Chatlog 2010-10-28

From RDFa Working Group Wiki
Jump to: navigation, search

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

13:58:52 <RRSAgent> RRSAgent has joined #rdfa
13:58:52 <RRSAgent> logging to
13:58:54 <trackbot> RRSAgent, make logs world
13:58:56 <trackbot> Zakim, this will be 7332
13:58:56 <Zakim> ok, trackbot; I see SW_RDFa()10:00AM scheduled to start in 2 minutes
13:58:57 <trackbot> Meeting: RDFa Working Group Teleconference
13:58:57 <trackbot> Date: 28 October 2010
13:59:08 <manu1> Agenda:
13:59:11 <manu1> Chair: Manu
13:59:23 <manu1> Present: Benjamin, Ivan, Knud, Manu, Steven, Nathan, Shane
14:00:05 <Zakim> SW_RDFa()10:00AM has now started
14:00:06 <Zakim> +??P36
14:00:24 <Zakim> + +3539149aaaa
14:00:24 <Zakim> - +3539149aaaa
14:00:24 <Zakim> + +3539149aaaa
14:00:54 <Zakim> +??P43
14:01:05 <manu1> zakim, I am ?P43
14:01:05 <Zakim> sorry, manu1, I do not see a party named '?P43'
14:01:08 <manu1> zakim, I am P43
14:01:08 <Zakim> sorry, manu1, I do not see a party named 'P43'
14:01:12 <manu1> zakim, I am ??P43
14:01:12 <Zakim> +manu1; got it
14:01:16 <Knud> zakim, I am aaaa
14:01:16 <Zakim> +Knud; got it
14:01:23 <Knud> zakim, mute me
14:01:23 <Zakim> Knud should now be muted
14:01:23 <nathan> zakim, I am ??P36
14:01:24 <Zakim> +nathan; got it
14:01:25 <Zakim> +ShaneM
14:01:42 <ShaneM> ShaneM has joined #rdfa
14:01:45 <manu1> zakim, who is on the call?
14:01:45 <Zakim> On the phone I see nathan, Knud (muted), manu1, ShaneM
14:05:17 <Zakim> +??P26
14:05:17 <Steven> zakim, dial steven-617
14:05:18 <Zakim> ok, Steven; the call is being made
14:05:18 <Zakim> +Steven
14:05:44 <Benjamin> zakim, I am ??P26
14:05:44 <Zakim> +Benjamin; got it
14:06:02 <Benjamin> zakim, mute me
14:06:02 <Zakim> Benjamin should now be muted
14:09:54 <manu1> Scribe: Nathan
14:10:01 <manu1> scribenick: nathan
14:10:46 <nathan> Topic: XHTML+RDFa preparation for LC
14:12:11 <nathan> ShaneM: I'm available over the next couple of weeks to get LC ready
14:14:09 <nathan> manu1: Knud and Thomas Steiner will have comments in by weekend, try to get a straw poll next thursday, Stephen and Ivan to send via mailing list as away next week
14:15:09 <nathan> manu1: provisional LC publishing date Tuesday 9th November
14:15:15 <nathan> manu1: Any objections to this approach and schedule?
14:15:15 <nathan> No objections noted.
14:15:17 <nathan> Topic: RDFa Core Test Suite
14:16:16 <nathan> manu1: thinking this time around we can spend less time discussing test suite on telecon and encourage test submission via mailing list and github from implementers
14:16:36 <ShaneM> q+ to talk about test suite control
14:17:02 <nathan> manu1: implementers can / will naturally provide review and feedback for Test Suite, no need for us to spend precious telecon time on that.
14:17:16 <manu1> ack shanem
14:17:16 <Zakim> ShaneM, you wanted to talk about test suite control
14:18:15 <nathan> ShaneM: I agree with process, think that if you're submitting a test case you should submit it for all host languages (XHTML1, HTML4 and HTML5)
14:19:39 <nathan> ShaneM: there was a bottleneck last time for getting tests in to the manifest, would be good to have this bottleneck cleared this time
14:20:22 <nathan> manu1: manifests are all on github and are easily managed, manifests contain full details of each test, clarification on bottleneck?
14:21:13 <nathan> ShaneM: perhaps bottleneck has been cleared previously.. how are you suggesting I add a test case?
14:21:29 <nathan> manu1: normal github process - check out source code, commit changes, request pull, etc
14:21:43 <nathan> ShaneM: okay, great
14:22:57 <ShaneM> (maybe reserve the right to have test issues raised to a call?)
14:23:08 <nathan> manu1: clarify, we are removing discussion of tests from telecons and moving to mailing list process, tests can be discussed on telecon if they're generating conflict and cannot be resolved on the list.
14:23:47 <nathan> Topic: RDFa API
14:23:53 <nathan> scribenick: manu1
14:24:29 <manu1> Nathan: There are several issues w/ RDFa API while implementing over the last 2-3 months
14:24:39 <manu1> Nathan: Some of them are small issues, fairly easy to fix
14:24:54 <manu1> Nathan: Some of them require a restructuring around DataStore - good chunk of time to sort that out.
14:25:03 <manu1> Nathan: Perhaps we can clear up some of the smaller issues today?
14:25:15 <manu1> Nathan: I sent an e-mail to mailing list yesterday about this
14:25:21 <nathan>
14:25:53 <manu1> Nathan: We have no way in the API at the moment to see if two nodes are equal.
14:26:09 <manu1> Subtopic: ISSUE-49 RDFNode type, equality and canonicalization
14:26:23 <manu1> Nathan: It's a simple proposal, add a .equals() method to RDFNode.
14:26:47 <ivan> zakim, dial ivan-voip
14:26:54 <Zakim> ok, ivan; the call is being made
14:26:56 <Zakim> +Ivan
14:27:51 <manu1> Nathan: Second sub-issue is how do we canonicalize?
14:29:05 <manu1>
14:30:11 <ivan> Nathan means, we should canonicalize according to NT, like so: <http://dfasfasd.asdfasda>
14:30:32 <manu1> Manu: I'm concerned that we're making a decision to officially support N-Triples.
14:30:48 <manu1> q+ to discuss NTriples for RDFTriples.
14:31:12 <manu1> Ivan: No, we're just stating a canonicalization format.
14:31:15 <manu1> ack manu
14:31:15 <Zakim> manu1, you wanted to discuss NTriples for RDFTriples.
14:33:02 <manu1> Manu: I know we're talking canonicalization, but we're also saying that we're going to serialize to NTriples to do that and provide an API hook to serialize to NT.
14:33:07 <manu1> Nathan: That's correct.
14:34:41 <manu1> Ivan: This is what happens in rdflib
14:35:02 <manu1> Ivan: rdflib will create NTriples to serialize the string.
14:35:04 <ivan> zakim, who is here?
14:35:04 <Zakim> On the phone I see nathan, Knud (muted), manu1, ShaneM, Benjamin (muted), Steven, Ivan
14:35:07 <Zakim> On IRC I see ShaneM, RRSAgent, Benjamin, Knud, Steven, trackbot, Zakim, manu1, ivan, nathan
14:35:29 <manu1> Nathan: The other issues is the NoInterfaceObject, which means typeof() will return "undefined"
14:35:44 <manu1> Nathan: There is no way to know if a plain literals is what it is.
14:36:08 <manu1> Nathan: We also need a way to expose which interface is implemented. nodeType property or something of that sort.
14:37:38 <Benjamin> I like and support these changes
14:37:41 <manu1> Ivan: We should call toNT() something like toCanonical()
14:39:49 <manu1> General agreement that the changes that Nathan proposes are good and the group wants to make them.
14:40:21 <manu1> Subtopic: ISSUE-55 Specifying that implementations must implement TypedLiteralConverters for all xsd numerical types 
14:40:28 <manu1>
14:41:38 <manu1> Nathan: Basically, we should support all the basic xsd: number types 
14:41:54 <manu1> Manu: So, the default context in RDFa API should provide all these type converters by default?
14:42:36 <manu1> Nathan: Yes, RDFa implementations MUST support typed literal converters for basic xsd: numerical types 
14:42:44 <manu1> Ivan: What about dates?
14:43:00 <manu1> Nathan: Supporting xsd:date is already in the spec.
14:43:09 <manu1> Nathan: We support xsd:date xsd:dateTime etc.
14:43:28 <manu1> Nathan: This change is specifically about numerical types.
14:44:18 <manu1> Manu: How do we get rid of typed literal converters?
14:44:22 <manu1> Nathan: That's another issue in the issue tracker.
14:44:57 <manu1> From the issue: support types like these (xsd:int/double/decimal/unsigned* positive* etc.)
14:46:09 <manu1> General agreement from the group that we want to support this mechanism.
14:47:42 <manu1> Subtopic: ISSUE-58: PlainLiteral and TypedLiteral value for strings 
14:48:20 <manu1> Nathan: What is the value for a plain literal when you have special characters? Should the output be encoded into JavaScript?
14:48:30 <manu1>
14:49:00 <manu1> Ivan: What is the rule in RDFa? We copy the bytes verbatim unicode?
14:49:04 <manu1> Shane: Yes, that is correct.
14:49:19 <nathan> Primarily, if we have the following triple:
14:49:19 <nathan>   <#nbsp> <> "Hello \n\tWorld." .
14:49:19 <nathan> Do we expect the related call to object.toString() and/or object.value to return
14:49:19 <nathan> "Hello \n\tWorld."
14:49:19 <nathan> or
14:49:20 <nathan> "Hello
14:49:22 <nathan>     World." 
14:49:23 <manu1> Shane: Actually, it's not in unicode, we don't change the document encoding - we copy directly.
14:50:02 <manu1> Manu: It's the second one.
14:50:17 <manu1> Shane: The data isn't transformed, if you need to transform it, you have to do it - JSON-encode it, whatever.
14:50:24 <manu1> Nathan: Great, that works.
14:51:24 <manu1> General agreement that we do not attempt to encode the plain literal data in any way, we pass the data through.
14:52:59 <manu1> Subtopic: ISSUE-57: TypedLiteralConverter Failures and Exceptions
14:53:01 <nathan>
14:54:01 <manu1> Nathan: I'm proposing that we don't generate any exceptions.
14:54:05 <manu1> Manu: I agree that we shouldn't generate any exceptions.
14:57:08 <manu1> Nathan: There is a third option - null or undefined.
14:57:18 <manu1> Nathan: Having out-of-band information may be messy.
14:57:36 <manu1> Nathan: When you use complex types, it could become messy.
14:57:58 <manu1> Nathan: Just returning the same thing that was passed in may be impossible.
14:59:18 <manu1> Manu: Perhaps we should use 'undefined' instead of 'null'
14:59:24 <ShaneM> I can't imagine a type literal that would need to map to undefined?
14:59:34 <manu1> Ivan: What's the style in webapps programming? 'null' or 'undefined'
15:00:02 <manu1> Nathan: This is a slightly different case - normally you'd be dealing w/ booleans, strings, integers, etc.
15:00:15 <manu1> Nathan: I think this is unique to the RDF API.
15:02:25 <Zakim> -Benjamin
15:02:45 <manu1> Manu: What about if you want to express "\0"? Typed literal converter should return 'null' in this case.
15:03:05 <Benjamin> Benjamin has joined #rdfa
15:03:24 <manu1> <> foaf:name "\0" .
15:03:52 <ShaneM> <> foaf:name "\NULL"^mytype:pointer
15:03:56 <nathan> Nathan: We do have the concept of rdf:nil
15:04:17 <manu1> <> foaf:name "^^mytype:null" .
15:05:05 <manu1> <> foaf:name "^^mytype:undefined" .
15:07:04 <manu1> Manu: I think we should use 'undefined'
15:07:22 <manu1> Nathan: I think we should use 'null'
15:07:47 <manu1> Shane: In XML, you have nil-able values - it's perfectly legal, null would be used in that case.
15:08:05 <manu1> Shane: We should marshall it as something - maybe the empty string, not "null" or something like that.
15:08:10 <manu1> Ivan: empty string is different.
15:08:43 <manu1> Nathan: Could you implement 'undefined' in Python?
15:08:53 <manu1> Ivan: I may be able to use 'None'
15:09:19 <manu1> Shane: You can explicitly check to see if its None, as you can in Perl.
15:09:33 <manu1> Ivan: This is a common idiom in Python.
15:09:35 <manu1> Ivan: 'None' in Python is a keyword and you can check directly for its value.
15:09:40 <manu1> General agreement that a failed conversion for valueOf() should be communicated in-band - using something like 'null' or 'undefined'. Nathan will ask WebApps for guidance for the type used.
15:09:42 <manu1> q+ to end the telecon
15:10:51 <manu1> ack manu1
15:10:51 <Zakim> manu1, you wanted to end the telecon
15:14:03 <Zakim> -Steven
15:14:04 <Zakim> -nathan
15:14:06 <Zakim> -Knud
15:14:10 <Zakim> -ShaneM
15:14:11 <Zakim> -manu1
15:14:12 <ivan> zakim, drop me
15:14:12 <Zakim> Ivan is being disconnected
15:14:14 <Zakim> SW_RDFa()10:00AM has ended
15:14:16 <Zakim> Attendees were +3539149aaaa, manu1, Knud, nathan, ShaneM, Steven, Benjamin, Ivan