Re: RDF Web Apps WG telecon minutes for 2011-10-13 (ISSUE-111)

I saw in the minutes that I was directly solicited on the @rev/@inlist issue...

- Yes, the pyRdfa version running on my machine does implement this. Implementation was not very complicated though, on the processing step level, there is a slight complication as for what information one has to store for hanging rels. Nothing that could not be handled, though 

(The one problem here is that we are adding small things to the processing rules. Nothing major in any step, but makes the whole processing more and more complicated. We should be careful with that...)

- My pros and cons on this are:

#1. Pro: it does seem a bit odd that @rev is excluded from a processing that is defined for @rel.
#2. Pro: Niklas has a clear use case
#3. Con: whilst the @rel/@inlist and the @property/@inlist combination works well, ie, it is possible to create lists with both literal and URI resource members (which is great, that is something RDF/XML cannot do for Literals and have been a huge issue for many!), @rev/@inlist does not have any counterpart with @property.

My gut feeling says that #3 may lead to lots of errors, ie, users that may expect the same flexibility in terms of resources and literals on list for @rev and for @rel. That is my main issue with this proposal as of now... 

Bottom line: I am still undecided:-(

Ivan

P.S. A complete aside. In order to lear mercurial I have installed mercurial on my own machine, and I use it to experiment with such features like @rev/@inlist. I must say the combination of mercurial and the possibility to roll back quickly through branches is really great. I wish I had done that earlier:-)





On Oct 13, 2011, at 18:54 , Manu Sporny wrote:

> Thanks to Stéphane for scribing! The RDF Web Apps WG telecon minutes for October 13th, 2011 are now available here:
> 
> http://www.w3.org/2010/02/rdfa/meetings/2011-10-13
> 
> If you would like to read minutes from this or previous meetings, the
> public record of all RDF Web Apps WG telecons is available here:
> 
> http://www.w3.org/2010/02/rdfa/wiki/Meetings
> 
> Full text log follows:
> 
> ----------------------------------------------------------------------------
> Agenda
>  http://lists.w3.org/Archives/Public/public-rdfa-wg/2011Oct/0020.html
> Seen
>  Gregg Kellogg, Henri Bergius, Knud Möller, Manu Sporny,
>  Niklas Lindström, Sebastian Germesin, Shane McCarron,
>  Stéphane Corlosquet
> Guests
>  Niklas Lindström, Henri Bergius
> Scribe
>  Stéphane Corlosquet
> Resolutions
>  1. Do not add functionality to @typeof where if the first token is an
>     IRI, that sets the default vocabulary for processing.
> Topics
>  1. ISSUE-111: Determine behavior when @inlist and @rev are used
>     together
>  2. ISSUE-108: Refine/deprecate Link relations
>  3. First IRI in @typeof determines the vocabulary used
>  4. Schema.org supports RDFa in Rich Snippet Tool
> 
> (Scribe set to Stéphane Corlosquet)
> (No events recorded for 11 minutes)
> 
> 1. ISSUE-111: Determine behavior when @inlist and @rev are used together
> 
> Manu Sporny: has anyone implemented @inlist or @rev in their processor?
> 
> Gregg Kellogg: I didn't include it yet... @rev, when used with @inlist, has no effect in the current spec.
> 
> Niklas Lindström: the idea was that @rev used with @inlist might be useful to make a link back to the list itself
> 
> ... it might introduce problems, though - you can only make reverse links to resources, not literals.
> 
> Manu Sporny: Ivan has implemented something, but there are issues with how to interpret that with other RDFa attributes?
> 
> Gregg Kellogg: @rev not doing anything with @inlist is not a proposal, it's the existing behavior. question is: do we want to keep it that way?
> 
> ... if we don't have an advocate for @inlist we can't make much progress today.
> 
> Manu Sporny: Ivan would be that advocate, though he said he didn't really like what it did.
> 
> Manu Sporny: The danger is we have an attribute which does not do what we expect it to do. @rev works in every other case in RDFa, so why not with @inlist?
> 
> Manu Sporny: one can argue that @inlist it's an advanced feature, and should only be used for advanced use cases
> 
> Niklas Lindström: I agree. haven't seen any use case for using the list as a subject - no use cases for @rev and @inlist.
> 
> ... if there were a real use case for it, we could supply the list as subject
> 
> ... we should postpone it until a real use case is found
> 
> Manu Sporny: it seems people feel uncomfortable with @inlist and @rev used together
> 
> Manu Sporny: anyone disagree?
> 
> Manu Sporny: We should not make a decision on this call today - let's wait until Ivan and Toby can give their input.
> 
> Sebastian Germesin: ok, for waiting
> Shane McCarron: You could send out a proposal to take effect in 7 days if nobody objects.
> 
> Shane McCarron: we should solicit Jeni's opinion
> 
> ACTION: Manu to write proposal to not support @rev/@inlist to not support for 7 days cc Jeni Tennison.
> 
> Trackbot IRC Bot: Created ACTION-98 - Write proposal to not support @rev/@inlist to not support for 7 days cc Jeni Tennison. [on Manu Sporny - due 2011-10-20].
> 2. ISSUE-108: Refine/deprecate Link relations
> 
> Manu Sporny: http://www.w3.org/2010/02/rdfa/track/issues/108
> Manu Sporny: one approach we can take it accept only the values specified by HTML5
> 
> Manu Sporny: another approach is that the initial list could be the same as the XHTML values we have now, and wait until the new values are standardized to add them
> 
> Manu Sporny: another approach is to remove stylesheet, since those triples are not really useful and people don't like them.
> 
> Gregg Kellogg: The Microdata spec removed alternate and stylesheet and replaced them with ALTERNATE-STYLESHEET
> Manu Sporny: maybe we should also remove alternate as it's not used the way we would expect it in RDF
> 
> Shane McCarron: alternate has valid use cases, linking to RSS documents, for instance.
> 
> Gregg Kellogg: Microdata used to do this in an earlier version of the spec: ALTERNATE-STYLESHEET
> 
> Shane McCarron: we don't care about the RDF generated by stylesheet
> 
> Manu Sporny: We could hold off on generating alternate or stylesheet triples until the processer would have processed all values in @rel to decide what value should be generated... alternate, stylesheet, or alternate-stylesheet
> 
> Manu Sporny: we want to generate useful triples for people on the semantic web. stylesheet and alternative are usually not useful.
> 
> ... people who need these would not use RDF for the purpose of alternative and stylesheet, they would use a different type of application framework.
> 
> Niklas Lindström: .. I agree: stylesheet without content is reasonably quite useless
> Niklas Lindström: .. i.e. without the html
> Manu Sporny: Shane made a point not to remove alternate as it can link to alternate representation (RSS)
> 
> Niklas Lindström: we could also say that if @rel contains stylesheet we would ignore the @rel
> 
> Manu Sporny: is anybody depending on @rel alternate? I would be surprised if there was any, but have no data to back that up.
> 
> Niklas Lindström: I think there is a potential for it, I would probably use dc:hasFormat for that use case though
> 
> Manu Sporny: RDFa has been around since 2008, if today there is no use case today, there probably won't be - we should remove it.
> 
> Manu Sporny: in the vast majority of the use case, it generates the wrong triples anyway. HTML->alternate->CSS is flat out wrong.
> 
> Shane McCarron: on ALTERNATE-STYLESHEET: collection of value in the vocab document. if the HTML WG is randomly introducing new terms with semantics, there is potential for conflicts. e.g. role. is anyone worried about that?
> 
> Manu Sporny: we would raise an issue in the HTML WG if this were to happen
> 
> ... that's all we can do, depend on W3C Process to prevent screw-ups like that.
> 
> Manu Sporny: the only thing we're talking about is the removal of stylesheet and alternate
> 
> Manu Sporny: anyone disagree? or want to add something?
> 
> ACTION: Shane to respond to mailing list with pointer to discussion today about alternate/stylesheet
> 
> Trackbot IRC Bot: Created ACTION-99 - Respond to mailing list with pointer to discussion today about alternate/stylesheet [on Shane McCarron - due 2011-10-20].
> 3. First IRI in @typeof determines the vocabulary used
> 
> Manu Sporny: http://lists.w3.org/Archives/Public/public-rdfa-wg/2011Oct/0016.html
> Stéphane Corlosquet: I'm not the first one to propose this - some background. [ Scribe Assist by Manu Sporny ]
> Stéphane Corlosquet: @vocab was introduced to make it easier to author content - to avoid using prefixes/CURIEs - so that was good, makes the markup easier. [ Scribe Assist by Manu Sporny ]
> Stéphane Corlosquet: I think a burden remains - people that come from Microdata - there is a new concept that they have to understand - vocabularies. It's one more thing that people have to learn. [ Scribe Assist by Manu Sporny ]
> Stéphane Corlosquet: Microdata solves this problem by just saying that the itemtype is the base vocabulary (implicitly). [ Scribe Assist by Manu Sporny ]
> Stéphane Corlosquet: I was wondering if RDFa could adapt the same approach as Microdata - give the option of extracting the vocabulary from the first item in @typeof [ Scribe Assist by Manu Sporny ]
> Stéphane Corlosquet: The only difference is that @vocab becomes optional - you can remove vocab and put the full URI of the type in @typeof... processors would infer the vocabulary from the first IRI in @typeof. [ Scribe Assist by Manu Sporny ]
> Stéphane Corlosquet: We could allow for CURIEs in @typeof if the prefix is described somewhere in the document. [ Scribe Assist by Manu Sporny ]
> Stéphane Corlosquet: So you could have something like typeof="skos:Concept" and then the skos URL would become the default @vocab. [ Scribe Assist by Manu Sporny ]
> Gregg Kellogg: I was uncomfortable with this feature in Microdata... in RDFa it seems dangerous - it doesn't solve all of the problems you want. It's common to use properties from other vocabs like Dublin Core with schema.org - it seems unnatural for RDFa to be able to do that. [ Scribe Assist by Manu Sporny ]
> Gregg Kellogg: RDFa just has other mechanisms to make this easier. [ Scribe Assist by Manu Sporny ]
> Stéphane Corlosquet: I'm not saying RDFa should drop everything else, just add this to make markup easier for those coming from a Microdata world. [ Scribe Assist by Manu Sporny ]
> Stéphane Corlosquet: This is just a shortcut to not use @vocab. [ Scribe Assist by Manu Sporny ]
> Stéphane Corlosquet: The idea is that people that come from Microdata could make the change easily. It's just a search/replace. [ Scribe Assist by Manu Sporny ]
> Niklas Lindström: I think I agree with Toby - who proposed this initially, and has since come around to not supporting this feature. There are problems with this approach - like what happens when you do chaining - there are technical issues. [ Scribe Assist by Manu Sporny ]
> Niklas Lindström: This might be a bit too magical, rather than the explicit use of @vocab. [ Scribe Assist by Manu Sporny ]
> Stéphane Corlosquet: I don't know if there is a problem with @vocab. re: Chaining - I believe that Microdata use cases will not use that. Raising this proposal from Microdata perspective. This is so that people can convert Microdata to RDFa easily. This feature will only be used with simple markup. [ Scribe Assist by Manu Sporny ]
> Manu Sporny: aq+
> Niklas Lindström: I kind of see your point - but there are issues. If I use full IRIs with this mechanism, that would also set the @vocab. @vocab specifies the namespace - but so does @typeof... that may be confusing - that's what Microdata kinda does. [ Scribe Assist by Manu Sporny ]
> Manu Sporny: How do we know how to process the @typeof IRI in a deterministic way?
> Manu Sporny: Is conversion from Microdata to RDFa a use case we care about?
> Niklas Lindström: How is the vocabulary determined? [ Scribe Assist by Manu Sporny ]
> Gregg Kellogg: Everything after the slash or hash - which is different from Hixie's spec. [ Scribe Assist by Manu Sporny ]
> Niklas Lindström: Yes, this is aligned with schema.org - that's what you'd expect. [ Scribe Assist by Manu Sporny ]
> Gregg Kellogg: and this is why we did it that way. [ Scribe Assist by Manu Sporny ]
> Niklas Lindström: Microdata dropped any conversion to RDF... where are we on that? [ Scribe Assist by Manu Sporny ]
> Gregg Kellogg: https://dvcs.w3.org/hg/htmldata/raw-file/24af1cde0da1/microdata-rdf/index.html#generate-the-triples 
> Gregg Kellogg: HTML Data TF has responsibility to document how to convert Microdata to RDF. [ Scribe Assist by Manu Sporny ]
> Niklas Lindström: This is an issue, no? <div vocab="http://purl.org/dc/terms/"><div typeof="bibo:LegalDocument"><p property="title">
> Manu Sporny: it is not if you don't know that you can use @vocab
> 
> Manu Sporny: I don't know if we should care about this use case, if people have a reason to migrate from Microdata to RDFa, they will find a way to do it. Having to learn about the concept of a vocabulary doesn't seem like a high barrier to me. It may be dangerous to do a global search/replace of @itemtype with @typeof. [ Scribe Assist by Manu Sporny ]
> Stéphane Corlosquet: I agree if you know that @vocab exists - it's not a problem... but if you don't know it exists - then it's difficult to learn that new concept. [ Scribe Assist by Manu Sporny ]
> Stéphane Corlosquet: However, with this, it's not as confusing - people don't have to learn about @vocab - they just use @typeof. [ Scribe Assist by Manu Sporny ]
> Manu Sporny: The mechanism to break the Microdata IRI into vocab is confusing, they'd have to learn that algorithm instead of @vocab. They already have to understand that there is a vocabulary somewhere. [ Scribe Assist by Manu Sporny ]
> Stéphane Corlosquet: RDFa has a mixture of attributes to be added to markup - in Microdata you just have @itemtype... in RDFa you have @vocab and @typeof. [ Scribe Assist by Manu Sporny ]
> Manu Sporny: Ok, we have discussed this thoroughly enough - is there any other information that we would need to understand to make an educated straw-poll on this item? [ Scribe Assist by Manu Sporny ]
> PROPOSED: Add functionality to @typeof where if the first token is an IRI, that sets the default vocabulary for processing.
> 
> Gregg Kellogg: -1
> Knud Möller: -1
> Sebastian Germesin: +1
> Niklas Lindström: -1
> Shane McCarron: -1
> Stéphane Corlosquet: +1
> 
> Manu Sporny: -1
> RESOLVED: Do not add functionality to @typeof where if the first token is an IRI, that sets the default vocabulary for processing.
> 
> 4. Schema.org supports RDFa in Rich Snippet Tool
> 
> Manu Sporny: Good news, schema.org supports RDFa in the Rich Snippet Testing Tool now. Let's all try to support them in implementing RDFa for schema.org
> http://openspring.net/blog/2011/09/30/schemaorg-rich-snippets-drupal-7-rdfa
> 
> --manu
> 
> -- 
> Manu Sporny (skype: msporny, twitter: manusporny)
> Founder/CEO - Digital Bazaar, Inc.
> blog: Uber Comparison of RDFa, Microformats and Microdata
> http://manu.sporny.org/2011/uber-comparison-rdfa-md-uf/
> 
> 


----
Ivan Herman, W3C Semantic Web Activity Lead
Home: http://www.w3.org/People/Ivan/
mobile: +31-641044153
PGP Key: http://www.ivan-herman.net/pgpkey.html
FOAF: http://www.ivan-herman.net/foaf.rdf