Chatlog 2012-01-26

From RDFa Working Group Wiki
Jump to: navigation, search

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

14:58:37 <RRSAgent> RRSAgent has joined #rdfa
14:58:37 <RRSAgent> logging to
14:58:39 <trackbot> RRSAgent, make logs world
14:58:39 <Zakim> Zakim has joined #rdfa
14:58:41 <trackbot> Zakim, this will be 7332
14:58:41 <Zakim> ok, trackbot; I see SW_RDFa()10:00AM scheduled to start in 35 minutes
14:58:42 <trackbot> Meeting: RDF Web Applications Working Group Teleconference
14:58:42 <trackbot> Date: 26 January 2012
14:58:53 <ShaneM> ShaneM has joined #rdfa
15:01:00 <Zakim> SW_RDFa()10:00AM has now started
15:01:04 <Zakim> + +1.540.961.aaaa
15:01:07 <Zakim> +OpenLink_Software
15:01:11 <gkellogg> gkellogg has joined #rdfa
15:01:11 <Zakim> - +1.540.961.aaaa
15:01:18 <niklasl> niklasl has joined #rdfa
15:01:25 <MacTed> Zakim, code?
15:01:25 <Zakim> the conference code is 7332 (tel:+1.617.761.6200, MacTed
15:01:28 <Zakim> +??P17
15:01:38 <manu1> zakim, I am ??P17
15:01:38 <Zakim> +manu1; got it
15:01:52 <manu1> zakim, who is on the call?
15:01:52 <Zakim> On the phone I see OpenLink_Software, manu1
15:01:59 <ivan> zakim, dial ivan-voip
15:02:00 <Zakim> ok, ivan; the call is being made
15:02:06 <Zakim> +Ivan
15:02:20 <Zakim> +??P24
15:02:20 <MacTed> Zakim, OpenLink_Software is temporarily me
15:02:30 <Zakim> +MacTed; got it
15:02:30 <niklasl> zakim, I am ??P24
15:02:46 <Zakim> +niklasl; got it
15:02:52 <Zakim> +??P27
15:02:52 <manu1> zakim, who is on the call?
15:02:56 <ShaneM> zakim, I am ??P27
15:02:58 <Zakim> On the phone I see MacTed, manu1, Ivan, niklasl, ??P27
15:03:02 <Zakim> +ShaneM; got it
15:03:33 <Zakim> +scor
15:03:49 <Steven_> Steven_ has joined #rdfa
15:03:50 <Zakim> +??P32
15:03:56 <gkellogg> zakim, I am ??P32
15:03:56 <Zakim> +gkellogg; got it
15:04:12 <ivan> scribenick: ivan
15:04:19 <ivan> scribe: ivan
15:05:06 <manu1> Agenda:
15:05:25 <ivan> manu: Any additions to the agenda?
15:05:25 <scor> scor has joined #rdfa
15:05:40 <ivan> ... Niklas raised the issue on invalid CURIEs in Facebook's Open Graph Protocol syntax.
15:05:50 <ivan> ... We should discuss that
15:05:59 <ivan> ... Stephane also sent comments, they were editorial
15:06:37 <ivan> ivan: One more agenda item - Should we explicitly disallow usage of xmlns: in RDFa Lite?
15:06:45 <ivan> manu: Yes, added to the Agenda. Shane, where are we with RDFa Core 1.1? Is it ready for Last Call?
15:07:36 <ivan> ShaneM: Niklas sent a reaction to the RDFa Core spec, I made some additional changes based on his comments
15:07:44 <ivan> ... if Niklas is satisfied then we are fine.
15:08:02 <ivan> ... Also, Manu asked that I add a note about the syntax for CURIEs possibly changing after Last Call if we change it now and screw something up. The intent on what we want to do for CURIEs is fairly clear, we just need to get the EBNF correct.
15:08:04 <Zakim> -niklasl
15:08:34 <manu1> Topic: Using xmlns: in an RDFa Lite 1.1 Document
15:08:51 <ivan> Ivan: Should we specifically disallow the usage of xmlns: in an RDFa Lite 1.1 document?
15:08:51 <ivan> manu: I do not think we need to say anything about it, we don't specifically disallow the other attributes you're not supposed to use. We specifically point out which RDFa attributes you can use and xmlns: is not in the set of allowable attributes.
15:09:27 <ivan> ... Also, there is now for HTML5/XHTML5 + RDFa 1.1/RDFa Lite 1.1 - the validator doesn't allow xmlns: in HTML5+RDFa 1.1 or HTML5+RDFa Lite 1.1, so I think we're fine.
15:09:51 <ivan> q+
15:10:01 <Zakim> +??P24
15:10:12 <niklasl> zakim, I am ??P24
15:10:12 <Zakim> +niklasl; got it
15:11:11 <ShaneM> zakim, who is on the call?
15:11:11 <Zakim> On the phone I see MacTed, manu1, Ivan, ShaneM, scor, gkellogg, niklasl
15:12:18 <MacTed> Zakim, who's noisy?
15:12:29 <Zakim> MacTed, listening for 10 seconds I heard sound from the following: MacTed (30%), niklasl (43%)
15:12:38 <manu1> zakim, mute niklasl
15:12:38 <Zakim> niklasl should now be muted
15:12:40 <MacTed> Zakim, mute me
15:12:40 <Zakim> MacTed should now be muted
15:12:52 <scor> q+
15:13:01 <manu1> ack ivan
15:13:40 <manu1> q+ to point out the issue.
15:13:58 <manu1> zakim, unmute niklasl
15:13:59 <Zakim> niklasl should no longer be muted
15:14:14 <manu1> ack scor
15:14:16 <ShaneM> q+ to talk about xmlns
15:19:02 <ShaneM> q-
15:19:27 <ShaneM> I am not in favor of disallowing "xmlns:" explicitly, the document already implies that it is not allowed.
15:19:27 <manu1> q+
15:19:51 <ivan> manu: For example, we do not explicitly disallow @resource or @datatype... but they are not allowed in RDFa 1.1 Lite documents.
15:20:04 <ivan> ... therefore, why explicitly refer to xmlns:?
15:20:16 <ivan> ivan: I understand your point
15:20:30 <gkellogg> q+
15:20:37 <ivan> ack manu1
15:20:37 <Zakim> manu1, you wanted to point out the issue. and to 
15:20:38 <manu1> ack manu1
15:20:43 <manu1> ack gkellogg
15:21:05 <niklasl> q+
15:21:06 <ivan> gkellogg: It is possible to have an XML-based host language that allows xmlns: and does not disallow using prefix 
15:21:36 <ivan> manu: If xmlns: exists, then the core processor must still use it...
15:21:48 <ivan> ... If there is an @xmlns: value and it is used to set a prefix, then the document isn't an RDFa Lite 1.1 document anymore.
15:22:33 <ShaneM> q+
15:22:50 <ShaneM> q-
15:23:02 <ShaneM> q+ to ask why we care of something can be labeled rdfa lite
15:23:13 <ivan> manu: We could add a note to RDFa Lite 1.1 stating that using @xmlns in a Host Language is allowed as long as you don't use it to set a prefix?
15:23:20 <manu1> ack niklasl
15:23:36 <ShaneM> q-
15:24:12 <ivan> niklasl: RDFa Lite 1.1 does not speak about html5, but not using xmlns: is implied I believe.
15:24:58 <ivan> manu: We can add a note explicitly that xmlns: is used in the host language then it is not to be used for prefix purposes in RDFa Lite.
15:25:50 <manu1> PROPOSAL: The RDFa Lite 1.1 spec should state that if RDFa Lite 1.1 is used in an XML-based language, that the usage of the xmlns facility is allowed as long as it is not used to declare CURIE prefixes.
15:26:00 <ivan> +1
15:26:06 <manu1> +1
15:26:07 <niklasl> +1
15:26:08 <scor> +1
15:26:09 <ShaneM> +1
15:26:10 <gkellogg> +1
15:26:15 <ivan> RESOLVED: The RDFa Lite 1.1 spec should state that if RDFa Lite 1.1 is used in an XML-based language, that the usage of the xmlns facility is allowed as long as it is not used to declare CURIE prefixes.
15:26:28 <ivan> Topic: Updating the CURIE Syntax
15:26:45 <ivan> manu: There are two issues here, the first is that Facebook uses multiple colons in their CURIEs, which is not allowed. The second is author confusion based on a prefix being defined to "http" without their knowledge of that happening, thus causing CURIE expansion on something that is supposed to be a plain URL.
15:26:49 <niklasl> q+
15:26:56 <ivan> ... How we do this can be left to the mailing list discussion.
15:27:04 <manu1> ack niklasl
15:27:15 <ivan> ... Shane's note is good enough to avoid last call again while we make sure that any changes we make today don't cause us to go through another Last Call.
15:27:20 <niklasl> This is of concern as well:
15:27:53 <ivan> niklasl: Apart from the change the syntax for ':', I believe the RDF WG comment can also be taken care of
15:28:30 <ivan> ... We can also address the mixing of 'http' prefixes with http:/// URIs.
15:28:46 <ivan> ... We cannot do the replacement with SPARQL PN_NAME, we discussed that related to issue 90 last week.
15:29:06 <ivan> ... and we had examples where PN_NAME would cause a big problem - namely with DBPedia and
15:29:40 <ivan> ... What Gavin was concerned about was that normal IRIs can be confused with CURIEs.
15:30:17 <ivan> ... I looked at the different URIs, 52% use the authority plus path format with a double slash.
15:31:00 <ivan> ... I am trying to figure out to see if we can also change the definition of CURIEs so that it disallows '//' at the start
15:31:23 <ivan> ... My belief that this will prevent a lot of potential bad markup in the case that a scheme IRI is defined.
15:31:39 <ivan> ... because these types of curies will not be seen as URIs with the CURIE syntax that I have outlined.
15:31:54 <ivan> ... and I have not seen any use for a curie with a '//' in the reference, either theoretical or real.
15:31:59 <manu1> q+ to support not allowing "//" starting CURIEs.
15:32:26 <manu1> ack manu1
15:32:26 <Zakim> manu1, you wanted to support not allowing "//" starting CURIEs.
15:32:37 <ivan> manu1: That was a convincing argument, I think we should make the "//" and ":" changes that Niklas proposes.
15:32:54 <ivan> ... The concern we had with Gavin's approach is that it changed too much
15:33:14 <ivan> ... what Niklas is proposing are minor tweaks in the CURIE definition, we know exactly why we're making these changes vs. the PN_NAME changes where we couldn't grasp the rammifications of the change.
15:33:35 <ivan> ... The data that Niklas is referring to is also very convincing. If we can prevent 52% of URI schemes from being accidentally overridden by prefix declarations, then we should do that.
15:34:22 <ivan> q?
15:34:22 <ShaneM> q+ 
15:34:26 <ivan> ack ShaneM 
15:35:01 <ivan> ShaneM:  While looking at the proposal, I was reminded that the tag required us to add a note into the document:
15:35:06 <ShaneM> When revising a language that has historically permitted URIs in certain locations (e.g., as values of a specific attribute), to ensure backward compatibility, language designers SHOULD NOT permit CURIEs (or safe_curies) as the datatype in the corresponding location, but SHOULD provide a new mechanism (e.g., a new attribute).
15:35:07 <ivan> ... We put that into the CURIE spec.
15:35:17 <ivan> ... We failed to put that in the RDFa Core spec.
15:35:38 <manu1> q+ to address TAG note.
15:35:44 <ivan> ... I don't feel strongly about this change, just pointing that out.
15:35:57 <niklasl> q+
15:36:08 <ivan> ... I maintain that the author cannot put a scheme into an RDFa attribute that will be misinterpreted in future.
15:36:22 <ivan> manu: What you say that the current algorithm is deterministic.
15:36:33 <ivan> ... it never will be misinterpreted.
15:36:52 <ivan> ShaneM: It will not be misinterpreted in a way that the doc author did not mean it.
15:37:04 <ivan> ack manu
15:37:04 <Zakim> manu1, you wanted to address TAG note.
15:37:15 <ivan> manu: We should not put that into the RDFa spec, it's unnecessary.
15:37:28 <ivan> ... We do not allow CURIEs in @href or @src
15:38:23 <manu1> ack niklasl
15:38:44 <ivan> q+
15:39:20 <ivan> niklasl:  ShaneM , you say that document author's CURIE will never be misinterpreted as an IRI?
15:39:22 <ivan> ... that is true
15:39:39 <ivan> ... but if somebody puts an IRI, then a prefix could overshadow it.
15:39:40 <manu1> q+ to say this is a slightly different issue.
15:40:04 <ivan> ShaneM:  but I control those prefix definitions.
15:40:22 <ivan> niklasl:  The unsettling bit to me is where people do not control the whole document, they don't control the prefix definitions.
15:40:49 <ivan> ... Internal to some companies, this is true as well - the markup can come from someone like Facebook.
15:41:06 <niklasl> Here's an example of what could happen based on a proprietary URI scheme used internally in a company:
15:41:07 <ivan> ... Those of the sorts of things we're trying to prevent from happening.
15:41:24 <ShaneM> q+
15:41:26 <ivan> ... Some people have a tendency to shoot themselves in the foot and we could prevent this if we just adopted the no '//' at the beginning of references rule.
15:41:32 <manu1> ack ivan
15:42:13 <manu1> Ivan: I don't know if this is necessary... if somebody that uses a string that looks like a CURIE or URI... the RDFa Processor will generate the same triples, but on different paths.
15:42:49 <manu1> Ivan: We have been losing an enormous amount of energy and time on this... the changes in the core document is minimal if we allow ':' and/or disallow '//'
15:43:12 <manu1> Ivan: We still refer to the RFC for the details. I propose that we should make these changes and move ahead.
15:43:16 <manu1> ack manu1
15:43:16 <Zakim> manu1, you wanted to say this is a slightly different issue.
15:43:43 <ivan> manu: It is important to note is that an RDFa Processor doesn't misinterpret what is written, but an author could misinterpret what they've written. We should try to prevent that from happening, if at all possible.
15:43:54 <ivan> ... This is about protecting document authors from themeselves.
15:44:10 <ShaneM> q-
15:44:30 <ivan> ... If we make no change, then the danger is that 52% of the URI schemes already out there could trigger this issue if a prefix is defined that overlaps with the URI scheme.
15:44:56 <ivan> ... If we make the change, we will prevent the authoring mistakes. There is no down-side other than people can't use '//' to start out a reference... and we have no use case that has come up over the past six years that requires that people need to be able to do that.
15:45:11 <ShaneM> For the record I note that I agree with Niklas, and the risk is far higher than just schemes that have an authority section.  consider 'widget'.  'widget:foo' is a CURIE, but there is a widget scheme out there that has no authority section anyway.  So it will be confused.
15:45:14 <MacTed> Can we get proposal as Ivan worded?
15:45:28 <MacTed> Let's resolve and move on...
15:45:39 <ShaneM> Here's my wording in the spec: <p class="note">The working group is currently examining the productions
15:45:39 <ShaneM>       for CURIE below in light of recent comments received from the RDF 
15:45:39 <ShaneM>       Working Group and members of the RDF Web Applications Working
15:45:39 <ShaneM>       Group.  It is possible that there will be minor changes to the production
15:45:39 <ShaneM>       rules below in the near future, and that these changes will be
15:45:41 <ShaneM>       backward <em>incompatible</em>.  However, any such incompatibility will be
15:45:42 <ShaneM>       limited to edge cases.</p>
15:46:13 <ivan> PROPOSED: Make the change on the CURIE definition in RDFa Core 1.1, according to Niklas' e-mail, allow for ':' and prevent the use of '//' in the reference portion of a CURIE.
15:46:21 <MacTed> +1
15:46:23 <manu1> +1
15:46:23 <niklasl> +1
15:46:27 <ivan> +1
15:46:29 <gkellogg> +1
15:46:39 <ShaneM> +1 if there is actually BNF
15:46:40 <scor> +1
15:46:43 <ivan> RESOLVED: Make the change on the CURIE definition in RDFa Core 1.1, according to Niklas' e-mail, allow for ':' and prevent the use of '//' in the reference portion of a CURIE.
15:47:21 <niklasl> Here's the BNF - Option C at the end of: 
15:47:33 <niklasl> curie       ::= [ [ prefix ] ':' ] reference
15:47:48 <niklasl> reference   ::= ( ipath-absolute / ipath-rootless / ipath-empty ) [ "?" iquery ] [ "#" ifragment ]
15:48:21 <MacTed> Zakim, unmute me
15:48:21 <Zakim> MacTed should no longer be muted
15:48:27 <niklasl> .. definitions from [RFC3987]
15:49:50 <ivan> Topic: Last Call Publication
15:49:50 <ivan> Manu: Any reason we shouldn't take RDFa Core 1.1, RDFa Lite 1.1 and XHTML+RDFa 1.1 to Last Call?
15:51:30 <ivan> No opposition to Last Call.
15:51:40 <manu1> PROPOSAL: Take RDFa Core 1.1, RDFa Lite 1.1 and XHTML+RDFa 1.1 to Last Call with a publication date of January 31st 2012, with a Last Call Duration of 3 weeks, ending on February 21st 2012.
15:51:53 <gkellogg> +1
15:51:54 <niklasl> +1
15:51:54 <ivan> +1
15:51:56 <scor> +1
15:51:56 <MacTed> +1
15:51:56 <manu1> +1
15:51:57 <ShaneM> +1
15:52:33 <manu1> RESOLVED: Take RDFa Core 1.1, RDFa Lite 1.1 and XHTML+RDFa 1.1 to Last Call with a publication date of January 31st 2012, with a Last Call Duration of 3 weeks, ending on February 21st 2012.
15:52:31 <scor> ShaneM, you will review my changes before sending to to LC right?
15:52:31 <ivan> Manu: Stephane, I'm sure Shane will, but your changes were editorial, so we don't need to make them right now... but I'm sure Shane will make them anyway.
15:53:33 <manu1> Topic: RDFa 1.1 Implementations
15:52:31 <ivan> Manu: Who is planning to do an RDFa 1.1 implementation?
15:53:53 <gkellogg> I already have an implementation in Ruby
15:54:01 <MacTed> "implementation" means so many things
15:54:06 <niklasl> I have started on one in Clojure
15:54:17 <ShaneM> I am planning to update SPREAD (Perl)
15:55:03 <gkellogg> We really need a JavaScript implementation, though.
15:55:50 <ivan> manu: do people feel they can finish their implementations before April?
15:55:57 <niklasl> Gregg, I could try to run it through ClojureScript. :) But I believe that Antonio Garotte has one in the making?
15:57:21 <niklasl> My rough clojure work is at btw; note that it's still quite incomplete)
15:57:21 <ivan> Manu: So we have 6 commitments to implementations of RDFa 1.1 - that's far more than the 2 necessary (which are already done - Ivan's and Gregg's), so we're in good shape for the Candidate Recommendation phase.
15:58:17 <ShaneM> q+
15:58:33 <manu1> ack shanem
16:00:06 <manu1> ack shanem
16:01:35 <Zakim> -ShaneM
16:01:39 <Zakim> -MacTed
16:01:43 <Zakim> -gkellogg
16:02:01 <Zakim> -scor
16:02:31 <Zakim> SW_RDFa()10:00AM has ended
16:02:33 <Zakim> Attendees were +1.540.961.aaaa, manu1, Ivan, MacTed, niklasl, ShaneM, scor, gkellogg