RIF Telecon 12-May-2009 -- 12 May 2009

Admin

<ChrisW> no kidding

<ChrisW> Scribe: AdrianP

<ChrisW> http://lists.w3.org/Archives/Public/public-rif-wg/2009May/att-0062/050509-rif-minutes.html

ChrisW: minutes from last week

<ChrisW> PROPOSED: accept last weeks minutes

<ChrisW> RESOLVED: accept last weeks minutes

Liason

Action Review

<ChrisW> action-808: closed

<trackbot> ACTION-808 make "1:1" for rdf:lists be at-risk notes added

<trackbot> If you meant to close ACTION-808, please use 'close ACTION-808'

<ChrisW> close action-808

<trackbot> ACTION-808 make "1:1" for rdf:lists be at-risk closed

close action-779

<trackbot> ACTION-779 Review SWC closed

close action-778

<trackbot> ACTION-778 Review FLD closed

close action-776

<trackbot> ACTION-776 Review bld closed

close action-775

<trackbot> ACTION-775 Review dtb closed

close action-767

<trackbot> ACTION-767 Review PRD closed

close action-762

<trackbot> ACTION-762 Amends the XML schema for Core. closed

close action-759

<trackbot> ACTION-759 Add the syntax and semantics of lists to BLD closed

close action-760

<trackbot> ACTION-760 Add the list builtins in DTB closed

DTB

PRD

ChrisW: Axel's review? not done, yet
... no review from Changai

Harold: editorial comments in my review
... specifity for conflict resolution not mentioned
... document should mention why it is not listed
... my old review comments are still open

csma: working on them
... conflict resolution strategy we agreed to have the three principle ones - common to most PR engines

cmsa: some also use specifity - but it is used differently in the different engines

Harold: could be explained in one abstract and why specifity is excluded

csma: then we would need to explain why we omitted others

Harold: could refer to a paper
... many share specificity

csma: will add a sentence about specificity not well defined

<ChrisW> csma: we chose the conflict resolution strategy shared by the most PR engines, NOT the one that "works best"

<ChrisW> ...not sure it makes sense to talk about which ones we didn't include, rather justify the reason for the one we retained

<ChrisW> ACTION: csma to respond to harolds comments [recorded in http://www.w3.org/2009/05/12-rif-minutes.html#action01]

<trackbot> Created ACTION-809 - Respond to harolds comments [on Christian de Sainte Marie - due 2009-05-19].

Harold: fine with going to LC for PRD

Axel: DTB last call actions I'm done except for refine all informal builtin definitions

<AxelPolleres> my reviews can be done until end of the week.

Jos: should be able to do the PRD reviews by end of this week

SWC

<josb> Michael did post regrets: http://www.w3.org/2005/rules/wiki/TeleconRegrets

<ChrisW> thanks, jos

BLD

<AxelPolleres> I promise to have my SWC+BLD reviews by the end of the week.

<ChrisW> ^I promise^I hereby promise^

csma: abstract syntax (mathematical syntax) should be normative

<ChrisW> sandro, looks like you are echoing

<Harold> "Such generalized open lists, similar to Lisp's s-expressions, make it unnecessary to restrict variable values in the tail to lists."

<Harold> (http://www.w3.org/2005/rules/wiki/BLD#Terms)

csma: BLD does not qualify elements and attributes; in PRD we sometimes qualify attributes

Harold: should not be qualified; makes the syntax ugly and verbouse

Sandro: we resolved that at the first LC

csma: then I will allign PRD

Harold: do not qualify elements, either

<sandro> sandro: In all RIF dialects, elements must be qualified -- although of course a default XMLNS can be used -- and attributes should not have a namespace. We agreed to this long time ago.

Harold: uses the default namespace

<sandro> root element has xmlns="http://www.w3.org/2007/rif#" right?

<Harold> We use:

<Harold> <Document

<Harold> xmlns="http://www.w3.org/2007/rif#"

<Harold> xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"

<Harold> xmlns:xs="http://www.w3.org/2001/XMLSchema#">

<Harold> . . .

<sandro> Yes, so that means the element are qualified, using the default namespace mechanism.

Jos: question to csma; difference between abstract syntax and presentation syntax

<Harold> Forall ?item ?deliverydate ?scheduledate ?diffduration ?diffdays (

csma: is the mathematical definition of the presentation syntax normative or not?

<Harold> )

csma: it says the presentation is normative

<ChrisW> "he presentation syntax is normative, but is not intended to be a concrete syntax for RIF-BLD. It is defined in "mathematical English," a special form of English for communicating mathematical definitions, examples, etc. The presentation syntax deliberately leaves out details such as the delimiters of the various syntactic components, escape symbols, parenthesizing, precedence of operators, and the like."

ChrisW: looks like some inconsistency in the description

<csma> Forall ?item ?deliverydate ?scheduledate ?diffduration ?diffdays (

is writting it like this normative?

csma: is writting it like this normative?

<Harold> Universal rule: If f is a rule implication and ?V1, ..., ?Vn, n>0, are distinct variables then Forall ?V1 ... ?Vn(f) is a formula, called a universal rule.

csma: object to this being normative

<Harold> (http://www.w3.org/2005/rules/wiki/BLD#Appendix:_XML_Schema_for_RIF-BLD)

<Harold> The use of "?" in ?V1, ..., ?Vn etc. is mathematical English: normative.

csma: abstract syntax for me is normative, not the presentation syntax which is used to write the rules for presentation
... not all PRD want PRD syntax normative

<sandro> (I'm confused.... and worried about how we're doing on time for this meeting.)

Jos: it is not a concrete syntax
... you need some mathematical symbols to refer to them

<sandro> Chris: how about if we say folks can use their own symbol for 'forall', etc?

csma: I object to being the abstract syntax being called presentation syntax
... this will not be understood outside of this RIF group

<sandro> (I agree it's extremely confusing and problematic)

<sandro> csma: In PRD I use the term "abstract syntax" for a normative intermedia syntax, and then provide a non-normative presentaiton syntax.

csma: confusing abstract syntax with presentation syntax - should be clearly rephrased

Harold: word abstract does not even occur in BLD
... we did have an abstract syntax but we abandoned it 2 yrs ago
... introduced mathematical syntax instead

ChrisW: would you be fine to call it abstract syntax

Harold: in OWL they use the term abstract syntax

cmsa: like in UML where they use it to define the abstract syntax

csma: mathematical English define abstract syntax for RIF
... confused then with the non-normative presentation syntax

<sandro> What if we are clear about having MULTIPLE presentation syntaxes, some more formally specified than others.

Harold: we have mathematical English (normative) and EBNG syntax (not normative)
... we need the normative English as normative to define the semantics

<Harold> The superconcept for both mathematical English and EBNF syntax is presentation syntax.

Hassan: agree with Christian - we do not define a normative abstract syntax

<Harold> So the presentation syntax divides into a normative and a non-normative version.

Sandro: add a very clear sentence which explains that only the XML syntax is normative

csma: replace "presentation syntax" is normative with "presentation syntax" is not normative

<Harold> The continuation of the para Chris showed is "Since RIF is an interchange format, it uses XML as its concrete syntax and RIF-BLD conformance is described in terms of semantics-preserving transformations."

Sandro: the reason was that we used the mathematical English

<josb> plus, the XML syntax is defined through a mapping from the presentation syntax

<Harold> (http://www.w3.org/2005/rules/wiki/BLD#Direct_Specification_of_RIF-BLD_Presentation_Syntax)

<ChrisW> ACTION: chris to draft a paragraph describing the status of the presentation syntax [recorded in http://www.w3.org/2009/05/12-rif-minutes.html#action02]

<trackbot> Created ACTION-810 - Draft a paragraph describing the status of the presentation syntax [on Christopher Welty - due 2009-05-19].

DTB

<AxelPolleres> http://www.w3.org/2005/rules/wiki/DTB

<sandro> 'We are NOT specifying any of the various presentation syntaxes used in these documents. Implementors are free to implement whatever RIF presentation syntaxes they like; users should NOT expect any interoperability using these presentation systaxes. For interoperability, either use XML or some other Rule syntax that *is* specified for interoperability.

<sandro> (that's my suggested text, Chris. Maybe a useful starting point.)

Axel: I added all data types, casting functions for datatypes

<AxelPolleres> http://www.w3.org/2005/rules/wiki/DTB#pred:literal-not-identical

<AxelPolleres> http://www.w3.org/2005/rules/wiki/DTB#Datatype_Conversion_and_Casting

Axel: 3.4.1. RIF does not require white space normalization
... this is not defined in XPath
... xs:anyURI is a cast from a subtype of string

Jos: not in Xpath 1.0

Axel: then it is fine
... XPath leaves open extent to which an implementation validates the lexical form of xs:anyURI to the implementation.
... but RIF requires all lexical forms of xs:anyURI appearing as constants in the xs:string symbol space to be castable to xs:anyURI

ChrisW: why?

Axel: cast function from String to AnyURI are implementation dependent in XPath/XQuery
... Xpath does not require to parse/implement it
... Editors note
... What to do about: "In casting to a date or time value, if the value is too large or too small to be represented by the implementation, error [err:FODT0001] is raised."
... Editor note "Casting from xs:float or xs:double to xs:decimal or its subtypes may raise implementation dependent errors [ERRFOCA0001] or [ERRFOCA0003]. Unclear how we avoid implementation dependance."
... was not clear to me how to avoid implementation dependency
... Editor note "Note that Section 17.1 of [XPath-Functions] says that for datatypes that do not have a canonical lexical representation defined an implementation dependent canonical representation may be used. We probably do not want that. This remark probably also applies to subtypes of xs:string."
... not sure what to do here

Jos: any examples?

ChrisW: what about datetime and timezone? Do they map to the same timepoint in the value space

<ChrisW> ACTION: axel to check that all RIF datatypes have a canonical representation [recorded in http://www.w3.org/2009/05/12-rif-minutes.html#action03]

<trackbot> Created ACTION-811 - Check that all RIF datatypes have a canonical representation [on Axel Polleres - due 2009-05-19].

Axel: cast functions only affect XML Schema datatypes
... no longer cast rdf:literal
... Editors note "Casting from rdf:text and rdf:XMLLiteral to xs:string are still under discussion. "?

<josb> I think only NOTATION and QName don't have a canonical representation

<ChrisW> proposed: extend meeting for 15 mins

<ChrisW> resolved: extend meeting for 15 mins

Axel: would leave it as it without casting of rdf:...

Sandro: why are we not casting from rdf:literal to string?

Axel: we would need to define it for RIF

<sandro> Jos: Every XMLLiteral has a unique lexical representation, which could be seen as the string form

<sandro> +1 yes, use that lexrep as the string form of XMLLiteral.

Axel: will see if I can define it in the general definition of cast functiosn

<sandro> (Yes, just use the lexical space.)

<ChrisW> ACTION: axel to define cast from xmlliteral to streing [recorded in http://www.w3.org/2009/05/12-rif-minutes.html#action04]

<trackbot> Created ACTION-812 - Define cast from xmlliteral to streing [on Axel Polleres - due 2009-05-19].

Axel: results where error can occur leave it undefined
... redefined numeric functions accordingly
... section 3.7. functions onf strings added
... function & predicates on datetime
... 3.8.1.14

<AxelPolleres> http://www.w3.org/TR/xpath-functions/#func-years-from-duration

Axel: 3.8.1.13 func:years-from-duration

<AxelPolleres> http://www.w3.org/TR/xpath-functions/#func-year-from-date

<josb> it's awkward that the former accepts xs:string as argument, but the latter does not

Jos: Let's ask the XPath working group if their definition is intended like it is

<ChrisW> ACTION: axel to find clarification on year-from-duration from XPATH wg [recorded in http://www.w3.org/2009/05/12-rif-minutes.html#action05]

<trackbot> Created ACTION-813 - Find clarification on year-from-duration from XPATH wg [on Axel Polleres - due 2009-05-19].

Axel: editors note "This and the following functions assume an implicit timezone provided by the dynamic context (See Section C.2 Dynamic Context Components.) to be present as part of the value, if not explicit timezone is given. How shall we proceed for RIF here? The current solution with not assuming any impliciet time zone seems unsatisfactory."

<StellaMitchell> yes

<ChrisW> proposed: continue for 15 more mins to finish dtb

<ChrisW> resolved: continue for 15 more mins to finish dtb

<AxelPolleres> http://www.w3.org/TR/xpath-functions/#func-subtract-dateTimes

Jos: Think you already implemented a good solution

Axel: timezones need to be explicility given
... Editor's Note: Is there any casting or promotion implicit here and in the following functions? That would affect the domain.

ChrisW: seems like a bug in XPath

<AxelPolleres> op:divide-yearMonthDuration(xs:yearMonthDuration("P2Y11M"), 1.5)

<josb> http://www.w3.org/XML/2007/qt-errata/xpath-functions-errata.html#E33

Jos: did you like at the errata document

Axel: will ask the XPath working group about this, too
... predicates are not finished yet
... some old editors note left there

<AxelPolleres> 3.8.2.25

Axel: Editor's Note: The introduction of less-than-or-eaual and greater-than-or-equal predicates for dayTimeDuration and yearMonthDuration still needs a WG resolution.

I think we wanted them

Axel: Editor's Note: Predicates for rdf:XMLLiteral such as at least comparison predicates (equals, not-equals) are still under discussion in the working group.
... leave 1 pred:XMLLiteral-equal
... Editor's Note: Issues which are still open in the rdf:text specification might imply future changes on the functions and predicates defined here. For instance rtfn:compare and rtfn:length are curently marked AT RISK. We could subsume these functions under a single func:compare and func:compare function, instead of defining separate functions for xs:string and rdf:text, or drop them alltogether for redundancy. Moreover, references and links to the [RDF-TEX
... still at risk

Sandro: will deal with that later
... onyl informal description right now

3.11.3 Predicates on RIF Lists

ChrisW: semantics are defined using formal mappings

Jos: right we need a formal mapping to complete the spec

ChrisW: these functions need to be aligned with the model-theoretic semantics of the other functions in DTB

Sandro: could we add a sentence about the general semantics of all list functions at the beginning of this list section

ChrisW: add a subbullet about formal mapping to each function

Sandro: think it is not needed, but do not object

<ChrisW> ACTION: josb to provide formal mapping for list preds & funs [recorded in http://www.w3.org/2009/05/12-rif-minutes.html#action06]

<trackbot> Created ACTION-814 - Provide formal mapping for list preds & funs [on Jos de Bruijn - due 2009-05-19].

<sandro> </rant> :-)

ChrisW: only thing remaining for DTB are the two questiosn to the working group

- DRAFT -

RIF Telecon 12-May-2009

12 May 2009

Attendees

Contents

Admin

Liason

Action Review

DTB

PRD

SWC

BLD

DTB

Summary of Action Items

Scribe.perl diagnostic output