ODF Errata discussion (earlier - Re: REMINDER AND DRAFT AGENDA - Teleconference meeting of WG 6, 2010-06-23 at 13:00 UTC)

Dennis E. Hamilton dennis.hamilton at acm.org
Thu Jun 24 22:14:33 CEST 2010


Thinking about this some more:

If we can find a rule/definition in [RNG] that accomplishes what is
intended, we should use that, since the appeal is really to the ODF RNG
schema, which operates atop the data model for the XML document.  That is
where we should be consistent.  We should not use terms like "element
content" and "content model" which are confused with technical terms of [XML
1.0] that are not intended.  

Since we are talking about a condition that is determined by the RELAX NG
Schema, and we expect the application to honor that schema however it
processes the XML instance, we should use the same nomenclature that RELAX
NG does.

As far as I can tell, The RELAX NG terminology is to speak of ELEMENT
CHILDREN, not element content.  An element child is either a non-empty
string or an element.

Then, what we could say is (for ODF 1.2):

  1. "An OpenDocument Consumer shall ignore all element children ([RNG]
section 5, Data Model) that are strings consisting entirely of whitespace
characters and that do not satisfy a text pattern in the ODF schema
definition for the element." 

In the context of ODF 1.0/1.1/IS 26300, how about

  2. "An ODF processor shall ignore all element children ([RNG] section 5,
Data Model) that are strings consisting entirely of whitespace characters
and that do not satisfy a text pattern of the ODF schema definition for the
element."

This still provides no guidance for elements not defined in the ODF Schema
(or have fake definitions that allow arbitrary element children), but I
think it is tied to the ODF Schema in a satisfactory way.

I'll include this proposal in my review of the current ODF 1.0 Errata draft
now under ballot at the ODF TC. 

[NOTE: The RNG Data Model is Section 5 of the referenced ISO/IEC
19757-2:2003 that is [RNG]; it is Section 2 of the OASIS Committee
Specification for RNG.  I expanded the reference so that those using a
different version of the RNG specification will likely find the correct
material either way.]

 - Dennis




More information about the sc34wg6 mailing list