W3C home > Mailing lists > Public > public-rdf-dawg-comments@w3.org > February 2005

Re: Question/Comment on FROM and WITH

From: Alberto Reggiori <alberto@asemantics.com>
Date: Wed, 23 Feb 2005 18:28:01 +0100
Message-Id: <82ea8bec229241a934a1ccfaab22fa6c@asemantics.com>
Cc: <andy.seaborne@hp.com>, <public-rdf-dawg-comments@w3.org>
To: "Geoff Chappell" <geoff@sover.net>

On Feb 23, 2005, at 6:12 PM, Geoff Chappell wrote:
>
> I'll predict that people are going to get pretty confused about 
> FROM/WITH if
> they remain as they are.

indeed - and your worry confirms other feelings around looking at 
current SPARQL RDF-dataset selection syntax. Lots of people keeps 
associating SPARQL to RDQL, which used FROM. And they will still do so 
for another while, till they will get used to a new syntax (and 
design).

>  I wonder if a slightly different model would make
> more sense to people - e.g. something like this:
>
> SELECT ?src
> FROM <http://example.org/foaf/aliceFoaf> 
> <http://example.org/foaf/bobFoaf>
> AS <http://example.org/foaf/allFoaf>

right - or add a NAMED modifier to FROM when is dealing with a named 
graph instead

which would mean a simple change into current spec

	s/FROM/FROM NAMED/
	s/WITH/FROM/

and would avoid lot of confusion

> I.e. use AS to rename one or more graphs; if no AS is specified each 
> graph
> is known by its own uri/name. You'd need to come up with a well-known 
> name
> for the default/background/nameless graph - e.g:

yes your AS modifier would play a similar trick too

>
> SELECT ?src
> FROM  <http://example.org/foaf/aliceFoaf>
> 	<http://example.org/foaf/bobFoaf>
>  		AS <http://example.org/foaf/allFoaf>,
> 	<http://example.org/otherData>
> 		AS <http://www.sparql.org/DefaultGraph>
> ...
>
> would have one named graph (allFoaf - containing bob and alice's foaf 
> data)
> and one background graph (DefaultGraph containing otherData).
>
> Then:
>
> WHERE
>      ( ?x ?y ?z )
>
> would match all triples in the default
> (<http://www.sparql.org/DefaultGraph>) graph.
>
> Or Alternatively, have no name for the background graph and say that 
> if no
> AS is specified, the graph(s) are in the background graph - e.g:
>
> SELECT ?src
> FROM  <http://example.org/foaf/aliceFoaf>
> 	<http://example.org/foaf/bobFoaf>
>  		AS <http://example.org/foaf/allFoaf>,
> 	<http://example.org/otherData>
> ...
>
> would have one named graph (allFoaf - containing bob and alice's foaf 
> data)
> and one background graph (containing otherData).
>
> I don't think the slight benefit you get from the FROM/WITH shorthand
> outweighs the confusion it will likely cause.

we fully support your view and worries here - and it would be good to 
find more common syntax/terminology to specify the RDF-dataset in the 
LC document. True, this is purely syntax problem, but very key to the 
wider adoption of SPARQL to us.

thanks for the comments Geoff - and to Andy to follow up the thread

Yours

Alberto

-
Alberto Reggiori, @Semantics S.R.L.
www.asemantics.com
Received on Wednesday, 23 February 2005 17:28:06 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 8 January 2008 14:14:47 GMT