- From: Andy Seaborne <andy.seaborne@epimorphics.com>
- Date: Tue, 12 Jul 2011 11:43:28 +0100
- To: public-rdf-dawg@w3.org
On 11/07/11 15:55, Polleres, Axel wrote: > Here goes my review of the CVS doc. > > > 1) In the abstract: > > "@@Add ASK as well? One line - true or false" > > I suggest to: > > a) s/This document describes their use for expressing SPARQL query results from SELECT queries./ > This document describes their use for expressing SPARQL query results from SELECT and ASK queries./ > > b) Add a new very short section before Section 3 as new Section 3 as follows: Are you proposing that the format does cover ASK results? Or listing the changes that would be needed without advocacy? > 3 Encoding results of ASK Queries as CSV/TSV > > The result of ASK queries as a CSV or TSV is either the single string > "true" or "false" followed by the end-of-line character. The following > sections explain how to encode the results table for a SELECT query in CSV > (Section 4) and TSV (Section 5). Steve and Greg have argued for requiring the header line in CSV files. The TSV format requires a header line. An ASK query of one word isn't compliant with th > > > and renaming current Section 3 and 4 to > > 4 Encoding results tables as CSV - Comma Separated values > > 5 Encoding results tables as TSV - Tab Separated values > > > c) A small side remark, I found a small problem in the "SPARQL Query Results XML Format" doc (http://www.w3.org/TR/rdf-sparql-XMLres/)... Is there an errata? Please add: http://lists.w3.org/Archives/Public/public-rdf-dawg/2010OctDec/0215.html > 2) In section 3.1: > > s/The SPARQL CSV Results Format should use of a header row./ > The SPARQL CSV Results Format should use a header row./ Fixed. > BTW, should we add an explicit remark, under which circumstances it might make sense to ommit the header row? E.g.: > "Note that, in general, i.e. when more than one output variable appear in the SELECT clause, removing the > header row will result in ambiguous order of columns in the result table. However, for instance when only a single output > variable is selected it may be desirable to omit the header row." See above. > 3) in Section 3.2 > > "Fields do not need to be quoted if they do not contain" > > Do I read this correctly that quoted and unquoted can be mixed arbitrarily, i.e. > > x > foo > bar > > x > "foo" > "bar" > > x > "foo" > bar > > are all valid serializations of the same result set ?x/"foo", ?x/"bar", yes? Yes. Standard CSV. > > 4) in section 3.3: > > It would be nice, if the final example for CSV would also contain some quoted strings. Note it already says: @@More complete example > > 5) in section 4: > > Is there a particular rationale for disallowing triple quoted forms from Turtle, and if so, can this rationale be added? A raw newline or raw tab would break the TSV format. By insisting on single quoted forms only, that can't happen. I've added something although generally I do not hold with explaining every design decision in a spec. A spec is a definition; justification leads to excessive text and can't not be complete. > > 6) in section 4.2 > > Why do we reference [RDF-SPARQL-QUERY] here instead of [SPARQL11-QUERY]? References are managed by the general W3C mechanisms and you can't ref working drafts. Has to be fixed later :-( > > 7) in section 4.2 > > s/Use of the samelabel indicates the same blank node within the results but has no significance outside the results./ > Use of the same label indicates the same blank node within the results but has no significance outside the results./ Done. > > 8) > > It would be nice, if the final example for TSV would also contain some \t, \n, or \r Note it already says: @@More complete example Providing those examples would be very helpful! ----- Can we please publish this as a WD-NOTE now? Andy
Received on Tuesday, 12 July 2011 10:43:59 UTC