RE: [Bug 4639] Allow cycle checking on element graphs as well as document graphs

Ginny,

The "mismatch" of which you speak is, I thought, a natural consequence
of our decision to drop sml:ref="true" from the schema.  The sml:acyclic
and target* attributes carry only an implicit assumption that the
instance element will be a SML reference, i.e., have a sml:ref="true".
I believe Sandy's proposal on how to treat these sml particles already
specifies the system behavior when these occur on non-reference
elements.

And if the instance document author forgets to add the sml:ref="true",
there will be more problems than simply having some attributes in the
schema that don't "do anything".  This was a problem from the day it was
decided to use the sml:ref attribute in instance documents rather than
examining the PSVI.  

But, yes, the situation gives one an uneasy feeling. 

Kirk Wilson, Ph.D.
Research Staff Member
CA Labs
603 823-7146
 
-----Original Message-----
From: public-sml-request@w3.org [mailto:public-sml-request@w3.org] On
Behalf Of Smith, Virginia (HP Software)
Sent: Friday, October 26, 2007 12:06 PM
To: public-sml@w3.org
Subject: RE: [Bug 4639] Allow cycle checking on element graphs as well
as document graphs


I like Kirk's revised definition. 

No, we didn't decide to re-introduce the sml:ref="true". That was just
me trying to make sense of the new syntax. It could be left out, in
which case the presence of sml:acyclic would be ignored, I suppose. But
then the document author would have to remember to add the sml:ref =
true in the instance document when the sml:acyclic is actually defined
in the schema. This mismatch doesn't make sense to me.

--
ginny

-----Original Message-----
From: public-sml-request@w3.org [mailto:public-sml-request@w3.org] On
Behalf Of bugzilla@wiggum.w3.org
Sent: Friday, October 26, 2007 8:39 AM
To: public-sml@w3.org
Subject: [Bug 4639] Allow cycle checking on element graphs as well as
document graphs


http://www.w3.org/Bugs/Public/show_bug.cgi?id=4639





------- Comment #7 from kirk.wilson@ca.com  2007-10-26 14:38 ------- I
would like to suggest a small editorial change to the definition of
cycle. 
(If you omit the phrase in commas in the current definition, the
definition no longer makes sense, and the temporal "when" offends my
sense that this is structural definition).  Suggested revision is quite
simple:

A cycle is formed for an element E if the path that is formed by
recursively following an SML reference that is a descendant of E to its
target leads back to E.

Also, I don't recall that part of the discussion at the F2F reintroduced
the sml:ref="true" attribute at the schema level (as represented in
Proposals 2 & 3).  Was this reintroduced during the subsequent
conference call?

Received on Saturday, 27 October 2007 17:46:42 UTC