Re: CDR: who is the audience?

On Mar 17, 2006, at 4:08 AM, Kevin E Kelly wrote:

>
> Who is the intended audience for this specification?
>
> Is it meant to be an actual specification that user agents should
> implement and that content authors should write to?
>
> [CDF] Yes.

I think it fails on this front. Particularly with the recent  
revisions, it does not provide useful conformance criteria for user  
agents.

> Or is it meant to  be just a meta-spec that defines requirements  
> that other specs should
> follow?



> [CDF] The Profiles build on the Framework and profile Core document.
> http://www.w3.org/TR/2005/WD-CDR-20051219/#cd-framework  (last  
> paragraph)
> http://www.w3.org/TR/2005/WD-CDR-20051219/#related-documents
> http://www.w3.org/TR/WICD/#scope (first paragraph)
> http://www.w3.org/TR/WICD/#related (document descriptions and  
> relationships for WICD)
>
> [CDF]  It seems as if we have spread the total answer to your  
> question across the Framework and WICD Core documents.
> [CDF]  Authoring Guidelines are in an Appendix to the documents,
> http://www.w3.org/TR/2005/WD-CDR-20051219/#authoring-guidelines
> http://www.w3.org/TR/2005/WD-WICD-20051219/#authoring-guidelines
> http://www.w3.org/TR/2005/WD-WICDMobile-20051219/#authoring-guidelines
> http://www.w3.org/TR/2005/WD-WICDFull-20051219/#authoring-guidelines
> [CDF]  No changes are being made in response to this comment.

I'm not sure if this is a "yes" or "no". I'd like the CDR document to  
clearly state the requirements for a conformant CDR profile. Are you  
saying it actually does define such a conformance class?

>
> I think it would be a sensible goal to define a baseline for cross-
> document inclusion and behavior that works regardless of "profile" or
> specific combination of languages being used. But currently, the spec
> as written appears to entirely lack testable, meaningful conformance
> criteria for either documents or user agents. I can't imagine how
> anyone could write a CDR validator for content or a CDR test suite
> for user agents. I do not see how one could check for the existence
> of interoperable implementations.
>
> [CDF]  The first set of documents are for compound documents by  
> reference only, so there is no compound document by inclusion  
> content yet other than some definitions and non-normative examples  
> in the Framework document.  Work has begun on the CDI documents.

Sorry for my imprecise language. When I said "inclusion" I did not  
mean CDI, but rather

> [CDF]  No changes are being made in response to this comment.
>
> [CDF]  There are testable assertions in the CDRF and WICD Core and  
> Profiles, and the CDF WG is building a test suite to test these  
> assertions.

I believe there are no testable assertions for user agents in the  
CDRF Last Call draft. I read it very closely and came to this  
conclusion. Can you provide any counter-examples? I think the latest  
draft may have the testable assertion that the UA must implement  
Window, but I'm not sure that really counts.

> [CDF]  The Framework document does not have many testable  
> assertions, but that is becasue it is a framework document.  One  
> such testable assertions is:
>
>           [assert-dom1: Compound Document profiles which leverage  
> the Compound Document
>             Framework and which support scripting must have scripting
>             interfaces that are compatible with the DOM Level 3  
> Core Specification.]

This is a testable assertion for profiles, not for user agents or  
content.

> [CDF]  The WIDC Core and Profile documents have many more testable  
> assertions:

Great, but my comment was on CDR lacking testable assertions for user  
agents and content. So none of these counter-examples are relevant.

[... snip ...]

>
> [CDF]  No changes are being made in response to this comment.

I disagree with this resolution, since the WG's justification appears  
to be based on an incorrect statement of fact (the premise that CDRF  
contains testable assertions for user agents).

> The spec should either be reworked to have useful conformance
> criteria, or it should clearly state that its target audience is
> other specs, and remove the conformance criteria for user agents and
> documents.
>
> [CDF]  The abstract texts for the Framework, WIDC Core, and  
> Profiles state the intent of the documents with additional  
> explanatory text in the
> related documents and scope sections.  The intended audience is  
> both user agent developers and content authors.  A separate  
> Appendix on
> Authoring Guidelines is in each document.  No changes are being  
> made in response to this comment.
>
> [CDF]  The Conformance Appendixes in the document will remain.  No  
> changes are being made in response to this comment.

I disagree with this resolution. It makes no sense to have  
Conformance appendices that do not state testable conformance  
requirements.

Regards,
Maciej