W3C home > Mailing lists > Public > public-rdf-dawg@w3.org > April to June 2005

Re: ordered results in the results format specs

From: Jeen Broekstra <jeen@aduna.biz>
Date: Tue, 19 Apr 2005 12:08:01 +0200
Message-ID: <4264D881.4090905@aduna.biz>
To: andy.seaborne@hp.com
Cc: public-rdf-dawg@w3.org

Seaborne, Andy wrote:

> Presumably the indexing information is only added if the query involved 
> ORDER BY.

Yes, that was what I intended.

> A query test with no implication of order would generate solutions in 
> any order and a test for graph equivalence would continue to work.

Right.

I've just ran into a small problem with this approach though. The 
ordering as currently specified in rq23 is not total: for example the 
order between two results with an unbound variable (if the sorting is 
on that variable) is not specified. See sorting test case 3 for an 
example.

The problem is not so much that there is no total order, but rather 
that our clever indexing information assumes there is. How do we index 
two unordered solutions?

In the currently checked in solution file, I'm just assigning 
unordered solutions the same index digit, leading to order numbering: 
1 1 3 4 (four solutions total, first two unordered).

In the XML format, this is a bigger problem, I just realized, since we 
don't use digits but rather just set 'order=true' for the entire 
result. This can only work properly for a total ordering.

Jeen
-- 
Jeen Broekstra          Aduna BV
Knowledge Engineer      Julianaplein 14b, 3817 CS Amersfoort
http://aduna.biz        The Netherlands
tel. +31 33 46599877
Received on Tuesday, 19 April 2005 10:06:06 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 16:15:23 GMT