Re: Preliminary EARL report

On Mar 14, 2012, at 1:28 AM, Ivan Herman wrote:

> Gregg,
> 
> this is a really great. My comments are really only cosmetic, and probably easy to add.
> 
> - Just to make it formally better, in terms of reporting at the end of the CR period, it is probably better to run the vocab section both with xml and xhtml1. I know the tests are identical, but better having that there instead of being asked to make additional explanations.

Yes, it's just a matter of collecting more results. I ran these results myself, and my intention is that implementors would send in their own reports to be incorporated. The tool will automatically add sections that it detects.

> - Actually, an SVG run would be nice as well. We know that, test-wise, it is the same as for XML, but good to have that, too (maybe in the Appendix, to avoid overcrowding the text)

See above; this was a preliminary run to get general comments on the output.

> - We should work on adding more info. I am not sure what is already available in the EARL reports. As far as I could see, the titles and the short descriptions of the tests are there; it may be worth adding these to the tables, additionally to the test numbers. Also, we should try to get some sort of an explanation for each of the sections (what do we mean by rdfa11.1-vocab, this sort of things). 

As you note, the title and descriptions are there, just hidden. We can work on the format of that.

I think the way to add descriptions generically, is to add a dc:description to each of the term definitions in[1], probably as an XMLLiteral that can be added in place into the report output. I'll get started on something for each, but others should feel free to improve this.

For the test cases, I think using the title instead of just the number, in each test is adequate, with a reference to an appendix where each generate test is described along with it's title, description, and other information from the test case definition.

> - This may have to be done manually, but we should also provide a short description for each of the processors: who did it, what does it offer, where is it available, etc. For example, the URI-s you provide there are, in some sense, not a reference to the tools themselves in general but, rather, the way an online service should be run. We should have links for both for each.

I agree; I think the best way is to relate the processor URL to a DOAP description. The DOAP description for rdf-rdfa (used by RDF.rb) is at [2]. Probably best to add something to the JavaScript file, so the doap:TestSubject references the doap:Project IRI, in my case <http://rubygems.org/gems/rdf-rdfa>. Note that, per EARL, we need to also assert earl:TestSubject on the same IRI.

By adding some more information to the JavaScript processor definition, we can report using that IRI instead of the processor IRI. Then, other information from the doap:Project description can be used to flesh out the processor description, in this case doap:description, dc:creator and doap:homepage.

If processor implementors can either provide a URL for such a DOAP project description, or send one to me, I'll make sure that it's used when creating the reports.

> - I would put the 'individual test results' list to the end. Hopefully we will have more than just these three, and that list may become very long. Maybe putting them in a tabular format is also a possibility. Also, we should add a remark that these links are to the EARL reports in RDFa (and maybe add a separate link to get directly to the turtle format of the individual reports).

Sure, I can automatically create Turtle versions of each of these reports and extract some information to make it clear the processor, version and host language the individual report is for.

Gregg

> As I said, I believe all these are really minor issues only. All in all, it looks great!
> 
> Huge thank!
> 
> Ivan

[1] http://rdfa.info/vocabs/rdfa-test.html
[2] https://raw.github.com/gkellogg/rdf-rdfa/master/etc/doap.html

> On Mar 14, 2012, at 01:46 , Gregg Kellogg wrote:
> 
>> I ran reports for Ivan's and my processors against XML, XHTML1, and HTML5 along with vocabulary-expansion for XML. (Note, XML1 has been renamed to just XML, at Ivan's suggestion). I collected all the reports in the /earl-reports directory, with a consolidated report available at [1]. There could obviously be some better explanatory information, but the core processing work is done. Note this information is (of course) encoded with RDFa, and is available through some alternative representations, as described in the report.
>> 
>> Note that between Ivan and I, we pass everything (except 0235 in XHTML1 for mine, for some odd reason). I think there may be some remaining XMLLiteral issues to be worked out.
>> 
>> About report generation:
>> 
>> Individual reports are generated as HTML+RDFa and parsed into a common graph, which is then output as both Turtle and JSON-LD. Note that the structure of the EARL report has been supplemented to allow for the JSON-LD framing. This is then used directly in a Ruby/Haml template to generate the HTML report.
>> 
>> Feedback appreciated!
>> 
>> Gregg
>> 
>> [1] http://rdfa.info/earl-reports/
> 
> 
> ----
> Ivan Herman, W3C Semantic Web Activity Lead
> Home: http://www.w3.org/People/Ivan/
> mobile: +31-641044153
> FOAF: http://www.ivan-herman.net/foaf.rdf
> 
> 
> 
> 
>

Received on Wednesday, 14 March 2012 17:17:43 UTC