Chatlog 2013-01-10

See CommonScribe Control Panel, original RRSAgent log and preview nicely formatted version.

14:53:47 <RRSAgent> RRSAgent has joined #rdfa
14:53:47 <RRSAgent> logging to http://www.w3.org/2013/01/10-rdfa-irc
14:53:49 <trackbot> RRSAgent, make logs world
14:53:49 <Zakim> Zakim has joined #rdfa
14:53:51 <trackbot> Zakim, this will be 7332
14:53:51 <Zakim> ok, trackbot; I see SW_RDFa()10:00AM scheduled to start in 7 minutes
14:53:52 <trackbot> Meeting: RDFa Working Group Teleconference
14:53:52 <trackbot> Date: 10 January 2013
14:59:21 <manu> Agenda: http://lists.w3.org/Archives/Public/public-rdfa-wg/2013Jan/0047.html
14:59:46 <gkellogg> gkellogg has joined #rdfa
15:00:45 <Zakim> SW_RDFa()10:00AM has now started
15:00:51 <Zakim> +??P16
15:00:52 <Zakim> +Shane_McCarron
15:01:00 <ShaneM> zakim, I am Shane_McCarron
15:01:00 <Zakim> ok, ShaneM, I now associate you with Shane_McCarron
15:01:09 <niklasl> zakim, I am ??P16
15:01:09 <Zakim> +niklasl; got it
15:01:22 <ivan> zakim, dial ivan-voip
15:01:23 <Zakim> ok, ivan; the call is being made
15:01:23 <Zakim> +Ivan
15:02:08 <Zakim> +??P21
15:02:13 <gkellogg> zakim, I am ??P21
15:02:14 <Zakim> +gkellogg; got it
15:02:34 <Zakim> + +1.347.439.aaaa
15:03:30 <Zakim> +??P26
15:03:33 <manu> zakim, I am ??P26
15:03:33 <Zakim> +manu; got it
15:05:43 <Zakim> +scor
15:06:28 <ShaneM> Scribe: ShaneM
15:06:28 <ShaneM> Guest: Sebastian Heath
15:07:15 <ShaneM> Topic: Administrivia
15:07:28 <ShaneM> It is very unlikely that we will get through all of the issues today. The group will switch to weekly telecons until we've processed all of the remaining issues. 
15:07:35 <ShaneM> The group is on track to take HTML+RDFa 1.1 to Last Call at the end of January 2013.
15:07:40 <manu> Topic: ISSUE-147: Preserve markup by default
15:07:49 <manu> http://www.w3.org/2010/02/rdfa/track/issues/147
15:08:55 <ShaneM> Manu: In RDFa 1.0 we used to generate an XMLLiteral by default if there was markup in content.  in RDFa 1.1 we decided to change the behavior because a number of people said that the data that was being gleaned from web sites had markup the authors probably did not intend.
15:09:08 <ShaneM> ... Sebastian has raised an issue that this is bad behavior.
15:09:49 <ShaneM> Sebastian summarizes the arguments made on the mailing list.
15:11:11 <ShaneM> ...  The system should not discard information by default.
15:12:07 <niklasl> q+
15:12:12 <manu> q+
15:12:14 <manu> ack niklasl
15:12:16 <ShaneM> ... Current version forces authors to propagate @datatype throughout a document when they want to preserve markup - error prone.
15:12:33 <gkellogg> q+
15:12:34 <scor> scor has joined #rdfa
15:13:23 <ShaneM> niklasl: Is it even possible to add the suggested attribute (from a process perspective)?  We have not been adding attributes in host language mappings.
15:13:54 <ShaneM> ... Is there a live demo of the data from Sebastian?
15:14:13 <ShaneM> Heath: Yes - I referenced it on the mailing list.  A good example is a journal.
15:15:04 <manu> here's one example of what Sebastian is working on: http://dlib.nyu.edu/awdl/isaw/isaw-papers/4/preprint/
15:15:25 <ShaneM> ... also geographical data.  Sometimes there are italics.  While this is presentational, it is also semantic.
15:15:41 <gkellogg> Preserving the Right-to-Left nature of markup was the most convincing to me.
15:16:06 <manu> ack manu
15:17:01 <gkellogg> As turtle: http://rdf.greggkellogg.net/distiller?format=turtle&in_fmt=rdfa&uri=http://dlib.nyu.edu/awdl/isaw/isaw-papers/4/preprint/
15:17:03 <manu> http://dlib.nyu.edu/awdl/isaw/isaw-papers/2/
15:17:22 <manu> That's the link that Sebastian was talking about.
15:18:44 <ShaneM> manu: I don't think that Sebastian is asking for a new attribute.  It is a property in the head that would switch on preservation of markup.
15:18:47 <gkellogg> We already have flags to affect the output graph and vocabulary expansion
15:19:03 <ShaneM> Heath: The worst case would be something that was a flag on a tool.
15:19:05 <manu> ack gkellogg
15:19:08 <manu> q+
15:19:57 <ShaneM> gkellogg: First, if we make this change we will create a further divergence between RDFa 1.1 Lite and Microdata.  That would be a disservice.
15:20:32 <niklasl> q+
15:20:57 <ShaneM> ... Second, what percentage of the markup is textnodes that are not whitespaces?   Is there a way to detect semantics of the embedded markup?
15:21:29 <ShaneM> ... Lastly, we do already have flags to effect processor behavior.  It is not unprecedented.
15:21:54 <ivan> q+
15:22:37 <ShaneM> manu: Restates Sebastian's request in preference order
15:23:29 <manu> ack manu
15:23:38 <manu> niklasl: Yes, that's the second suggestion.
15:23:41 <gkellogg> q+
15:23:45 <manu> the second suggestion to be something like: … <head> ... <link property="rdfa:defaultDatatype" resource="rdf:HTML" />
15:23:48 <manu> ack niklasl
15:25:03 <ShaneM> niklasl: If we were to inspect the content to try to determine whether to extract text or preserve markup, the way to do that reasonably would be to have a limited list HTML elements that have semantics
15:26:04 <manu> ack ivan
15:26:21 <ShaneM> ... might be a lot of opinions about what those semantic elements are.
15:26:56 <ShaneM> ivan: I am afraid that we are discussing something where "the boat has already sailed".  The goal we have with HTML5+RDFa is to define a host language for RDFa 1.1.
15:27:21 <ShaneM> ... that means make the minimal customizations.
15:29:26 <ShaneM> ... Such a flag would require a change on all processors.  But it could cause a divergence between HTML+RDFa and XHTML+RDFa or SVG+RDFa.
15:30:44 <manu> ack gkellogg
15:30:46 <ShaneM> ... Finally, if we put in a flag like this we are opening up the floodgates.  We are introducing a way to modify the behavior of the processor from the document source.  If we proviide such a mechanism then we could end up with all sorts of additional requests for modificiations to the behavior.
15:31:01 <ShaneM> q+ to point out that lots of processors have "pragma" directives.
15:32:28 <ShaneM> gkellogg: Adding a property means modifying the document.  If you are modifying the document then why not just add the datatypes?  Is it because the content you are including in the document comes from elsewhere?
15:33:07 <manu> q+ to ask if a processor flag would help for raptor?
15:33:54 <ShaneM> Heath: My documents are "xhtml".  In my world I want to preserve the markup.  It is well-formed XML.  Putting datatype on every element would have a cost.
15:34:46 <ShaneM> ... I have beginning users doing this.  In 1.0 I didn't need to give them any guidance.  With 1.1 now those users need to decide whether datatype should be set or not.
15:34:58 <niklasl> q+ to ask about when the choice of what is meaningful is to be made
15:35:22 <ShaneM> ... I realize there are ways to achieve this via pre-processing.  But I shouldn't have to do this for what is to me a simple, common use case.
15:37:01 <ShaneM> ... The implementation mechanism is pretty clear in terms of logic if there were a property added.
15:38:49 <ShaneM> ... I have real use cases and hope they are compelling.  The working group is doing something else that has effectively no use cases already.  That seems unfortunate.  There should be some way to accommodate this use case.
15:39:34 <ShaneM> gkellogg: There are use cases for the itemref stuff.
15:39:59 <manu> ack Shane_McCarron
15:39:59 <Zakim> Shane_McCarron, you wanted to point out that lots of processors have "pragma" directives.
15:40:45 <manu> Shane: I appreciate that people don't want to add a property. For those of you that have been involved in this for a long time, we have discussed this sort of thing before - a way to automatically modify the behavior of the processor through document content or profiles.
15:41:19 <ivan> q+
15:41:23 <manu> Shane: This isn't a new thing for us - every time we discuss it, we come to the decision that there is no way we could do it correctly. Most processors have #pragmas. I don't mind the concept of allowing a #pragma.
15:41:36 <manu> Shane: I don't think we should say that we should never do this because it would open the floodgates.
15:41:45 <manu> ack manu
15:41:46 <Zakim> manu, you wanted to ask if a processor flag would help for raptor?
15:41:49 <manu> ack niklasl
15:41:49 <Zakim> niklasl, you wanted to ask about when the choice of what is meaningful is to be made
15:42:10 <manu> q+ to ask if a processor flag would help for raptor?
15:42:25 <ShaneM> niklasl: your needs to preserve the markup seems to be about it being hard in your situation what is meaningful when the markup is tagged.
15:43:09 <ShaneM> Heath: No - we know that it is all meaningful in my use case.
15:43:38 <ShaneM> niklasl: And you have a way to do that with @datatype.
15:44:21 <ShaneM> Heath: Yes, and if that is the ultimate solution then I lose.  Because the rules for @datatype are confusing.  If there is @property you do one thing.  If there is @rel there is another.
15:45:05 <ShaneM> ... it is error prone.  Even experts will forget to put it on places.  No wysiwyg editor supports this, so people are going to be forced to use 'pointy brackets'.
15:46:17 <ShaneM> niklasl: We had lots of real world examples that caused us to change the behavior for RDFa 1.1.  Also, in RDFa 1.1 we have the ability to process the nested content, so we can't change the default behavior without breaking that.
15:47:13 <manu> ack ivan
15:48:21 <ShaneM> ivan: I would like to stop the discussion about default behavior.  We can't change it right now procedurally.  The only thing we could do now is introduce some sort of 'pragma' and have this feature as the first example of such a pragma.
15:48:51 <ShaneM> ... this means such a departure for the processors that it can invalidate the behavior of all processors that are already deployed.
15:50:19 <ShaneM> manu: Straw polls....
15:51:07 <tinkster> tinkster has joined #rdfa
15:51:43 <manu> PROPOSAL: HTML+RDFa 1.1 should preserve markup as typed literals with the datatype "rdf:HTML" when emitting triples.
15:51:49 <manu> -1
15:51:52 <ivan> -1
15:51:53 <niklasl> -1
15:51:54 <gkellogg> -1
15:52:01 <ShaneM> Shane: -1
15:52:09 <ShaneM> Heath: +1
15:53:07 <manu> PROPOSAL: Add a pragma directive to HTML+RDFa that switches the RDFa processor into a mode where markup is preserved as a typed literal with the datatype of "rdf:HTML" when emitting triples.
15:53:27 <gkellogg> +0.2
15:53:28 <manu> +0
15:53:29 <ShaneM> Heath: +1
15:53:33 <scor> +0
15:53:48 <ivan> -1
15:54:11 <ShaneM> Shane: +0
15:54:23 <niklasl> +0
15:55:35 <niklasl> q+
15:55:51 <niklasl> q-
15:56:22 <manu> PROPOSAL: Allow RDFa processor authors to add a flag that allows a default datatype for complex markup to be set when invoking the processor.
15:56:25 <manu> +0.4
15:56:32 <gkellogg> +0.2
15:56:37 <ivan> +0.7
15:56:53 <ShaneM> Shane: +0
15:56:56 <niklasl> -1
15:57:28 <scor> +0.8
15:59:08 <manu> RESOLVED: Allow RDFa processor authors to add a flag that allows a default datatype for complex markup to be set when invoking the processor.
16:03:50 <manu> Niklas: Future RDFa 2.0 could consider in-document pragmas, based on observing if a lot of energy is needed in the wild for keeping XML-literals. The same observance should be made about white-space preserving, inheriting language etc.
16:05:56 <Zakim> -manu
16:05:57 <Zakim> -gkellogg
16:05:57 <Zakim> - +1.347.439.aaaa
16:06:01 <Zakim> -Ivan
16:06:07 <Zakim> -scor
16:06:08 <Zakim> -niklasl
16:11:09 <Zakim> disconnecting the lone participant, Shane_McCarron, in SW_RDFa()10:00AM
16:11:12 <Zakim> SW_RDFa()10:00AM has ended
16:11:12 <Zakim> Attendees were Shane_McCarron, niklasl, Ivan, gkellogg, +1.347.439.aaaa, manu, scor
16:12:58 <manu> rrsagent, make logs public
# SPECIAL MARKER FOR CHATSYNC.  DO NOT EDIT THIS LINE OR BELOW.  SRCLINESUSED=00000147