this proposal covered in the latest doc? If not, was this subsequently discussed in some email that I missed?


-----Original Message-----
From: public-sml-request@w3.org [mailto:public-sml-request@w3.org] On Behalf Of Smith, Virginia (HP Software)
Sent: Thursday, October 25, 2007 3:06 PM
To: public-sml@w3.org
Subject: [w3c sml] Bug 4639 (Action 136) updated cycle proposal document

Attached is the updated cycle proposal. Let me start by saying that this problem is hard and my impression is that we could really screw this up if we don't give this some serious thought.

After spending a lot of time thinking through these new proposals, it seems crazy to me (maybe it's just my headache) that a schema author will intentionally create cycles (acyclic or not!) that consist of completely different types of nodes connected by completely different types of links. Anyone know of a good use case for that?

Perhaps we are trying to allow for every possible cycle a person can dream up and I'm not sure that is productive.

The only thing that is changed is section 3. The 2 new proposals from the f2f have been added with some thoughts.

--
ginny

Received on Tuesday, 30 October 2007 04:17:46 UTC