Re: [office] Motion to certify that all schema and XML instancesincluded in the ODF 1.2 specification are valid

Hi Michael,

Are you sure that a vote is needed for this?  I don't see a requirement 
for a vote anywhere.   In any case, this would be an odd thing for someone 
to vote for, since it is not really a question of preference or approval. 

"Certification" is an odd word to use here.  The TC Process uses it in 
only one other place, to speak of TC Admins certifying a Work Product with 
"Designated Cross-Reference Changes".  If you go back to the August 2006 
version of the Process, they had a requirement for "Certification by at 
least three OASIS member organizations that they are successfully using 
the specification".  The language for this requirement was changed in 2007 
to "Statements of Use" rather than "certification".    In that case 
certification does not imply a vote.

Another problem is in  "all expressions are valid".  This makes zero 
sense.  "Expression" is not a term of art in the XML specification.  Relax 
NG defines "expression" but that is an abstraction of the grammar,  In 
Relax NG an "element" is what is valid with respect to a schema. 

Further, I think it is not reasonable to expect that "fragments" of XML 
will be valid. A fragment for example, may be shown in an figure in a 
specification, as an illustration, perhaps collapsing irrelevant 
attributes, omitting namespace declarations, or without showing a 
necessary parent element or elements, or targets of an IDREF.   This is 
common practice.  But it should be encouraged to require that these be 
"feasibly valid", which is  a broader concept that says that with 
additions of attributes and elements (but no other changes) that the 
fragment could be made valid.

There is not much we can do about this requirement at this point.  But 
something for OASIS to consider is restating this requirement as something 
like this:

Definitions:

Formal notation: Specification of normative requirements using a notation 
other than prose or mathematical notation, e.g., schema definitions, XML 
instances, EBNF, etc.

Validation Report: A report listing all normative formal notations in a 
Committee Specification, whether by inclusion or reference, along with a 
list of corresponding and governing formal notations, as well as a output 
from a validator demonstrating that each instance of formal notation 
conforms with the requirements of the governing notation with which it 
claims adherence.

And then state the requirement as:

"(c) Approval by Simple Majority Vote by the TC of a Validation Report for 
the Committee Specification"

That would make more sense, since is easier to see what approving a report 
entails than trying to imagine what it means for each of us as individuals 
to "certify" to dozens of validation micro-facts than few of us will 
personally verify.  It also eliminates any requirements on non-normative 
example fragments.  It also acknowledges that some standards have a formal 
notation that is not XML, for example in OpenFormula. 

In any case, no harm in having a vote on this.

-Rob




MICHAEL BRAUER <> wrote on 04/01/2011 07:37:50 
AM: