Re: RDF syntax variations and EARL test suites

Hi,

At 18:00 25-11-2011, Shadi Abou-Zahra wrote:
>Hi Carlos and all,
>
>[[
>That is why any attempt to treat RDF as XML is damned to fail.
>]]

Sure, but AccessODF's EARL code was read as RDF, otherwise the 
OpenOffice.org RDF library would have choked on the EARL code 
produced by AccessODF in LibreOffice. The reading of "RDF as XML" was 
merely the effect of extracting the EARL from ODF files processed by 
LibreOffice and OpenOffice.org and comparing them in an editor.



>I wasn't aware you felt so strongly about this. Actually one of the 
>assumptions we had (if my recollection doesn't fail me) is that we 
>wanted to use the XML serialization in our code examples to support 
>tool developers who may be more XML aware than, say, Turtle.
>
>RDF'ers much prefer Turtle and claim it is more readable. I believe 
>there have been comments in this direction in the past. Do you think 
>it's time to switch our examples from XML to Turtle, and mention XML 
>serialization in the Guide (the opposite of what we currently have)?

Section 1.1 of our spec says: "This document assumes that the reader 
is familiar with the Resource Description Framework (RDF) and can 
read its XML serialization." But we never say that the code examples 
are merely XML serialisations of EARL/RDF graphs. Perhaps we should 
emphasize that more, e.g. in Section 1.2 (Document conventions) by 
adding something like: "This document uses XML as the serialization 
format for EARL. EARL serialized in XML syntax is a representation of 
RDF graphs, and EARL test results should be compared by comparing the 
RDF graphs they represent, not by comparing their concrete syntax."


>Similarly, is it time to reconsider conformance requirements B and C?
>  - <http://www.w3.org/TR/EARL10-Guide/#conformance>

So these conformance requirements insist on the use of XML serialisation.
What is the intent of our requirements? To make sure that tools that 
only support XML serialisation can really parse EARL as RDF, not just 
as XML? Or is the intent to require XML syntax, even if/when we 
switch over our examples to Turtle?

Getting ready to produce test suites really focuses the mind ;-)

Best regards,

Christophe



>Regards,
>   Shadi
>
>
>On 25.11.2011 08:38, Carlos A Velasco wrote:
>>hi Christophe,
>>
>>These are different serializations of the same report. Any tool parsing
>>EARL should be able to read those two "models" (in Jena jargon) and find
>>them equivalent. That is why any attempt to treat RDF as XML is damned
>>to fail.
>>
>>I tried to highlight this issue in the Guide.
>>
>>If you don't configure your tool adequately, for instance in Jena, it
>>may be that you become as prefixes j.0, j.1, etc. And being honest, it
>>does not matter :-)
>>
>>On 24/11/11 14:42, Christophe Strobbe wrote:
>>>Hi,
>>>
>>>The discussions about using the dc or the dct namespace for Dublin Core
>>>metadata terms are essentially about syntax. This brings up the
>>>question: what about bigger variations in syntax?
>>>
>>>For example, when I use AccessODF
>>><http://sourceforge.net/projects/accessodf/> in OpenOffice.org 3.3 and
>>>in LibreOffice 3.4.3, the EARL syntax is not the same. The EARL reports
>>>are saved in RDF files inside the ODF files, so it is possible to
>>>extract the reports and compare them.
>>>
>>>One thing that is striking is that the dct, earl, foaf and doap
>>>namespace prefixes are nowhere to be seen: you just get ns0, ns1,
>>>etcetera, accompanied by a namespace declaration as in: <ns0:result
>>>xmlns:ns0="http://www.w3.org/ns/earl#" ...>.
>>>
>>>Another striking thing is the difference in syntax between LibreOffice
>>>and OpenOffice.org. In the RDF from LibreOffice, everything is
>>>completely flattened, like so:
>>>
>>><rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#">
>>><rdf:Description rdf:nodeID="r1r4520r357">
>>><ns0:assertedBy xmlns:ns0="http://www.w3.org/ns/earl#"
>>>rdf:resource="http://docarch.be/accessibility/ooo/InternalChecker"/>
>>></rdf:Description>
>>><rdf:Description rdf:nodeID="r1r4520r357">
>>><ns0:subject xmlns:ns0="http://www.w3.org/ns/earl#"
>>>rdf:nodeID="r1r4520r359"/>
>>></rdf:Description>
>>><rdf:Description rdf:nodeID="r1r4520r357">
>>><ns0:test xmlns:ns0="http://www.w3.org/ns/earl#"
>>>rdf:resource="http://www.docarch.be/accessibility-checker/checks#E_NoDefaultLanguage"/>
>>>
>>>
>>></rdf:Description>
>>><rdf:Description rdf:nodeID="r1r4520r357">
>>><ns0:result xmlns:ns0="http://www.w3.org/ns/earl#"
>>>rdf:nodeID="r1r4520r358"/>
>>></rdf:Description>
>>><rdf:Description rdf:nodeID="r1r4520r357">
>>><rdf:type rdf:resource="http://www.w3.org/ns/earl#Assertion"/>
>>></rdf:Description>
>>><rdf:Description rdf:nodeID="r1r4520r358">
>>><ns0:date
>>>xmlns:ns0="http://purl.org/dc/elements/1.1/">2011-11-23T19:23:49+0100</ns0:date>
>>>
>>>
>>></rdf:Description>
>>>(...)
>>></rdf:RDF>
>>>
>>>In OpenOffice.org, it looks a little bit nicer:
>>>
>>><rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#">
>>><ns1:TestCase xmlns:ns1="http://www.w3.org/ns/earl#"
>>>rdf:about="http://www.docarch.be/accessibility-checker/checks#E_NoDefaultLanguage"/>
>>>
>>>
>>><ns2:TestCase xmlns:ns2="http://www.w3.org/ns/earl#"
>>>rdf:about="http://www.docarch.be/accessibility-checker/checks#A_FormulaWithoutAlt"/>
>>>
>>>
>>><ns3:TestCase xmlns:ns3="http://www.w3.org/ns/earl#"
>>>rdf:about="http://www.docarch.be/accessibility-checker/checks#E_ImageAnchorFloat"/>
>>>
>>>
>>><ns4:TestCase xmlns:ns4="http://www.w3.org/ns/earl#"
>>>rdf:about="http://www.docarch.be/accessibility-checker/checks#E_UnsupportedImageFormat"/>
>>>
>>>
>>><ns5:TestCase xmlns:ns5="http://www.w3.org/ns/earl#"
>>>rdf:about="http://www.docarch.be/accessibility-checker/checks#A_ImageWithoutAlt"/>
>>>
>>>
>>><ns6:TestCase xmlns:ns6="http://www.w3.org/ns/earl#"
>>>rdf:about="http://www.docarch.be/accessibility-checker/checks#A_FakeTable"/>
>>>
>>>
>>><ns7:TestCase xmlns:ns7="http://www.w3.org/ns/earl#"
>>>rdf:about="http://www.docarch.be/accessibility-checker/checks#E_EmptyHeading"/>
>>>
>>>
>>><ns8:TestCase xmlns:ns8="http://www.w3.org/ns/earl#"
>>>rdf:about="http://www.docarch.be/accessibility-checker/checks#A_MergedCells"/>
>>>
>>>
>>><ns9:TestCase xmlns:ns9="http://www.w3.org/ns/earl#"
>>>rdf:about="http://www.docarch.be/accessibility-checker/checks#A_BreakRows"/>
>>>
>>>
>>><ns10:TestCase xmlns:ns10="http://www.w3.org/ns/earl#"
>>>rdf:about="http://www.docarch.be/accessibility-checker/checks#A_FakeHeading"/>
>>>
>>>
>>><ns11:TestCase xmlns:ns11="http://www.w3.org/ns/earl#"
>>>rdf:about="http://www.docarch.be/accessibility-checker/checks#A_NoHyperlinkText"/>
>>>
>>>
>>><ns12:TestCase xmlns:ns12="http://www.w3.org/ns/earl#"
>>>rdf:about="http://www.docarch.be/accessibility-checker/checks#A_JustifiedText"/>
>>>
>>>
>>><ns13:Checker xmlns:ns13="http://www.docarch.be/accessibility-checker/"
>>>rdf:about="http://docarch.be/accessibility/ooo/InternalChecker">
>>><rdf:type rdf:resource="http://www.w3.org/ns/earl#Assertor"/>
>>></ns13:Checker>
>>><ns14:TestCase xmlns:ns14="http://www.w3.org/ns/earl#"
>>>rdf:about="http://www.docarch.be/accessibility-checker/checks#A_SmallText"/>
>>>
>>>
>>><ns15:Assertion xmlns:ns15="http://www.w3.org/ns/earl#">
>>><ns15:assertedBy
>>>rdf:resource="http://docarch.be/accessibility/ooo/InternalChecker"/>
>>><ns15:result>
>>><ns15:TestResult>
>>><ns16:date
>>>xmlns:ns16="http://purl.org/dc/elements/1.1/">2011-11-24T14:23:11+0100</ns16:date>
>>>
>>>
>>><ns15:outcome rdf:resource="http://www.w3.org/ns/earl#failed"/>
>>></ns15:TestResult>
>>></ns15:result>
>>><ns15:subject>
>>><ns17:Document
>>>xmlns:ns17="http://www.docarch.be/accessibility-checker/types#">
>>><rdf:type rdf:resource="http://www.w3.org/ns/earl#TestSubject"/>
>>></ns17:Document>
>>></ns15:subject>
>>><ns15:test
>>>rdf:resource="http://www.docarch.be/accessibility-checker/checks#E_NoDefaultLanguage"/>
>>>
>>>
>>></ns15:Assertion>
>>>(...)
>>></rdf:RDF>
>>>
>>>We need to be aware of this when thinking about our test suite.
>>>
>>>On second thought, I also think this syntax difference provides an
>>>opportunity for the documentation of implementations: the RDF libraries
>>>inside LibreOffice and OpenOffice.org provide different outputs
>>>(syntactically), but appear to be able to read each other's format (as
>>>they should, anyway). You can test this by checking an ODF document in
>>>the first office suite, then saving it (to save the EARL report), and
>>>then opening it in the other office suite, and checking whether
>>>AccessODF displays the same list of errors and warnings in its UI. I
>>>have tested this for the first time, and it works. Maybe we can use this
>>>as evidence for support of both output and input.
>>>
>>>Best regards,
>>>
>>>Christophe
>>>
>
>--
>Shadi Abou-Zahra - http://www.w3.org/People/shadi/
>Activity Lead, W3C/WAI International Program Office
>Evaluation and Repair Tools Working Group (ERT WG)
>Research and Development Working Group (RDWG)

-- 
Christophe Strobbe
K.U.Leuven - Dept. of Electrical Engineering - SCD
Research Group on Document Architectures
Kasteelpark Arenberg 10 bus 2442
B-3001 Leuven-Heverlee
BELGIUM
tel: +32 16 32 85 51
http://www.docarch.be/
Twitter: @RabelaisA11y
---
Open source for accessibility: results from the AEGIS project 
www.aegis-project.eu
---
Please don't invite me to Facebook, Quechup or other "social 
networks". You may have agreed to their "privacy policy", but I haven't.

Received on Friday, 25 November 2011 17:44:47 UTC