- From: Dave Reynolds <der@hplb.hpl.hp.com>
- Date: Tue, 20 Mar 2007 12:36:12 +0000
- To: "Public-Rif-Wg (E-mail)" <public-rif-wg@w3.org>
Here are comments on the frozen draft. Dave ** Issues: i1) Despite the new words, it still comes across as not being a web language because all the examples use the unsorted syntax and there is not a URI in sight. Checking the f2f actions I can't see any relevant ones so I'm not sure how we disposed of this issue. How about putting a caveat before example 3: "The examples of BNF and XML rule syntax given here use the unsorted version of the condition syntax and fail to illustrate the use of URIs for constants. This will be addressed in a future working draft." ? i2) The syntax for primitive sorts is still: _sortname"value" I thought we had agreed to switch that to: "value"^^sortname as a minor way to indicate we are aware of relevant existing approaches. However, I can't find any record of it in the minutes. I repeat my offer to do the edits myself if the only barrier is editing time. ** Editorial (or minor prior comments not yet addressed): e1) The "Abstract" still doesn't reflect the status of the BNF and XML syntax. Suggest replacing: "A human-oriented syntax, an XML syntax, and the semantics of the condition language and of the Horn rule language are given." By "We give an abstract syntax and semantics for the condition language and Horn rule language, we also give example concrete syntax forms (BNF and XML) which are for illustrative purposes only." e2) Section 2.1. Suggest deleting from para 2 the text: "the intent behind this condition language is that it will be shared among the bodies of the rules expressed in future RIF dialects. Possible dialects that have been considered so far include LP, FO, PR and RR; the condition language could also be used to uniformly express integrity contraints and queries" This is the third time in 1.5 pages this same sentiment has been expressed and it seems needlessly repetitive. e3) Section 2.1. One of the XSD type URIs is still incorrect. Replace http://www.w3.org/TR/2004/REC-xmlschema-2-20041028/datatypes.html#long by: http://www.w3.org/2001/XMLSchema#long e4) Section 2.1.1.1 the UML diagram still shows all roles as private (-formula etc) I thought we agreed they were should be shown as public. Happy to leave unchanged for this working draft if changing is a problem. e5) Section 2.1.3. s/BSignature ssociates/BSignature associates/ e6) Section 2.1.3.2. There is a pair of broken font changes around the text: "Long integers also have a short notation, which does not require the" e7) Section 2.1.3.2. I suspect that "constant" in "Note that constant is a symbol from ..." is supposed to refer to "NAME" in the signature examples. ** Comments for future working drafts: f1) The UML diagram in section 2.1.1.1 can't be derived from the asn06. If we continue to use asn06 we will need a notation for fixed length lists and ordering. f2a) The discussion on well-formedness should be expanded in a future working draft. It is unclear what explicit declarations are needed in order to determine the B/ASignature functions. Are we really saying this is a strongly statically typed language? f2b) The description of "Specifying Arrow and Boolean Sorts" does not give an abstract syntax, nor a BNF or XML syntax (just a template example presumably intended to be in the human readable syntax). f2c) What is the scope of signature declarations? f3) Drop the non-sorted examples and replace with examples which conform to the multi-sorted syntax (see i1). f4) Other W3C documents seem to use a referencing style based on an internal document link [ref] and a separate References section that gives the reference for each [ref]. Perhaps we should switch to that style?
Received on Tuesday, 20 March 2007 12:37:25 UTC