Re: Protocol validator

I've addressed both of the issues noted below and now see the latest Fuseki snapshot build passing all protocol tests.

.greg


On Aug 27, 2012, at 4:54 PM, Gregory Williams wrote:

>> I'm now at 30/34 failing these tests.
>> 
>>   update_dataset_default_graphs
>>   update_dataset_named_graphs
>>   update_dataset_full
>>   bad-update-missing-direct-type
>> 
>> The first 3 are for the same reason.
>> 
>> Issue 1:
>>   update_dataset_default_graphs
>>   update_dataset_named_graphs
>>   update_dataset_full
>> 
>> Could you explain these please?  From logging the Fuseki server requests received I don't see how they are supposed to return the ASK result of true that is expected:
> 
> It sounds like I just made a mistake in the tests. I'll look into it and make adjustments as necessary.
> 
>> Issue 2:
>>  bad-update-missing-direct-type
>> 
>> This test is sending a SPARQL Update without content type.
>> 
>> With a no "Content-type" request, Fuseki assumes "application/sparql-update" which makes it easier for simple clients to open a POST stream and send an update.
>> 
>> I don't see anything in the spec that requires an error if the content type is not 'application/x-www-url-form-urlencoded' and not 'application/sparql-update'.
>> 
>> My reading is that it is simply outside the spec at this point so a server that does something is not non-conformant, just helpful.
> 
> Yeah, that's a reasonable understanding of the spec. I wrote a bunch of tests that were closer to best-practice tests than normative tests, but that one seems to have been included in the normative set. I'll remove it (or make it optional or something).

Received on Sunday, 9 September 2012 23:31:14 UTC