W3C home > Mailing lists > Public > public-rdf-dawg@w3.org > April to June 2011

Re: JSON Results doc : First complete draft ready

From: Andy Seaborne <andy.seaborne@epimorphics.com>
Date: Tue, 17 May 2011 14:54:30 +0100
Message-ID: <4DD27E16.3040009@epimorphics.com>
To: public-rdf-dawg@w3.org


On 17/05/11 14:23, Axel Polleres wrote:
> Just to understand the issue about warnings/comments...
>
>>> Are there any examples of system that do add anything to the JSON format?
>>
>> Yeah, 4store apparently, it adds a "warnings": block after "results":, and puts an array of strings there: https://github.com/garlik/4store/blob/master/src/frontend/results.c#L1909
>>
>> I'm not sure who did that, but it's quite a good idea!
>
> Wouldn't that be non-conformant/problematic with the current spec?
>
> As a work-around,  could we allow to "add more keys" on top-level, i.e. do we/could we
> (i) state somewhere that adding keys except "results","boolean" on top level is allowed
> and (ii) give "warnings:" block from 4store as an (informative) example?
>
> Axel

(i) is just as non-conformant to the Note/spec.

Given the discussions awhile ago about the format of error messages, so 
they can be processed and displayed to people or used by software to act 
on them (and also language issues), I think any loose, under-constrained 
feature merely begs the question of format.  An example in a spec is 
tantamount to specification.

So, at most, note other keys may be present, or leave as-is.

	Andy

>
> On 17 May 2011, at 12:05, Steve Harris wrote:
>
>> On 2011-05-17, at 11:40, Andy Seaborne wrote:
>>
>>> Edits done.
>>>
>>>
>>> On 17/05/11 09:26, Steve Harris wrote:
>>>> A quick review:
>>>>
>>>> 3.1.1 "vars"
>>>>
>>>> The "var" member - Should be "vars" I think. A variable is not
>>>> necessary ->   A variable is not necessarily
>>>>
>>>> 3.2.1 "bindings"
>>>>
>>>> Possibly worth including an example of a solution with no bindings,
>>>> the text /could/ be interpreted as meaning it should be omitted.
>>>> Minor point.
>>>
>>> Example added, text reworded.
>>>
>>>> In XML it's possible to include runtime warnings and errors inline in
>>>> XML comments without messing up processing. It would be good if there
>>>> was some way to do something similar in JSON, e.g if there was an
>>>> "_comment" key (anything disjoint from variable names) which was
>>>> defined to be ignored. JSON doesn't have "native" comments. Minor
>>>> point.
>>>
>>> Tricky.
>>>
>>> The XML comments aren't an XML element - different processing model - and can go anywhere.
>>>
>>> I support there is nothing to stop an implementation adding "_comment" or "_warning" or etc if it wants to.
>>>
>>> We would have to define where it can go.
>>>
>>> We can:
>>> 1/ Leave the doc as-is - maximum backwards compatibility
>>> 2/ Define a key or keys and add some general text about it can go in any object.
>>> 3/ Define a key or keys and add to the doc like other keys.
>>>
>>> I'm inclined to go for (1) [do nothing]
>>>
>>> Are there any examples of system that do add anything to the JSON format?
>>
>> Yeah, 4store apparently, it adds a "warnings": block after "results":, and puts an array of strings there: https://github.com/garlik/4store/blob/master/src/frontend/results.c#L1909
>>
>> I'm not sure who did that, but it's quite a good idea!
>>
>> On the downside you can't include warnings inline, you have to save them up to the end, but it's less likely to confuse old processors.
>>
>> - Steve
>>
>>>> On 2011-05-12, at 11:34, Andy Seaborne wrote:
>>>>
>>>>> http://www.w3.org/2009/sparql/docs/json-results/json-results.xml
>>>>>
>>>>> There is now a complete 1st draft document (and fairly unchecked).
>>>>>
>>>>> It addresses the comments made (Richard, Michael, Paul) by
>>>>> restructuring the whole document and copying across content from
>>>>> before.  I wil go back and check the comments again but by the time
>>>>> the ties to the XML format are removed, it's a rather different
>>>>> document.
>>>>>
>>>>> I have tried to strike a balance and retained a more informal style
>>>>> but it should also be accurate.
>>>>>
>>>>> If anyone wants to have a go at JSON schema,
>>>>> http://json-schema.org/ then please do - I don't plan on doing that
>>>>> this time round.
>>>>>
>>>>> It's time for reviews.  It's not a long document.
>>>>>
>>>>> Andy
>>>>>
>>>>
>>
>> --
>> Steve Harris, CTO, Garlik Limited
>> 1-3 Halford Road, Richmond, TW10 6AW, UK
>> +44 20 8439 8203  http://www.garlik.com/
>> Registered in England and Wales 535 7233 VAT # 849 0517 11
>> Registered office: Thames House, Portsmouth Road, Esher, Surrey, KT10 9AD
>>
>>
>>
>
>
Received on Tuesday, 17 May 2011 13:55:02 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 16:15:46 GMT