W3C home > Mailing lists > Public > public-rdf-dawg@w3.org > October to December 2010

Re: Format of SPARQL Update results

From: Andy Seaborne <andy.seaborne@epimorphics.com>
Date: Thu, 23 Dec 2010 10:44:16 +0000
Message-ID: <4D132800.8030904@epimorphics.com>
To: Alexandre Passant <alexandre.passant@deri.org>
CC: Steve Harris <steve.harris@garlik.com>, SPARQL Working Group <public-rdf-dawg@w3.org>


On 23/12/10 10:32, Alexandre Passant wrote:
>
> On 23 Dec 2010, at 10:28, Andy Seaborne wrote:
>
>>
>>
>> On 23/12/10 10:20, Steve Harris wrote:
>>> On 2010-12-23, at 10:07, Alexandre Passant wrote:
>>>
>>>> Hi,
>>>>
>>>> Reading the discussion about JSON serialization of query results, I figured out that we don't have a proper format for results of SPARQL Update queries.
>>>> Since we agreed to send only success / failure as a result, I suggest to use the boolean return form (as for ASK) with true / false values depending on the success / failure (so that it does not require updates to the XML / JSON result formats).
>>>>
>>>> I don't think we need a formal vote for that, but if anyone objects or have other opinions, please let me know.
>>>> If not, I'll include that to the next round of publishing.
>>>
>>> Due to user requests we added some text along the lines of "12 triples added, 7 removed" in the response to a SPARQL Update requests in 5store. So, whatever format is specified, I'd like to be able to include some human readable text somewhere.
>>>
>>> Worst case it can go in a XML comment, but that's not ideal.
>>>
>>> - Steve
>>>
>>
>> The query protocol uses HTTP status codes.  200 for success , 400 for parse error and 500 for execution problems or any other appropriate HTTP status code.
>>
>> We could define the update protocol similarly, then leave the body content, if any, to implementations. c.f. query parse error messages.
>
> That would be ok, but what if one is using SPARQL Update without HTTP calls, e.g. a command-line client ?

The program returns success or failure via the usual mechanism of the 
OS?  exit(0) and exit(1)

Or the command line client can say "OK" or "that went badly" as it 
chooses - we don't specify the "protocol" for command line use in query.

My client dumps the dataset if you ask it to so you can play and see 
what the updates did.  Use only on small datasets!

> Or maybe we could extend the results with a "comment" field in the header (in addition to the current "link" one)

Maybe.  The results part of the format isn't going to be used so it 
seems to me to be a bit odd to use it. (Process point: we aren't 
revising the XML result format and the doc isn't reissued this time round.)

Steve's example shows a use case for certain information.  What's 
convenient to use for feedback is going to depend on the implementation. 
  e.g. whether that was a "change in size of a graph by 12" or "12 
triples sent" or "12 unique triples sent"

	Andy

>
> Alex.
>
>>
>> 	Andy
>>
>
> --
> Dr. Alexandre Passant
> Digital Enterprise Research Institute
> National University of Ireland, Galway
> :me owl:sameAs<http://apassant.net/alex>  .
>
>
>
>
>
>
Received on Thursday, 23 December 2010 10:44:54 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 16:15:44 GMT