Re: Comments on Query Results XML Format: Extraneous Elements, Boolean Results, and the Namespace

On Sun, 12 Jun 2005 21:16:53 -0700 (PDT), James Cerra <jfcst24_public@yahoo.com> wrote:

> 
> I am a little confused why the 'head' and 'results' elements exist.  What's
> their purpose beyond encapsulating the 'variable' and 'result' (or 'boolean')
> elements exactly?

The purpose of head at present to list the variable binding names and
give their order.  The purpose of results is to encapsulate the actual
variable binding or boolean results.  These are separate because they
have different roles, and that separation is a good enough reason.

>   I don't think that is a sufficient reason to include them as
> it unnecessarily complicates the XML tree.

It hardly complicates it, head and body are familiar concepts to anyone
who's seen an HTML document.  head is where the metadata goes about the main
content of the document, the body or in this case, the results.  Similar to
SOAP too with an envelope and contents.

head may also be useful for other things, as has just been commented on:
  http://lists.w3.org/Archives/Public/public-rdf-dawg-comments/2005Jun/0028.html

Other things the WG has discussed include passing query warning messages
to the results.  These would have not be for errors causing the query to fail -
a protocol error that HTTP for example, could handle - but some other info.
However, at this point the WG has found no useful warnings to give.

> Also, the 2.3.2. Boolean Results example indicates a head section even though
> it has no content.  How wasteful!  If you need it, I suggest using a document
> like:
> 
> <boolean xmlns="...">true</boolean>
> 
> That saves bandwidth and removes all unnecessary clutter.

I don't believe this is a strong argument.  The difference in bytes is
so small it probably does not make the document fit in one less network packet
or one less disk block, for most systems.

As for your claim of clutter, head is useful for the same reasons for boolean
results as it is for variable bindings.  Which brings you back to the structure
the current draft has.

> Finally the namespace for the XML format does not follow the conventions used
> by the W3C.  Why not use "http://www.w3.org/2004/12/sparql-query-results" which
> seems to follow the conventions of other standards like OWL and RDF Schema?

Yes, this is something we are fixing.  It'll likely change in the next WD.

Thanks for your comments

Dave

Received on Wednesday, 15 June 2005 08:50:26 UTC