W3C home > Mailing lists > Public > public-rdf-dawg@w3.org > January to March 2006

Re: comment: QName production in SPARQL grammar; reconsider punctuationSyntax?

From: Dan Connolly <connolly@w3.org>
Date: Wed, 04 Jan 2006 08:45:45 -0600
To: andy.seaborne@hp.com
Cc: RDF Data Access Working Group <public-rdf-dawg@w3.org>
Message-Id: <1136385945.20839.307.camel@dirk.w3.org>

On Thu, 2005-12-15 at 11:08 +0000, Seaborne, Andy wrote:
> 
> Dan Connolly wrote:
> > JJC asks that we re-consider our position on punctuationSyntax;
> > (a) he asks that foo:123 be allowed
> 
> The restriction to no leading digits in the local part does catch people out. 

I don't doubt it. The question is: was the WG aware of that when
we closed punctuationSyntax? Or is this new information?

For myself, I was well aware of the limitations of XML name
syntax that SPARQL was inheriting; I assume others were as
well, but if not -- i.e. if you see new information --
please say so.


>   There is no necessity in SPARQL for the restriction.
> 
> A more conservative change is:
> 
> OLD::
> [90]   	NCNAME 	  ::=   NCCHAR1 ((NCCHAR|'.')* NCCHAR)?
> 
> NEW::
> [90]   	NCNAME 	  ::=   (NCCHAR1 | [0-9]) ((NCCHAR|'.')* NCCHAR
> 
> JJC's suggested change would make other thing legal as well (and make
> 
>   ns:2  legal (he included leading dots and  minuses as well as #x00B7 | 
> [#x0300-#x036F] | [#x203F-#x2040] in the local part).
> 
> 
> This still means that an IRI like:
> 
> <http://www.w3.org/2005/08/sparql-protocol-query/#wsdl.interface(SparqlQuery)>
> 
> can't be abbrevated except with the whole thing in the prefix.
> 
> 
> I checked - this change does not fail any DAWG tests.
> 
> This works because lexing is greedy so "ns:2" it isn't seen as "ns:" and the 
> number "2".  We depend on this in quite a few ways (e.g. the letters of 
> keyword like SELECT inside a qname).
> 
> <insert type="request for negative syntax tests"/>please<insert>
> 
> > (b) he suggests editorial changes that disconnect
> > the N3/turtle/SPARQL foo:bar notation from XML QNames
> 
> I'm not inclinded to change the test names.
> 
> I'm inclined to make the wording in the document use qname less (it's just 
> 2.1.1 and 2.1.2)
> 
> I'm (mildly) not inclined to change the grammar rule names and the text about 
> the rules.
> 
> > 
> > http://lists.w3.org/Archives/Public/public-rdf-dawg-comments/2005Dec/0018.html
> > 
> > It's not obvious to me that there's sufficient new information
> > to reconsider the decision. I'm inclined to let him know that
> > we have previously considered this issue and show him our
> > decision records, but anyone who is persuaded by JJC's arguments
> > that we should take another look will please say so.
> > 
> > These tests make it pretty clear to me that we considered all sorts
> > of details about qname syntax...
> > 
> > http://www.w3.org/2001/sw/DataAccess/tests/#syntax-qname-07-rq
> > syntax-qname-07.rq
> >         Approved
> > syntax-qname-08.rq
> >         Approved
> > syntax-qname-09.rq
> >         Approved
> > syntax-qname-10.rq
> >         Approved
> > syntax-qname-11.rq
> >         Approved
> > syntax-qname-12.rq
> >         Approved
> > syntax-qname-13.rq
> >         Approved
> > syntax-qname-14.rq
> >         Approved
> > http://www.w3.org/2001/sw/DataAccess/tests/#syntax-qname-14-rq
> > 
> > 
> 
-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/
D3C2 887B 0F92 6005 C541  0875 0F91 96DE 6E52 C29E
Received on Wednesday, 4 January 2006 14:45:55 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 16:15:25 GMT