RE: Updated EARL 1.0 Schema Editors' Draft

 
 
> Carlos Iglesias wrote:
> > * 1.1 Structure of EARL Results
> > 
> > "Test Subject
> >   This can include: Web pages, tools..."
> > 
> > IMO it should be: Web CONTENT, tools...
> 
> These are only examples. Or do you dislike the word "pages"?

Yes, I'd like to have more web content examples (not only "pages") or use "Web content" instead. (Yes, I know, I'm quite fussy)
 
> > "Test requirement
> >   What are we evaluating the subject against?..."
> > 
> > IMO we should adopt the new wording to avoid confusion (i.e. "Test 
> > Criteria" instead of "Test requirement").
> 
> Yep, but the singular is "criterion".

Ok, "Test Criterion".

> > The same applies to the rest
> > of the document (Examples 2 and 3, the introduction at the 2.1 
> > Assertion section and the earl:result description at the 
> same section)
> > 
> > Other times "Test Statement"  is used to refeer "Test 
> Criteria", again 
> > I suggest use always "Test Criteria" (This happens in the earl:test 
> > description
> 
> Didn't we want to call the property earl:testable?

I'd prefer earl:test for the property. Testable sounds like a true/false property.

> > at the 2.1 Assertion section and in the earl:fail and 
> > earl:notApplicable descriptions at the 2.6.1 Validity level section)
> > 
> > 
> > * 2.2 Assertor
> > 
> > "foaf:Agent
> >   ... An Agent is a super class of foaf:Person, 
> foaf:Organization and 
> > foaf:Group which can all be used to describe an Assertor..."
> > 
> > Due to this description I think that the previous 
> foaf:Person section 
> > is redundant and not necessary (why a section for Person 
> and not for 
> > Organization or Group?)
> 
> How about something like:
> 
>    The assertor must be a foaf:Agent or a subclass thereof (e.g.
>    foaf:Person, foaf:Organization, foaf:Group or earl:Software).
> 
> Following some information on foaf:Agent and subclasses.

It's OK for me.

> > * 2.5 Test Mode
> > 
> > "earl:mixed
> >   Where there is no detailed information about the test mode 
> > available..."
> > 
> > IMO should be something like:
> > 
> >   Where THE TEST WAS PERFORMED BY AN UNKNOW COMBINATION OF AGENTS 
> > AND/OR TOOLS...
> > 
> > To avoid confusion with an "unknow" mode (no information at all)
> 
> Do we have an "unknown" mode?

No, we haven't. But IMO the current earl:mixed definition is more close to this non-existen unknown mode. What I want is to avoid people reading the definition and thinking "OK, this means unknow" because there is an essential difference:

Unknow --> No idea how the test where performed (only tools, only humans, tools and humans...)
Mixed --> The test was performed by an unknow combination of Agents AND Tools (sorry, ignore the "OR" in my previous message) i.e we don't have details but we KNOW that both (human and tools) where involved.

Additionally maybe we should add explicity an earl:unknow mode (based on the previous definitions)

> > * 2.6.2 Confidence Level
> > 
> > "...This may be used where a tool wants to assert..."
> > 
> > Is the confidence level supossed to be just for tools?
> 
> s/a tool/an assertor/

OK

> > * 2.7 Software Tool
> > 
> > I find the Example 13 obscure. It says:
> > 
> > "The software which was an Assertor in example 5 is now a 
> Test Subject"
> > 
> > But that's not clear in the example. IMO could be better:
> > 
> > The software which was an Assertor in example 5 could also 
> be used as 
> > a Test Subject and make use of the inherited properties
> 
> and then:
> 
>    <earl:subject>
>      <earl:Software rdf:about="#tool">
>        ...
>      </earl:Software>
>    </earl:subject>
> 

I don't think it's neccessary, but can live with it.

Regards,

CI.

 
--------------------------------------

Carlos Iglesias

CTIC Foundation
Science and Technology Park of Gijón
33203 - Gijón, Asturias, Spain 

phone: +34 984291212
fax: +34 984390612
email: carlos.iglesias@fundacionctic.org
URL: http://www.fundacionctic.org

Received on Thursday, 14 September 2006 10:40:45 UTC