- From: Dave Beckett <dave.beckett@bristol.ac.uk>
- Date: Tue, 15 Feb 2005 23:48:50 +0000 (GMT)
- To: "Seaborne, Andy" <andy.seaborne@hp.com>
- cc: Alberto Reggiori <alberto@asemantics.com>, 'RDF Data Access Working Group' <public-rdf-dawg@w3.org>
On Tue, 2005-02-15 at 17:02 +0000, Seaborne, Andy wrote: > Thanks for the note - for now I have made some changes for publication. > > > For this week's publication the state is: > 1/ BOUND(?x) - no change > > 2/ Left as AND for now. > > 3/ Just OPTIONAL, not [] > > (note that OPTIONAL can take a {} pattern or a plain triple pattern - > restricting to just {} would be OK). > > 4/ No change to functions/casting/builtins although this does look like > it can be simplified - it just requires more time than is available before > publication. > > 5/ WITH done. > > 6/ Clause order unchanged > > Andy > > Alberto Reggiori wrote: > I basically support everything alberto said (snipped). Keep BOUND(?x) AND should stay like that for now and for later - I see no reason to change this. OPTIONAL { } is OK, dropping []. My preference only OPTIONAL { } and not with plain triple. Casting etc. - I want to see what you come up with. Clause order. I don't care where BASE, PREFIX, WITH, FROM and LIMIT come as long as they are well defined and they execute in query syntax order. So sticking to syntax things, and not doing an rq23 doc review (skipping reading most of the words) this is my current syntax comments. 2.1 found typo 'ELECT' in one of the examples 2.5 "BNodes can't appear in a SPARQL query." yes they can. 10.2 shows a construct query with bnodes. "Update when XML result syntax document is available and use the syntax there." Please can you link to it now. end of 3 "Open: whether to allow "foo"@?v or ?v@fr or ?v^^xsd:integer or "foo"^^?v" I'd say let's close this. No. You can't query inside RDF's "atoms". It looks to me that 3.1 syntax examples decides this anyway. 5.1 OPTIONAL{ (?s ?p ?o } my preference as I said above 5.2 "Constraints can be given in optional blocks" This surprises me. All constraints are global - this seems to imply a new feature of scoped constraints. 6 UNION I have problems with since I objected to disjunction. 8.4 uses FROM and GRAPH to make a dataset without explanation before section 9 that explains what it means. Should be WITH anyway? 9.1 reads that WITH only may allow multiple URIs but 9.2 gives a multiple URI example for FROM implying that this is OK? 10.1 you can link to the syntax in the VBR document 10.2 UNION construct - same problems. 10.4 yes or no "Align results to XML results format" Did we decide this? I feel there might be a dropped for me action here somewhere. Table 11.1 The links are just too tiny in Operator column I'd prefer a new column Definition and larger links say "X.Y" section numbers Can the sub-sections XQuery Connectives etc. be highlighted or centred for emphasis? Appendix A grammar It's still not in the XML EBNF form and that disclaimer has been there too long. See for example [61] or anything using ~, "\uHHHH", "A"-"Z" etc. The link to XML1.1 goes to XML Namespaces instead The grammar's <T> syntax is just confusing. The split between lexical tokens and grammar ones is application specific. This form is not XML EBNF either. Lots of clutter like (foo)? (bar)* instead of just foo? bar* (for example [1], [2]) The grammar defines keywords in UPPERCASE. I'd like words that allow all lowercase too (NOT mixed case) as it quickly gets tedious typing that out like that. Early demos have shown this. [19] GRAPH * in the grammar has no meaning I can see (section 8) [30] BaseDecl - BASE has no defined meaning (CVS 1.209) "Perl 5 regular expression" has no citation, could point to some XQ F&O I expect. General Can we have at least one example with $vars please? Please can these examples be updated in CVS - this found a typo for the first WD.
Received on Tuesday, 15 February 2005 23:49:41 UTC