Re: Test framework for Update

On 22/02/2010 5:43 PM, Paul Gearon wrote:
> Hi Andy,
>
> On Mon, Feb 22, 2010 at 4:46 AM, Andy Seaborne<andy.seaborne@talis.com>  wrote:
>>
>> The work was done by several of us (DanBri, Libby ...) then adopted by DAWG
>> (Steve, Lee, Jeen), Eric added annotation support for noting intersting
>> facts and the rdf:type got added (syntax test, bad synatx test, query test).
>>
>> mf:entries
>>    (
>>      [  mf:name    "Path-01" ;
>>         rdf:type   mfx:TestQuery ;
>>         mf:action
>>         [ qt:query<path-01.arq>  ;
>>           qt:data<data-1.ttl>  ] ;
>>         mf:result<path-01.srx>
>>      ]
>>   ....
>>
>> The structure of tests needs tweaking but it is basically there:
>> action-result, where the action has query-specific vocab for dataset
>> description and the query itself.
>
> Ah yes. I have ample experience with this, having implemented my own
> framework for parsing these tests and running them.
>
> The issue here is that it is not (yet) adequate for updates, hence the question.
>
>> What could work for update is the same overall theme:
>>
>> graph store - update script - result
>
> We're thinking along similar lines, though I was thinking more of:
>
> graph document - update operation(s) - graph document
>    or
> graph document - update operations(s) - [ query + result ]

What's a graph document?

Don't qt:data and qt:graphData work?  Or (better - because the graph 
name can be specified) the dataset description in service description? 
Or an NQuads file (TriG is OK too - NQ is just lowest common denominator).

Or put the initial data in a SPARQL Update script :-)

>
>> where the result is an NQuads file (lowest implementation cost).  The
>> dataset description would work for a graph store, I think, including empty
>> graphs (with care).
>>
>> You can get to all the scripted test via:
>>
>> http://jena.svn.sourceforge.net/viewvc/jena/ARQ/trunk/testing/ARQ/
>>
>> but there isn't scripted support for update.
>
> Darn. I was hoping you'd have something for testing SPARUL that we
> could build on. So I take it that this hasn't been automated?

It's automated in Java.

	Andy

>
> Regards,
> Paul Gearon
>
> ______________________________________________________________________
> This email has been scanned by the MessageLabs Email Security System.
> For more information please visit http://www.messagelabs.com/email
> ______________________________________________________________________

Received on Monday, 22 February 2010 21:16:51 UTC