Re: Update

good morning;

> On 2015-10-12, at 14:54, Karima Rafes <karima.rafes@gmail.com> wrote:
> 
>>> […]
>> 
>> 
>> Each of us may well think some choices are better than others, or that the
>> specs should have said something else, but that's not the point when
>> curating the W3C test suites.
> 
> For the tests, I can use the Request Content Type (in the reference)
> in the header HTTP but the basic developer JS uses the specific
> parameter HTTP GET of triplestore and each triplestore implement a
> different manner:
> output=json
> output=application/sparql-json...
> format=jsonp
> etc...
> the reference of protocol SPARQL  is again not enough…

where a test suite which is associated with a protocol, its intent is to demonstrate conformance to that protocol.
this applies, by extension, to a framework which intends to evaluate the behavior of a service under that suite,
should it be necessary to observe and adhere to such variations in order to obtain results, then the respective service is non-conformant.

if the framework’s intent is to document variations and inform implementers as to how they may need to adapt their clients, rather than to evaluate conformance to the test suite, this principle might not apply.

best regards, from berlin,
---
james anderson | james@dydra.com | http://dydra.com