- From: Jeremy Carroll <jjc@hplb.hpl.hp.com>
- Date: Fri, 25 Oct 2002 13:41:51 +0200
- To: "Patrick Stickler" <patrick.stickler@nokia.com>, <w3c-rdfcore-wg@w3.org>
PatrickS: > Please begin vigorous backpeddaling ;-) > Brian: > ... I suggest the WG should prefer an > acceptable alternative Having achieved the near miracle of getting Brian and Patrick to agree on something ... I won't backpedal before today's telecon but thought I should sketch the possibilities: 0) rewording to satisfy Patrick's "the lexical form is not a pair" I stuck the lexical form is a pair in very late in the day, to try and get section 2.4.4 to flow more smoothly. See below. 1) (only very slight change) Explicitly prohibit datatypes other than the two predefined types from using the language identifier in the map. This could be accompanied by (0). 2) back to 'majority' view This is like (1) except that the two predefined datatypes are instead expanded as two alternative types of literal. The language IDs are present and ignored. 3) back to (nearly) 'minority' view In the datatyped literals there is no language identifier. (This seems to be Brian's preference) ==== lexical form explanation: The text I had problems with was: [[[ The datatypes used in RDF have either: + A lexical space consisiting of a set of strings (for example any datatypes from XML Schema). Or: + A lexical space consisiting of a set of pairs: each being a string with a language tag (for example the two predefined types). ]]] and some of the other text in 2.4.4. This is where the pair (string + lang) originated from. I could change this to be either a set of stings (first bullet) or a set of literals (second bullet) [a literal being a triple: including the (constant) datatypeURI]. That way I could drop the pair/pair way of expressing a triple. === I note that the chair is weighting concerns about potential feedback we might receive more highly than the actual feedback we have already received. Jeremy
Received on Friday, 25 October 2002 07:41:59 UTC