RDFa Working Group Teleconference

30 Sep 2010

See also: IRC log


Ivan, +1.612.217.aaaa, ShaneM, +1.441.592.aabb, NathanR, Knud, Steven
Manu, Mark


<trackbot> Date: 30 September 2010

<ivan> yes

<Steven__> trackbot, start telcon

<trackbot> Meeting: RDFa Working Group Teleconference

<trackbot> Date: 30 September 2010

<ivan> scribe: Knud

<ivan> scribenick: Knud

Ivan: welcome to Nathan, we know him from the mailing list as an active contributor
...: Nathan joins as an invited expert

Issue 40


"Empty element should not create literals"

Nathan: I think empty triples should be generated

<Steven__> I agree
...: there might be situations where you _want_ an empty string, like not having a middle name

<Zakim> ShaneM, you wanted to agree that the empty triples should be generated
...: this would be just moving the problem

ivan: it would still possible to generate an empty string with @content

shane: agree with Nathan - RDFa is just another serialisation, why should there be a special case for it?

steven: I also agree. It would be wrong to not create empty literals

<ivan> PROPOSED: empty elements should generate empty literals (Issue-40)

<Steven__> +1

<ivan> +1

<NathanR> +1

<ivan> PROPOSED: not to accept the change request in Issue-40

<Steven__> +1

<ivan> +1

<NathanR> +1

<ivan> RESOLVED: not to accept the change request in Issue-40

ivan: not formally closed, because too many active people missing



"Consider relationship between LinkTypes in RDFa and the IETF LinkType registry"

shane: there is work having to do with http headers. creating a registry for values of @rel
...: however, what we do is slightly different than usually.
... our values of @rel are terms, and is extenable
... hostlanguages can define their own terms
... hostlanguages could worry about linking, registering, etc. those terms with the IETF

ivan: however, we plan to define a default profile (of terms) for XHTML

shane: exactly. We define terms, can look for correlations, but we should not have a dependency to the IETF

<ivan> PROPOSED: RDFa does not introduce an explicit dependency on the LinkTypes registry (Issue-35)

<ShaneM> +1


<ivan> +1

<Steven__> +1

<NathanR> +1

knud: should we explain that in the specs?

<ivan> RESOLVED: RDFa does not introduce an explicit dependency on the LinkTypes registry (Issue-35)

ivan: that's probably not necessary
...: it was Manu who raised this initially

shane: on a related note - this has to do with case-sensitivity of terms
...: in the IETF, they are case-sensitive
... in RDFa core, they are case-insensitive

URI strings in literal

"Any plain literal that does not have a (possibly empty) @datatype that matches the regex for a URL MUST be treated as an IRI object in RDFa."


ivan: many people do not make a difference between URIs appearing as a literal, or as an attribute value of special attributes

steven: can we not apply a datatype to make a literal a URI?

ivan: this is not possible at the moment
...: there is currently no corresponding xsd datatype
... and we cannot use "anyURI"
... there are legitimate cases where I want to have a literal, a string, which conforms to the URI spec
... anyway, this would not help people like FB, because they won't ask their developers to use this datatype

nathan: we should go with the common case. basically agree with the proposal

steven: this means that, every time we have a non-datatype string, we have to check if the string is an acceptable URI

<ShaneM> fubar == xsd:anyURI
...: problem: "fubar" is an acceptable relative URI

ivan: we may have to restrict to absolute URIs

<NathanR> ../foo & iso: & "" (0-length-string) all == xsd:anyURI

<NathanR> relative0uri*

<Zakim> ShaneM, you wanted to ask about datatypes

steven: there would still be lots of things that might be parsed as a URI. lots of string with a colon in it, for example

shane: I also see this problem. Also, we will probably need to allow relative URIs. e.g., resource="#foo", resource="picture.jpg"

<Steven__> <span property=my:datatype>xsd:integer</span>ยจ

<Steven__> xsd:integer would be seen as a URI

<NathanR> <span property="foaf:name">nathan</span> <--- would be a valid uri ref

<ivan> <meta property="og:sfdssf" content="http://....." />

<Steven__> Not for an abs URI

<Steven__> xsd:integer

nathan: I also agree this is a problem. maybe less of a problem with absolute URIs

<ShaneM> mailto:shane

steven: but even absolute URIs (that are meant to be URIs) are very hard to detect

<ShaneM> urn:...
...: isbn:43290489fj is an absolute URI

ivan: so, we might have to agree on a family of URI schemes that have to be matched
...: is there an official registry of schemes?

shane: yes, there is the IETF scheme

<NathanR> skype:

steven: but not everybody uses this registry. Like Apple.

ivan: people using unregistered URI schemes could still use the "normal" way of specifying URIs, via @resource

<scribe> ACTION: shane to find reference to the IETF registry [recorded in http://www.w3.org/2010/09/30-rdfa-minutes.html#action01]

<trackbot> Created ACTION-38 - Find reference to the IETF registry [on Shane McCarron - due 2010-10-07].

<ShaneM> Information on registring URI schemes: http://tools.ietf.org/html/rfc4395

<ShaneM> ACTION-38: closed

<trackbot> ACTION-38 Find reference to the IETF registry notes added

<trackbot> If you meant to close ACTION-38, please use 'close ACTION-38'

<ShaneM> close ACTION-38

<trackbot> ACTION-38 Find reference to the IETF registry closed

steven: I would really like to find a general solution, not a hacky one

shane: we can still define a term to instruct the RDFa processor to interprete something as a URI

<ivan> <meta property="og:asfasdf" datatype="rdfa:resource" content="http://sfsd" />

<ShaneM> <link property='og:lala' datatype='xv:thisIsAURI'>http://example.com</link>

<ivan> <link rel="og:lala" resource="http://example.com"/>

ivan: this is probably just as difficult for the FB-esque community

nathan: it's also not backwards compatible

shane: if the OGP had an RDFa profile, and if the profile could define that certain properties have their values interpreted as a URI?
...: so we could invoke "follow-your-nose". the definition of og:fdhskfj could specify that it's value is an rdf:resource

ivan: this is impossible to define in the current framework

nathan: long-term solution: a way for properties to define their ranges as needed here

Summary of Action Items

[NEW] ACTION: shane to find reference to the IETF registry [recorded in http://www.w3.org/2010/09/30-rdfa-minutes.html#action01]
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.135 (CVS log)
$Date: 2010/09/30 14:59:28 $

Scribe.perl diagnostic output

[Delete this section before finalizing the minutes.]
This is scribe.perl Revision: 1.135  of Date: 2009/03/02 03:52:20  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: RRSAgent_Text_Format (score 1.00)

Found Scribe: Knud
Found ScribeNick: Knud
Default Present: Ivan, +1.612.217.aaaa, ShaneM, +1.441.592.aabb, NathanR, Knud, Steven
Present: Ivan +1.612.217.aaaa ShaneM +1.441.592.aabb NathanR Knud Steven
Regrets: Manu Mark
Found Date: 30 Sep 2010
Guessing minutes URL: http://www.w3.org/2010/09/30-rdfa-minutes.html
People with action items: shane

WARNING: Input appears to use implicit continuation lines.
You may need the "-implicitContinuations" option.

[End of scribe.perl diagnostic output]