- From: Graham Klyne <Graham.Klyne@MIMEsweeper.com>
- Date: Fri, 09 Nov 2001 12:45:31 +0000
- To: Brian McBride <bwm@hplb.hpl.hp.com>
- Cc: Brian McBride <bwm@hplb.hpl.hp.com>, Art Barstow <barstow@w3.org>, w3c-rdfcore-wg@w3.org
I have reviewed this document (http://www.w3.org/2001/08/rdf-test/), and I think it's fine for publication as a working draft. I have the following editorial comments that might be considered for some future revision: Section 1.1:, final sentence, suggest: "... if such test cases are donated to the W3C, it may add them to the overall set of test cases." Section 2.3: (I've not reviewed the actual test cases cited.) I would suggest that test cases with .rdf and .nt files each be presented in a single row of the table, e.g. Issue Test case Status ----- --------- ------ amp-in-url test001.rdf, .nt Approved etc. The "Approved" column content seems like redundant text. Maybe using the date of approval as the anchor text? (This would make it easier to locate the appropriate message(s) in offline email archives.) Section 3.1, 1st para, final sentence. I find the reference to [RDFMT] does not read very clearly. I suggest: "The meaning of these terms is defined in the model theory [RDFMT] being developed as part of the RDF Core WG activity." Section 3.3: I have been in discussion with some I18N folks about the issue of representation on non-ASCII URI characters in CC/PP. Based on those discussions (which are almost but not quite finalized) I offer something the following: [[[ RFC 2396 [URI], section 2.1, discusses the use of non-ASCII characters in URIs, and notes in particular that a URI may be represented as an original character sequence or as a URI character sequence. N-triples uses only US-ASCII characters, so the representation of URI values in N-triples should be as a URI character sequence. This is obtained from an original character sequence value expressed using Unicode characters by applying the rules defined in [Charmod] section Character Encoding in URI References. That is, disallowed characters are represented in UTF-8 and then encoded using the %HH format, where HH is the byte value expressed using hexadecimal notation. ]]] If N-triples is changed to use UTF-8 encoding, then (following XML, see http://www.w3.org/XML/xml-V10-2e-errata#E26) it may be preferred to defer transforming the URIs to use just US-ASCII. (This approach is also indicated for XML schema datatype 'anyURI'.) The second paragraph might then be changed to: [[[ N-triples uses UTF-8 character encoding, so the representation of URI values in N-triples should be as an original character sequence expressed by Unicode characters, coded using UTF-8. When the URI is required in the form of a URI character sequence (e.g. for retrieving a resource referenced by the URI), it is transformed by applying the rules defined in [Charmod] section Character Encoding in URI References. That is, disallowed characters are represented in UTF-8 and then encoded using the %HH format, where HH is the byte value expressed using hexadecimal notation. ]]] Section 3.4: The example is un-printable: the NT text is too wide to fit on a page. Actually, this is a general problem I have with N-triples as defined: the insistence that each statement be on a single line makes the test case text hard to read (and hence hard to review). I think the extra effort required for a machine implementation to read an N-triple statement that splits over 2 or 3 lines would be trivial compared with the added difficulty the single line form creates for human readers. I would like to see additional line breaks permitted between the URIs; e.g. triple ::= subject (ws* eoln)? ws+ predicate (ws* eoln)? ws+ object ws* '.' ws* References: Include reference to RFC 2396 as [URI] ? #g -- ------------------------------------------------------------ Graham Klyne MIMEsweeper Group Strategic Research <http://www.mimesweeper.com> <Graham.Klyne@MIMEsweeper.com> ------------------------------------------------------------
Received on Friday, 9 November 2001 07:49:44 UTC