Re: Conformance to EARL

On Wed, 25 May 2005 13:35:06 +0200, Shadi Abou-Zahra <shadi@w3.org> wrote:

> Yes, we should certainly provide possibility to conform to EARL as well  
> as to define what conformance is. However, I think it is difficult to  
> require any tools to adopt other specifications on which EARL does not  
> really depend on, even if they are as important and relevant as UAAG or  
> ATAG.

I am using the SVG spec as a precedent. I don't think it is out of scope.  
It makes it harder to make a formally conforming tool, but it also means  
that we are clearly fitting in with W3C specs, not just picking and  
choosing things that we do.

Likewise, if we get EARL to recommendation I expect all subsequent W3C  
Recommendations conforming to SpecGL to include an explanation of how to  
make a conformance statement in EARL...

On the other hand I agree that these are up for discussion.

> It seems we need to define the following:
>
> 1. A formal and machine readable representation of EARL (aka an RDF  
> Schema) so that tools can validate their output (and input).

Well, tools can validate that something is RDF. This is independent of  
anything in EARL (except making sure the EARL schema is valid - which is  
why I would like to see the new draft)

> 2. A "minimal EARL", a set of classes and properties so that  
> interoperability of reports that contain these can be guaranteed.

Yeah, this is the idea of having OWL constraints, etc.

A further possible requirement is that a consuming tool can validate EARL  
according to whatever conformance criteria we set.

> 3. Tools are conformant to EARL when they produce at least valid  
> "minimal EARL", and/or interpret at least valid "minimal EARL" depending  
> on their nature (writers/readers).
>
> Is that sufficient with regard to the scope?

I don't think so. I think we should specify that things like producing  
"valid EARL plus other made-up EARL" as non-conformant behaviour, since  
this can harm interoperability (less so because EARL is RDF and not XML,  
but is still a bad idea).

Likewise I think that if there is minimal stuff and more advanced stuff,  
we should specify more than one level of conformance, so we know when to  
expect interoperability of the advanced stuff, rather than only clearly  
pushing it at the basic level.

cheers

Chaals

-- 
Charles McCathieNevile                              chaals@opera.com
          hablo español - je parle français - jeg lærer norsk
   Here's one we prepared earlier:   http://www.opera.com/download

Received on Wednesday, 25 May 2005 14:05:17 UTC