This is an archived snapshot of W3C's public bugzilla bug tracker, decommissioned in April 2019. Please see the home page for more details.

Bug 5679 - fix disparity between SML & SML-IF specs regarding SML ref targets
Summary: fix disparity between SML & SML-IF specs regarding SML ref targets
Status: RESOLVED FIXED
Alias: None
Product: SML
Classification: Unclassified
Component: Interchange Format (show other bugs)
Version: LC
Hardware: PC Windows NT
: P2 normal
Target Milestone: ---
Assignee: Virginia Smith
QA Contact: SML Working Group discussion list
URL:
Whiteboard:
Keywords: resolved
Depends on:
Blocks:
 
Reported: 2008-05-07 21:42 UTC by Kumar Pandit
Modified: 2008-06-19 18:58 UTC (History)
0 users

See Also:


Attachments

Description Kumar Pandit 2008-05-07 21:42:52 UTC
The LC draft of SML-IF requires that an SML reference pointing to a schema document must be treated as unresolved. This is defined in section 5.3.4 URI Reference Processing (http://www.w3.org/TR/2008/WD-sml-if-20080303/#URI_Processing). That section does not define what happens when an SML reference targets a rule document. 
  
Proposal:
Allow a ref to target a schema doc as well as a rule doc. This does not require adding new verbiage. Just remove the existing text that defines this restriction.

Reasons:
1. The member submission draft does not define any such restriction. As far as I know, we did not specifically take a separate decision on this. It could also have resulted due to an editor error.
2. The SML spec itself does not place any restriction on where an SML ref could point to. It is valid for a model to have an SML ref pointing to a schema/rule doc. However, given the current SML-IF spec, there will be no way to serialize such a model into SML-IF. This would mean that not all SML models can be represented using SML-IF. This is clearly not what we intended.
Comment 1 John Arwe 2008-05-08 13:21:41 UTC
I believe this crept in during early formulations of the adjustments we made to handle schemalocation in general, but not in instance documents (the idea being that schemalocation in an instance doc might be trustworthy enough for things like schema-based end-user input guidance, but not enough for "things we care about" like model validation).

While the general notion of pointers from instances to classes hurts my head, I'd rather have the flexibility (loaded gun, pointed at, foot) and not use it than need it and not have it.

i.e. +1
Comment 2 Pratul Dublish 2008-05-22 19:05:12 UTC
Resolution in 5/22 call - fix as per the proposal in WG
Comment 3 Virginia Smith 2008-06-05 20:53:23 UTC
Changed section 5.3.4 as follows:

1 - bullets 1 and 2 are unchanged.
2 - bullet 3 is now:

---------
3. If D has a value, then

   a. If UR is within category #1 (schemaLocation), then UR has a target if and only if all of the following are true.
         i. D is a schema document that is also a model definition document in the interchange model.
         ii. UR does not contain a fragment component. 
   2b. If UR is within category #2, then
         i. If UR does not contain a fragment component, then it targets the root element of D.
         ii. Otherwise (UR contains a fragment component), the fragment component of UR is applied to the root element of D, which may result in 0, 1, or many target elements.
-------------

3 - bullets 4 and 5 have been removed.

See the diff at:
http://www.w3.org/2007/10/htmldiff?doc1=http%3A%2F%2Fdev.w3.org%2Fcvsweb%2F%7Echeckout%7E%2F2007%2Fxml%2Fsml%2Fbuild%2Fsml-if.html%3Frev%3D1.156%26content-type%3Dtext%2Fhtml%3B%2520charset%3Diso-8859-1&doc2=http%3A%2F%2Fdev.w3.org%2Fcvsweb%2F%7Echeckout%7E%2F2007%2Fxml%2Fsml%2Fbuild%2Fsml-if.html%3Frev%3D1.157%26content-type%3Dtext%2Fhtml%3B%2520charset%3Diso-8859-1
(however, for some reason, the diff does not show the removed bullets 4 and 5)