Re: First draft of test cases

El 20/05/13 16:03, Eric Prud'hommeaux escribió:
> * Raúl García Castro <rgarcia@fi.upm.es> [2013-05-10 14:55+0200]
>> Dear all,
>>
>> Miguel and myself have prepared a first draft of the LDP test cases.
>> You can find it here:
>> https://dvcs.w3.org/hg/ldpwg/raw-file/default/Test%20Cases/LDP%20Test%20Cases.html
>>
>> The document includes:
>> .- Some design issues (important to read).
>> .- Description of the testing process.
>> .- How to describe testing artifacts in RDF.
>> .- Description of test cases.
>>
>> It also identifies different issues to be discussed regarding the
>> current specification:
>> https://dvcs.w3.org/hg/ldpwg/raw-file/default/Test%20Cases/LDP%20Test%20Cases.html#Feedback
>>
>> Some of these issues are related to others currently under
>> discussion, so I suggest you to take a look at them.
>> Also, I don't know whether to create issues in the issue tracker for
>> them or to discuss them previously on the mailing list. Arnaud, any
>> suggestion?
>>
>> I know that Eric is currently taking a look at the document and that
>> Sergio is willing to do so. Nevertheless, anyone is free to comment
>> on it, don't be shy.
>
> Eric's belated review:
>
> Excellent job! now the annoying nits:

Hi Eric,

Thanks a lot for your comments.

I have implemented them in the current version of the document. Some 
detailed comments follow below.

> globally, see if you can substitute "concrete implementation" with "specific implementation" in e.g. "dependent on concrete implementations".
>
> https://dvcs.w3.org/hg/ldpwg/raw-file/default/Test%20Cases/LDP%20Test%20Cases.html#ProtocolVsData
>    I think LDPCs required either
>      { <C> rdfs:member <eltⁿ> . }
>    or
>      { <C> ldp:membershipPredicate <mp> ; <mp> <eltⁿ> }
>    in addtion to rdf:type.

Whether rdfs:member is required or not is being discussed in another 
thread. Once that issue is solved, I will update the document.

> https://dvcs.w3.org/hg/ldpwg/raw-file/default/Test%20Cases/LDP%20Test%20Cases.html#Coverage
>    use markup to hilight "LDP Core", perhaps following conventions in RDF Concepts
>      https://dvcs.w3.org/hg/rdf/raw-file/default/rdf-concepts/index.html#section-rdf-graph

Thanks for the pointer. I have followed the conventions there and used 
<dfn>.

> https://dvcs.w3.org/hg/ldpwg/raw-file/default/Test%20Cases/LDP%20Test%20Cases.html#ResultVsAssertion
>    s/successfully pass a test all the
>     /successfully pass a test, all of the/
>
> https://dvcs.w3.org/hg/ldpwg/raw-file/default/Test%20Cases/LDP%20Test%20Cases.html#ResultVsAssertion
>    If SPARQL tests are a precedent, we don't need to explain why we demand pass/fail claims intead of execution results. I propose
> [[
> Testors submit execution results, e.g. the HTTP message returned by a POST, and an <a href="...">EARL</a> assertion of either <code>earl:passed</code> or <code>earl:failed</code>.
> ]]

OK, I have changed the process, now we don't make the assertions but 
they are already included in the data that they report.

> https://dvcs.w3.org/hg/ldpwg/raw-file/default/Test%20Cases/LDP%20Test%20Cases.html#Traceability
>    does "is "should be related to those documents that are relevant for it" to prev" mean "link to the related specs"?
>    if so, i propose instead (with some fiddling around for @@1 and @@2):
>
>      Invoking the tests produces produces resulting @@1 and @@2s.
>      An <Implementation Report> includes @@1 and @@2 and rdf:seeAlso links to related documents, e.g., specifications, uses cases, etc.

Tests are related to specifications, use cases, etc. Therefore, any 
assertion that is related to a certain test will be also related to the 
corresponding document through the test.

>    I note that you used "test execution report" instead of e.g. "Implementation Report". I can't remember if in SPARQL tests, "implementation report" referred to a submittedEARL report for an implementation or the summary of those which is called an "implementation report" in the W3C process. It would be nice to align with SPARQL if they used some principled nomenclature.
>
> https://dvcs.w3.org/hg/ldpwg/raw-file/default/Test%20Cases/LDP%20Test%20Cases.html#TestingProcess
>    s/not compulsory/not required/ # "required" is a more appropriate term of art

I'm also using the W3C naming of "Implementation report". With the 
change about it should be now clearer.

>    I think #2 sounds like a spec for a summary report. I Propose to move it there once it's created.

Sure, we can move the part related to the testing process to the 
implementation report once it is ready.

Kind regards,

-- 

Dr. Raúl García Castro
http://delicias.dia.fi.upm.es/~rgarcia/

Ontology Engineering Group
Departamento de Inteligencia Artificial
Universidad Politécnica de Madrid
Campus de Montegancedo, s/n - Boadilla del Monte - 28660 Madrid
Phone: +34 91 336 36 70 - Fax: +34 91 352 48 19

Received on Monday, 3 June 2013 14:06:50 UTC