W3C home > Mailing lists > Public > public-rdf-wg@w3.org > October 2013

Re: Comments on Last-Call Working Draft of RDF 1.1 Concepts and Abstract Syntax

From: Antoine Zimmermann <antoine.zimmermann@emse.fr>
Date: Sun, 20 Oct 2013 10:47:33 +0200
Message-ID: <526398A5.3000905@emse.fr>
To: Sandro Hawke <sandro@w3.org>, Pat Hayes <phayes@ihmc.us>, RDF Working Group <public-rdf-wg@w3.org>
Le 20/10/2013 05:40, Sandro Hawke a écrit :
> Pat Hayes <phayes@ihmc.us> wrote:
>> My 2c on these comments added inline.
>> On Oct 19, 2013, at 4:46 PM, Michael Schneider wrote:
>>> Dear Working Group!
>>> This is my review of the Last-Call Working Draft of the "RDF 1.1
>> Concepts and Abstract Syntax" specification.
>>> Unfortunately, I only learnt about the existence of the LCWDs
>>> from
>> their announcement on the SWIG mailing list as of 3 October, with
>> an - already extended - deadline set to 17 October, which was a
>> very short time and did not give me enough time to complete the
>> review in time. I hope that you will still accept my review. I will
>> send another review for the RDF Semantics specification, which I
>> will hopefully finish till tomorrow. To ease the process for you, I
>> have, after a mail exchange with Sandro Hawke, created my review
>> based on the most-current editor drafts of the two documents:
>>> *
>> <https://dvcs.w3.org/hg/rdf/raw-file/default/rdf-concepts/index.html>
* <https://dvcs.w3.org/hg/rdf/raw-file/default/rdf-mt/index.html>
>>> I'd like to add that I have a high personal and professional
>>> stake in
>> these two documents.
>>> In general, I consider the "Concepts" document pleasing and
>>> don't
>> have any real show-stopping issues to report. Still, there are a
>> few issues that I consider "major", and a couple that I consider
>> "minor".
>>> Major issues: -------------
>>> * § 3.3: "a datatype IRI, being an IRI that determines how the
>> lexical form maps to a literal value." I do not think that it is
>> correct to state that an IRIs would determine how the
>> lexical-to-value mapping works. IRIs generally do only denote some
>> resource, in this case a datatype. It's the datatype specification
>> which determines the mapping, not the IRI that denotes the
>> datatype.
>> Correct. Better wording would be "..being an IRI that identifies a
>> datatype which determines..."
>>> * § 3.1: I believe that the "NOTE" about IRIs, literals and
>>> blank
>> nodes being distinct should be formally specified in some way, and
>> not just "noted".
>> It should be normative, for sure.
>>> * § 5.1: Several of the XML datatypes listed seem to be
>>> incompatible
>> with the definition of a "lexical-to-value mapping" in the
>> beginning of §5. According to the definition, "each member of the
>> lexical space is paired with exactly one value, and is a lexical
>> representation of that value. However, for example the lexical
>> forms of the datatype "xsd:time" do not uniquely denote a single
>> time value, but denote an infinite number of recurrent time values,
>> one per day (for a fixed time zone). Further, for the datatype
>> "xsd:date", it is not clear whether a lexical form denotes a single
>> point on the timeline (e.g. the starting point of a day), or rather
>> a whole interval of values (the whole day). Variants of these
>> problems exist for several of the other listed time-related
>> datatypes.
>> This is all very arguable. The text is however correct in
>> describing what RDF assumes about datatypes. This has some
>> consequences for how the XML datatypes must be interpreted. The
>> value space of xsd:time is the 24-hour daily clock, not timepoints
>> on an infinite timeline. The value space for xsd:date is either
>> points or intervals (if you care, ask the XMLSchema authors), but
>> either way the value is required to be unique for a given lexical
>> form. Variants of these comments apply to the other time-related
>> datatypes.
>> I propose that we make no changes in response to this comment. The
>> statement is accurate, and to get into this much detail is
>> inappropriate in Concepts.
> I agree with you, but I know Michael has thought about this a lot.
> I'd be more comfortable if we first asked him if there's a material
> reason we can't just think about it as you suggest here, Pat.
> Given our current urgency, I think it would be best to ask asap.   I
> guess I'll take a crack at it in the morning (us/eastern) if you
> don't get to it first or say why we shouldn't.
> I might try to look up whether xsd is 1-1 (between values and
> canonical lex rep) first.

XSD 1.1 says that a datatype comprises:
  - a value space,
  - a lexical space,
  - a collection of functions, relations or procedures associated with 
the datatype, *including* a lexical mapping, which is a mapping from the 
lexical space into the value space.

Then it is said that:

For the ·primitive· and ·ordinary· atomic datatypes, the ·lexical 
mapping· is a (total) function on the entire ·lexical space· onto (not 
merely into) the ·value space·: every member of the ·lexical space· maps 
into the ·value space·, and every value is mapped to by some member of 
the ·lexical space·.

xsd:date, xsd:time, etc are primitive datatypes, so all literals of 
these types must map to exactly one value in their value spaces.


> - Sandro
>>> Minor issues: -------------
>>> * § 1.2, par 1, states that the term "resource" "is synonymous
>>> with
>> entity". I did not find the term "entity" being mentioned elsewhere
>> in the document, nor would I say from my experience that it is
>> widely used in the RDF world as a synonym for "resource". So I
>> suggest to remove the cited phrase.
>> Whatever.
>>> * § 1.3, 2nd item: double word: "what what"
>>> * § 3.2, the NOTE: "... that permits a much wider range...": the
>>> word
>> "much" is redundant and should in my opinion not appear in a
>> specification.
>> Agree.
>>> * §4.1: In the introductory sentence, the two datasets D1 and D2
>>> each
>> have only a single named graph NG1 and NG2, respectively. This
>> would be unnessesarily restrictive. Therefore, and from condition 6
>> of the definition, I guess that NG1 and NG2 are really meant to be
>> /sets/ of named graphs for the two datatsets?
>>> * §4.1: Typos in condition 4 of the definition: - comma missing
>>> in the set defining G - the set defining G runs to "1n", which
>>> should be "tn" or "Tn" - the set defining M(G) runs to "M(T(n))":
>>> be careful to use the
>> same
>>> term ("tn" or "Tn" that is used in the set defining G.
>>> * §5.4: "Semantic conditions of RDF MAY recognize other datatype
>> IRIs...". The term "semantic condition" is specific to the RDF
>> Semantics
>>> and as far as I can see does not appear elsewhere in the
>>> Concepts
>> document. I think that the sentence should appear only in the RDF
>> Semantics and should be removed in the Concepts document.
>> I have no opinion on the above comments, but the typos should
>> obviously be fixed.
>> Pat
>>> Best regards, Michael Schneider
>> ------------------------------------------------------------ IHMC
>> (850)434 8903 home 40 South Alcaniz St.            (850)202 4416
>> office Pensacola                            (850)202 4440   fax FL
>> 32502                              (850)291 0667   mobile
>> (preferred) phayes@ihmc.us       http://www.ihmc.us/users/phayes

Antoine Zimmermann
ISCOD / LSTI - Institut Henri Fayol
École Nationale Supérieure des Mines de Saint-Étienne
158 cours Fauriel
42023 Saint-Étienne Cedex 2
Tél:+33(0)4 77 42 66 03
Fax:+33(0)4 77 42 66 66
Received on Sunday, 20 October 2013 08:48:52 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 17:04:33 UTC