W3C home > Mailing lists > Public > public-rdf-dawg-comments@w3.org > October 2006

Re: JSON from SPARQL for multiple assertions

From: Lee Feigenbaum <feigenbl@us.ibm.com>
Date: Sat, 7 Oct 2006 17:49:18 -0400
To: conor325 <conor@the325project.org>
Cc: public-rdf-dawg-comments@w3.org
Message-ID: <OFF8AF3610.22D54472-ON85257200.00776A87-85257200.0077DEE3@us.ibm.com>
public-rdf-dawg-comments-request@w3.org wrote on 10/07/2006 05:20:41 PM:

> thx Lee.
> BTW, does the JSON format have to directly ape the XML form - so that 
> if the XML form has multiple bindings then the JSON form must too? 
> The extra binding - or many extra bindings if there are many multiple 
> assertions - doesn't seem to serve the purpose of JSON which I think 
> is to be trivial for a client to process?

There is no requirement in the DAWG's charter to produce a JSON 
serialization of results, so it's not really appropriate to speak of what 
the JSON format did or did not "have to" be. Rather, the editors 
(including myself) felt that it was appropriate to make the JSON format as 
predictable as possible, and we felt this would be accomplished if it were 
as similar as was reasonable to the XML format.

<takes off DAWG member hat>

As for the triviality of processing, we prefer the JSON format in our 
browser-based applications of SPARQL due to the lightweight nature of 
parsing the structures. Instead of wielding an XML parser, the JSON 
structure requires a regular expression substitution, a regular expression 
check, and a call to eval() to be in a usable state. Once there, we do 
find it very straightforward to transform between the SPARQL Query Results 
JSON format and simpler JavaScript data structures. For example, the 
transformation which makes the selectValues() form of querying possible in 
the sparql.js library is nothing more than:

    selectValues: function (o) {
        var v = o.head.vars[0]; // assume one variable
        var values = [];
        for (var i = 0; i < o.results.bindings.length; i++)
        return values;

Received on Saturday, 7 October 2006 21:49:28 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:52:07 UTC