and
TextThis is an automatically generated IRC chat log made by the logger bot from the Semantic Web Interest Group IRC chat at irc://irc.freenode.net/swig (also known as server irc.freenode.net channel #swig if that URI does not work for you).
See also the Semantic Web Interest Group IRC Scratchpad for the collaboratively written weblog and ESW wiki.
Semantic Web Interest Group Logs > 2005 > 2005-01 > 2005-01-12 (Latest) (Search)
00:03:19 <ows> night
00:07:20 * DanC wanders off to family time...
01:24:16 <masaka>http://www.kanzaki.com/works/2005/imgdsc/flickr2rdf
01:24:16 <dc_swig> A: http://www.kanzaki.com/works/2005/imgdsc/flickr2rdf from masaka
01:24:33 <masaka> A:|Flickr photo info to RDF image description
01:24:33 <dc_swig> Titled item A.
01:25:08 <masaka> A: This service uses Flickr API to extract metadata from Flickr's photo repository, and generates an RDF description. Uses flick tag as foaf:topic.
01:25:09 <dc_swig> Added comment A1.
01:25:19 <masaka> A: If flickr 'note' presents, it will be translated into W3Photo Image Region Vocabulary, and be demonstrated via XSLT.
01:25:20 <dc_swig> Added comment A2.
01:26:19 <eaon> masaka: cool - but shouldn't that use foaf:maker instead of dc:creator there?
01:27:45 <masaka> hi eaon:, yes, but i wonder danbri's note on dc:creator...
01:28:20 <masaka> here it is: http://danbri.org/words/?p=63
01:30:12 <masaka> I'm fine to change it to foaf:maker though.
01:31:56 <kota> whoa it keeps my notes! cool!
01:34:11 <masaka> kota: note is one of the most fantastic features of flickr :)
01:35:42 <masaka> that makes image (region) annotation so easy for everyone
01:36:15 <md-afk> md-afk is now known as PhUrl
01:36:24 <kota> yeah :) ah isn't the note thing based on w3photo's?
01:37:01 <masaka> flickr note you mean, kota ?
01:37:45 <kota> yep, i think i've read some blog posts saying that...
01:38:10 <masaka> flickr doesn't use RDF in its native format
01:38:46 <masaka> so flickr2rdf translates its information into RDF, with foaf and w3photo vocab
01:39:30 <masaka> ah, you mean fotonotes ?
01:40:18 <kota> yeah, maybe that one. that was a time ago so i might be confused.
01:40:48 <masaka> it could be. find the blog and tell us :)
03:05:14 * DanC reads about the new Mac goodies...
03:05:19 <phenny> DanC: 02:26Z <sbp> tell DanC that the 2004-12-13 CWM/N3 chat did indeed take place as scheduled*, and that TimBL has asked me to arrange another though I haven't got around to it yet. hope you like n3p. * http://ilrt.org/discovery/chatlogs/swig/2004-12-13.html#T18-31-53
03:43:31 <Mo666> hi! i'm looking for some sort of rdf/rss to irc bot code. Some bot which sends headline and url to an irc channel on updates of an rdf feed.
03:43:56 <Mo666> have you ever heard of anything like that?
03:44:36 <crschmidt> hm, sbp, didn't Morbus write one of those?
03:44:42 <crschmidt> Or did he just teal it from somewhere?
03:45:47 <sbp> I think he stole it
03:45:53 <crschmidt> steal, yeah
03:45:54 <sbp> but I can't remember. hang on
03:45:55 <crschmidt> not the color
03:46:12 <Mo666> cool
03:46:35 <Mo666> i only know about the logger bot and the scrachpad so far, but they are both the other direction
03:46:39 <sbp> ah: http://rssbot.sourceforge.net/
03:46:45 <crschmidt> hey, look taht that
03:46:49 <crschmidt> .g rssbot
03:46:57 <phenny> rssbot: http://rssbot.sourceforge.net/
03:47:11 <Mo666> aah, that's great, thanks so much!
03:47:19 <crschmidt> google probably works wonders in this case
03:47:21 <crschmidt> yep
03:48:08 <Mo666> crschmidt, i must have chosen bad terms. what would you search for?
03:48:29 <Mo666> i tried: rss OR rdf OR newsfeed irc
03:55:09 <sbp> well, rss-bot seems to work :-)
03:55:28 <sbp> but yes, I don't think this is a case of JFGI
03:55:41 <sbp> doing RSS Bot, RSS IRC Bot etc. don't give much
03:59:48 <crschmidt> yeah
03:59:48 <crschmidt> i wawsn't trying to complain
04:19:16 <Mo666> JFGI?
04:19:25 <Mo666> what's that?
04:19:49 <Mo666> java'ized fancy google imitation?
04:20:11 <bijanp> bijanp is now known as bijan
04:20:12 <Mo666> just for grat information?
04:20:44 <Mo666> oh, found it on google
04:20:46 <Mo666> hehe
04:20:50 <sbp> :-)
04:21:41 <Mo666> << german
04:22:02 <Mo666> (>> bad english)
06:19:10 <CloCkGrr> CloCkGrr is now known as CLoCkWeRX
09:34:57 <[GNU]> moin
09:42:34 <eaon> moin is sooo german :)
09:43:12 <[GNU]> ah... noddeudsch, the bavarian may not say moin :)
09:53:45 <[GNU]> [GNU] is now known as gnu][jabber
10:17:18 <dajobe> "it will be a combination of Apache2 mod_rewrite rules, PHP5, XSLT, MySQL, and the Sesame RDF engine," -- http://www.ftrain.com/InstallingUpdates.html
10:17:37 <dajobe> paul's asking for trouble there. or fun :)
10:20:37 <JimJibAct> JimJibAct is now known as JibberJim
10:46:55 <PhUrl> good morning
10:53:49 <gnu][jabber> moin
10:54:04 <PhUrl> gnu][jabber: moin moin
10:54:15 * PhUrl collects the semantic web news for the day:
10:55:25 <PhUrl> News: http://www.alwayson-network.com/comments.php?id=7889_0_5_0_C eHealthcare Focus Area at ebxmlforum.org/2004 In Review : the Artemis Project documented a Semantic Web Service-based P2P Infrastructure for the Interoperability of Medical Information systems http://healthcare.xml.org/resources/Artemis%20-%20A%20Semantic%20Web%20Service-based%20P2P%20Infrastructure%20for%20the%20Interoperability%20of%20Medical%20Information%20systems.ppt
10:56:05 <PhUrl> News : http://www.linuxjournal.com/comment/reply/8025 Geotagging Web Pages and RSS Feeds : Various semantic Web projects still are solidifying geospatial tagging standards, but several techniques already have become common and supported. This article presents these current techniques.
11:05:26 <eaon> the geotagging article is rather ... nonsensical? from my point of view. discussing syntax not data itself...
11:07:38 <PhUrl> well, it has a couple of links, i guess it is superficial.
11:13:38 <PhUrl> the link to www.a2b.cc is nice
12:23:26 <dajobe> dajobe has changed the topic to: Semantic Web Interest Group hack-n-chat - UTF-8 charset please - Weblog: http://swig.xmlhack.com/ - Logs: http://ilrt.org/discovery/chatlogs/swig/ - please identify with NickServ
13:04:07 <crschmidt> A: Bah, you beat me to it! :p
13:04:07 <dc_swig> Added comment A3.
14:21:18 <SimonW> danbri: are you still involved in web annotation research?
14:35:04 <CLoCkWeRX> CLoCkWeRX is now known as CloCkWeRX
16:06:58 <balbinus_> balbinus_ is now known as balbinus
16:31:24 <kao_home_> kao_home_ is now known as kao_home
17:13:13 <timbl> sbp?
17:13:29 <crschmidt> He's just stepped out for a bit
17:15:17 <sbp> just stepped back, as luck would have it
17:17:01 <sbp> the bulk is in n3proc.py
17:20:51 <sbp> as for scheduling a meeting, I was a little worried that attendance would zero out: the previous meeting attendance level was a bit stark, as I recall
17:30:24 <timbl> I was going to ask about upcoming scheduled topic chats for N3 syntax ... maybe to discuss lists
17:30:43 <sbp> you want to add a set syntax?
17:30:45 <timbl> n3proc.py is impressive, btw... I was just trying to find my way around it.
17:31:01 <sbp> thanks. n3p.py fires off events that n3proc.py picks up on
17:31:11 <timbl> set syntax ...yes, I have come to the conclusion that RDF in general really needs sets.
17:31:27 <sbp> based on the grammar. it's a bit of a hack since it just matches the fragID, but it wouldn't do too difficult to register the URIs properly
17:31:30 <timbl> What connects prodtcions to events?
17:32:10 <sbp> grep for on(Start|Finish|Token) in n3p.py
17:32:10 <timbl> Ok, so the name of the production is the nam,e of the python routine to call?
17:32:36 <sbp> well, by default n3p.py just prints out a kind of XML-like tree for the productions it receives
17:32:44 <sbp> but n3proc.py overrides that and uses the information as events
17:33:04 <sbp> as I say, I bind fragIDs to methods at the moment
17:33:18 <sbp> but it'd be better to do d[productionURI] = self.method
17:33:52 <timbl> Maybe .
17:34:00 <timbl> Why make it more complicted if it works?
17:34:05 <sbp> e.g.
17:34:05 <sbp> prod = self.productions.pop()
17:34:05 <sbp> handler = prod + 'Finish'
17:34:05 <sbp> if hasattr(self, handler):
17:34:05 <sbp> getattr(self, handler)()
17:34:19 <timbl> yes.
17:34:31 <sbp> yeah, that's what I thought. it'd be nice to do it properly for it being derived, but on the other hand the code is very N3-specific at the moment so I doubt anyone would want to derive from it :-)
17:34:40 <timbl> But the delightful thing is to get the whole parser down to a screenfull or two.
17:34:49 <sbp> absolutely
17:35:02 <sbp> and to separate parsing from processing
17:35:15 <timbl> Making it n3-specific .... one thought was to define whitespace, makbe comment handling also in RDF.
17:35:23 <sbp> clearly defined layers, and all RDF at the root!
17:35:35 <sbp> yeah, I remember you commenting on one of the talks about that
17:35:58 <sbp> that'd be fine, I think. the tricky part is the @keywords munging
17:36:10 <timbl> re sets ... there are lots of places wher unordered things are needed, and using ordered things for them is non-optimal.
17:36:12 <sbp> that's a seriously heavy-duty extension to the BNF
17:36:23 <sbp> agreed
17:37:05 <sbp> shame there are no complementary delimiters left
17:37:15 * dajobe oh no, not more syntax
17:37:16 <sbp> you reserved ' etc., but didn't think of pairs! :-)
17:37:20 <timbl> now http://www.w3.org/DesignIssues/N3Alternatives.html has some stuff about sets
17:37:27 * sbp looks
17:37:36 <timbl> well, I reserved ` and '
17:37:44 <sbp> dajobe: it's alright for you, you can choose to omit it from Turtle...
17:38:01 <timbl> and it has been suggested that they should be used for formulae, but they are unreliable in typography
17:38:04 <sbp> hmm. ``this'' is one of my pet i18n-fuelled peeves thouhg :-)
17:38:15 <sbp> s/uhg/ugh/
17:38:34 <timbl> The idea vis that if you declare something as being owl:oneOf a list, then it is the unordered class of those things which are in the list.
17:39:28 <timbl>http://www.w3.org/DesignIssues/N3Alternatives.html#Sets
17:39:28 <dc_swig> B: http://www.w3.org/DesignIssues/N3Alternatives.html#Sets from timbl
17:39:28 <sbp> "{$ 1, 2, 3 $}" isn't too bad, under the circumstances
17:39:38 <timbl> That's what I thought.
17:40:05 <sbp> "@Set{1, 2, 3}"?
17:40:26 <timbl> looks like Set([1,2,3]) a little.
17:40:30 <sbp> too verbose probably, but yeah
17:41:14 <sbp> nice that it's being mapped onto OWL/rdf:List too--means you can round trip to and fro' RDF/XML
17:41:41 <sbp> rather than introducing another core RDF Term
17:41:52 <sbp> which is what I was assuming up until now you'd do
17:42:21 <sbp> perhaps it'd be better modelled like Seq though
17:42:36 <timbl> I'm thinking of the rdf:first stuff as being jsut reification syntax for lists.
17:42:37 <sbp> [ a :Set; :member "p"; :member "q"; :member "r" ]
17:42:51 <sbp> right, but sets are... sets, not lists :-)
17:42:58 <dajobe> [ are you going to schedule a talk sometime. I've got to run now, some advance warning, pls? ]
17:43:04 <sbp> OTOH, then you don't get the OWL semantics
17:43:28 <sbp> dajobe: will a CC on the announcement do?
17:43:44 <dajobe> I'm on cwm-talk
17:43:53 <sbp> excellent. it'll definitely go there
17:44:34 <timbl> Yes, do organize another sheduled topic chat.
17:45:14 <sbp> "Weird and unconventional to start with a comma"--bit RPNish
17:45:15 <timbl> Sean, have you tried outputting test results in Yosi's ontology fo n3 parser tests etc?
17:45:53 <timbl> (or even making a test manifest from your tests?)
17:45:56 <sbp> ontology: you mean the one for reifying formulae so that you can compare to reference N-Triples output?
17:46:27 <timbl> No ... ontology of "this implementation had this result on this test"
17:46:45 <sbp> the filenames are the manifest: you just need os:listdir, str:scrape, and a few other bits... :-)
17:46:52 <sbp> ah, then nope
17:46:57 <dajobe> hmm, I do have some time now. Maybe I'll add """quotes""" to turtle.
17:47:00 <sbp> we did chat about the flattening ontology
17:47:13 <sbp> URI for the test ontology?
17:47:34 <timbl> Yosi's http://www.w3.org/2000/10/swap/test/n3/test_results.html is generated from RTDF. I think, yosi?
17:48:01 <sbp> ah: http://www.w3.org/2000/10/swap/test/n3/Makefile
17:48:07 <sbp> from test_results.n3
17:49:17 <sbp> (and hence -> http://www.w3.org/2004/11/n3test#)
17:49:27 <sbp> (and Not Found)
17:49:33 <timbl> The namespace is http://www.w3.org/2004/11/n3test but there isn't (yet) am ontology there, it seems. But anyway, ... the OWL tests were doine by people generating test reports in RDF, and SDandro merged them to make a table.
17:49:57 <timbl> so it seems to me that it would be good to do the same for N3.
17:50:12 <sbp> I vaguely remember that too... I was hacking on an OWL Syntax Checker some time ago
17:50:15 <sbp> agreed
17:50:35 <timbl> There is an ontology for the N3 tests ... like for RDF tests.
17:51:55 <sbp> heh: n3test:N3Parser
17:52:19 <sbp> looks straightforward
17:52:47 <timbl> ... like http://www.w3.org/2000/10/swap/test/n3parser.tests
17:52:53 <sbp> interoperability between test suites means we should be able to pool them then too
17:53:07 <sbp> heh: "# yuk - under protest - no hash"
17:53:32 <sbp> right. reminds me of RDF Test's MANIFEST.rdf, as you say
17:53:44 <crschmidt> that NS isn't even used in the document
17:53:53 <timbl> that?
17:53:56 <crschmidt> is there a reason for the prefix to be defined that I'm missing?
17:54:08 <crschmidt> "@prefix dc: <http://purl.org/dc/elements/1.1/>. # yuk - under protest - no hash"
17:54:17 <sbp> probably a copy-'n'-paste job
17:54:24 <sbp> .gcount "yuk - under protest - no hash"
17:54:27 <phenny> "yuk - under protest - no hash": 68
17:54:31 <crschmidt> heh
17:54:33 <timbl> you are right ... lots fo files have some dc:author stuff
17:54:36 <crschmidt> Probably so
17:56:39 <timbl> I think n3proc.py is nifty. What is your opinion of te technology? Do you think it would be good to put it in cwm as a default parser?
17:57:30 <timbl> (Or does loading the whole pickled branch table take too long?)
17:57:31 <sbp> there's still quite a reliance on custom code, so it's best not to be too complacent that n3.n3 is a cure-all; but on the other hand, it's definitely a step in the right direction w.r.t. maintainability
17:57:40 <sbp> oh, no, the branch table is zippy
17:57:52 <sbp> and I think I found that the written module is faster than pickle
17:58:05 <sbp> Python'll make a *.pyc out of it which is marshaled out. much faster
17:58:15 <sbp> and probably still faster even if not
17:59:22 <sbp> I've been waiting for a more formalised notation3.py since DanC started hacking on the YAPPS and Spark stuff, so I'd definitely welcome it :-)
17:59:26 * timbl lost ... even if not ...?
17:59:53 <sbp> even if Python doesn't make .pyc files, the .py files are probably faster-loading than pickles
18:00:11 <sbp> sorry for the ambiguity
18:00:32 * bengee wonders if SPARQL is going to consider/support something like "pattern extensions" for quad/5-tuple/6-tuple stores or if it is limited to triple stores (plus the hard-coded SOURCE keyword).
18:00:54 <sbp> I'd be happy if it just had some nice builtins for list handling...
18:01:29 <bengee> or some reification shortcuts..
18:01:33 <timbl> On the subject of making th etokenizer more general: One could build things like the list of double-character tokens from the grammar itself.
18:01:35 <sbp> SPARQL needs shredding. viva la N3QL
18:01:52 <timbl> And the whitespace and notNameChar epressions could be stored in the grammar file.
18:01:59 <sbp> ah, that's a point
18:02:14 <sbp> though again, @keywords is the biggest hurdle I think
18:02:21 <timbl> Yes.
18:02:25 <sbp> doesn't seem much point in separating the other parts out if you can't do that too
18:02:32 <timbl> Is it @keywords worth the pain?
18:02:54 <sbp> hmm. hard to tell really
18:03:12 <sbp> I've been authoring lots of little test cases, obviously, so I'd say it's be worth it for me :-)
18:03:18 <timbl> Or, you say we will use @@ for @keywords, and make the keyword process totally a preprocessopr.
18:03:22 <sbp> others may disagree though. we could talk about it in the chat
18:03:39 <sbp> ah, pre-processing is an interesting option
18:03:45 <sbp> then it leaves the door open for other macros
18:03:57 <timbl> Yosi played with preprocessing.
18:04:07 <sbp> anything on the Web?
18:04:31 <timbl> I don't know. And he doesn't seem to be around at the moment.
18:04:36 <sbp> aye
18:04:54 <sbp> can defer that to the chat also, at any rate
18:05:35 <sbp> sometime within the next week or two for the chat?
18:05:38 <timbl> Ummm. what worked last time?
18:05:43 * timbl looks
18:05:44 <sbp> I should probably line it... right
18:05:52 <sbp> Monday 18:30Z or something
18:06:22 * DanC tunes in, catches up... mostly hacking on rdfizing my procmail rules and logs
18:06:32 <sbp> yep. first was 19:00Z, second was 18:30Z
18:06:41 <sbp> hi DanC
18:07:10 <sbp> 2005-01-17T18:30Z would be fine for me
18:07:12 <timbl> 2004/12/13 13:30ET was last time/
18:07:24 <sbp> yep
18:09:10 <timbl> 18:30Z is I guess 18:30 London, 13:30 ET... DanC and I would have competing meeting after 90 mins
18:09:15 <timbl> But that is generally OK
18:09:47 <sbp> would 18:00Z or earlier be preferable then?
18:09:57 <sbp> we ran to two and a half hours on the first meeting...
18:10:32 <sbp> though how you could think the TAG competes with N3, I don't know! :-)
18:10:38 <DanC> 17th Jan... (a) quite soon, (b) conflicts with http://www.w3.org/2005/01dc-hel/
18:10:58 <DanC> 1 week notice is traditional
18:11:04 <timbl> for me, I can do any time after 11:00 ET, 16:00 Z
18:11:18 <sbp> yeah. it just appeared that timbl was itching for it to happen
18:11:21 <timbl> So 1/24 maybe, DanC?
18:11:47 <sbp> and about 17:00Z? that'll give ample clearance
18:12:00 <DanC> 24 Jan... still a monday (ugh... monday meetings suck) but I'm in office mode
18:12:15 <timbl> Well, sbp was complaining that he "had no control over n3.n3" in earlier logs, when he doesn-- by chairing a topic chat -- have wuite a lot!
18:12:29 <sbp> we're only sticking to Mondays because the previous two meetings where held then and turned out to be fairly successful
18:12:42 <sbp> heh. DanC should be the chair though
18:12:42 * DanC looks up 1700Z on 24 Jan in Chicago time...
18:12:46 <sbp> I'm just verbose
18:12:57 <timbl> Monday meetings: Maximal energy?
18:13:11 <DanC> minimal prep
18:13:20 <sbp> presumably 11:00 CST
18:13:52 <DanC> 11:00 AM Chicago per http://www.timeanddate.com/worldclock/fixedtime.html?month=1&day=24&year=2005&hour=17&min=0&sec=0&p1=0
18:14:11 <DanC> can do
18:14:24 <sbp> you could take Friday off and prep on Sunday!
18:14:28 <DanC> logger, pointer?
18:14:28 <DanC> See http://www.ilrt.bris.ac.uk/discovery/chatlogs/swig/2005-01-12#T18-14-28
18:14:53 <sbp> actually, sabbath. forget that
18:15:03 <DanC> "cwm/N3 ScheduledTopicChat Monday, January 24, 2005" -- http://calendar.sidekick.dngr.com/event?id=3284&date=2005-01-24
18:15:42 <timbl> Looks good .. tentative oro firm?
18:15:45 <DanC> (not a very cool URI... cookie-specific... hmm... we need a tag issue on my.yahoo.com , since it comes up in ws-addressing)
18:15:50 <sbp> DanC: ooh. pointer to information about that?
18:15:50 <dajobe> I can't make that, but I'm not critical.
18:16:08 <timbl> darn
18:16:17 <sbp> dajobe: what times are open for you?
18:16:18 <DanC> pls counter-propose, dajobe? the meeting's much more valuable to me if you can attend
18:17:00 <timbl> indeed
18:17:01 <DanC> pointer to info about which, sbp?
18:17:02 <dajobe> not monday 24th 1600-1800Z
18:17:25 <sbp> about the sidekick URI. asks me to login, but I presume it's related to your ever growing army of wearable gizmos
18:17:30 * DanC prefers wednesdays for #swig events
18:17:41 <dajobe> weds ar good for me
18:17:55 <timbl> 26th?
18:18:00 <dajobe> fine
18:18:05 <DanC> yes, I'm the only person authorized to read that page, sbp
18:18:14 <sbp> I'd prefer 19th, but 26th is alright
18:18:31 <sbp> can never tell how much work I'll be lumbered with past a week
18:18:34 <dajobe> some of us'll be in HEL
18:18:37 <timbl> 19th the DAWG f2f is on , may affect some intersted parties
18:18:52 <sbp> ah, okay
18:19:00 <DanC> Wednesday, January 26, 2005 at 17:00:00 UTC http://www.timeanddate.com/worldclock/fixedtime.html?month=1&day=26&year=2005&hour=17&min=0&sec=0&p1=0
18:19:02 * sbp tries to mentally decode HEL...
18:19:17 * tlr waves finnish flag at sbp
18:19:18 <dajobe> wfm
18:19:23 <DanC> HEL refers to an airport in Helsinki, Finland
18:19:23 <timbl> Helsinki Finland ... you are there on the 26th, DaJoBe?
18:19:25 <sbp> aha
18:19:38 <dajobe> yea
18:19:39 <sbp> who'd've know that HEL would be in Finland...
18:19:39 <dajobe> h
18:19:44 <sbp> s/know/known/
18:20:01 <timbl> Not for the DAWG f2f?
18:20:08 * dajobe answers the right question. 26th 1700 is good for me
18:20:20 <sbp> 2005-01-26T17:00Z in stone then
18:20:41 <DanC> phpht. I'm holding 2005-01-26T17:00Z for something else.
18:21:00 <DanC> so I'm at risk at best
18:21:20 <sbp> shouldn't we have all our schedule information in RDF by now? can't we just throw it into CWM and have it spit out a time at us? :-)
18:21:25 <timbl> An hour later help, danc?
18:21:35 <dajobe> anytime that wed ok for me
18:21:58 <DanC> yes, sbp, 5000 points for whenIsEverybodyFree.n3
18:22:06 * sbp still remembers DanC and em's poolGame.n3
18:22:12 <dajobe> sbp managed to pick the only 2 hours that week I had a meeting
18:22:39 <crschmidt> what would be the best format for that? is ical the only standard for RDF calendaring data?
18:23:49 * timbl has his calendar data in RDF, but it i snot that simple. DOesn't have m,eetings marked with star-rating as to how easy they are to move
18:23:50 <sbp> "@forAll :lawyer, :person . { <universe:the> log:notContains { :lawyer a :Lawyer }; log:contains { :person a :Person } } => { ?Person a :Free } ." - whenIsEverybodyFree.n3
18:24:04 <DanC> yes, but ical can't express all the foafy stuff you'd need to prove you're authorized to read my calendar
18:24:45 <crschmidt> DanC: are the times you're available sensitive? or just the actual activities?
18:24:48 <timbl> I could make you a calendar with free-busy info only (ie all summary, description and URL zapped). B
18:25:22 * crschmidt has never been a very private person, publishes any available appointment info in RDF
18:25:23 <DanC> good question, crschmidt
18:25:32 <crschmidt> however, since I'm eternally disorganized, that's usually rdf:nil ;)
18:26:01 <DanC> but as timbl mentioned, free/busy is really not just one bit. free for _what_?
18:26:25 <DanC> I could give you a 1-bit masked schedule, but I'd be busy almost all the time.
18:27:09 <crschmidt> hm, that's a point: for example, currently I'm working on a website: I can't go out to lunch with you, but I may be able to participate in an IRC chat
18:27:27 <dajobe> irony - this is Wed, at the time scheduled and we've been talking about scheduling for 20mins...
18:28:13 <sbp> it may be too complex to solve entirely computationally, but the knack is in getting it useful
18:28:38 <sbp> it's going to be harder to get an answer the less essential the thing you're trying to schedule
18:28:49 <crschmidt> heh
18:28:52 <dajobe> later
18:29:29 <yosi_> clearly, we need to schedule a chat for scheduling the next chat
18:29:36 <sbp> agreed
18:29:40 * crschmidt agrees
18:29:41 <sbp> but when?
18:30:39 <yosi_> maybe we can set up a committee for planning a series of meetings for choosing committe members for scheduling the scheduling chat
18:31:05 <sbp> scheduling meetings all the way down
18:31:09 * DanC is counting on sbp to wisely choose a time
18:31:21 <sbp> fool :-)
18:31:37 <sbp> python -c "import random..."
18:32:16 <sbp> I'll do my best. 16th was okay except for DanC risk, right?
18:32:22 <sbp> er, 26th
18:32:33 <sbp> so maybe a day either side... though I risk putting everyone out then
18:32:34 <DanC> right, and an hour earlier or later does help me
18:32:45 <PhUrl> PhUrl is now known as md-afk
18:33:16 <sbp> I should plot a graph of the cumulative probabilities of everyone attending over the next two weeks and then choose the time that corresponds to the highest peak; though I'll have to build in error margins and...
18:34:10 <yosi_> also rate the relative importance of the different people; if you can't show up, but everyone else can ...
18:34:20 <sbp> ah, good point
18:35:29 * sbp will need to write to Tufte just to figure out how to present it
18:36:26 * timbl goes looking for food
18:36:37 <DanC> sbp, I have a simpler schedulding protocol. maybe I can teach it to you.
18:36:43 <yosi_> timbl, just ask zakim
18:37:11 <DanC> it goes like this: (a) whoever wants the meeting to happen owes one (or 2 or 3) specific proposals for when to meet, down to the minute, including duration
18:37:28 <JimJibber> JimJibber is now known as JibberJimm
18:37:30 <JibberJimm> JibberJimm is now known as JibberJim
18:37:59 <DanC> critical resources owe either "yes i can do that time" or "here's my counterproposal"
18:38:21 <DanC> non-critical resources have 2 more choices: regrets, and not responding
18:40:14 <DanC> if there's only one proposal outstanding, and all the critical resources are available, the proposer may declare that the meeting time, and should do so, if there's support from lots.
18:41:20 <sbp> and if not?
18:41:53 <edd> for resource being wn:Person not rdf:Resource? :)
18:41:57 <DanC> hmm... let's say each proposer must order their proposals; they can declare a lesser proposal to be the meeting time only if they justify why their greater proposals wouldn't work.
18:42:09 * edd has a bee in his bonnet about that usage
18:42:20 <DanC> no, rdf:Resource. teleconference bridges count too. as do IRC channels.
18:42:33 <Arnia> edd: About wordnet?
18:42:39 <edd> But they can't send regrets? :)
18:42:57 <DanC> they may, per this protocol
18:43:24 <sbp> words may have more than one sense; they're not URIs!
18:43:25 <DanC> if not, nothing happens
18:44:00 <sbp> so this is a kind of weighted democracy system?
18:44:25 <DanC> so... 2005-01-26T17 was proposed... by whom? and am I a critical resource? let's say yes
18:44:38 <DanC> it's not very democratic, no
18:45:06 <DanC> whether somebody's a critical resource or not is not formalized
18:45:12 <Arnia> sbp: Yes, which is my major issue with using Wordnet as a lexicon. You can, however, IMO use a synset-marked version as an easy-to-grok conceptual system as long as you don't forget that the URIs aren't words, they're conceptual descriptions... but that's all my own opinion and most will probably scoff :)
18:45:44 * DanC concludes that sbp proposed 2005-01-26T17
18:46:10 <DanC> pop quiz, sbp: who owes a next move?
18:48:18 <sbp> sorry, watched Deep Impact launching
18:48:42 <sbp> okay, the required resources owe the next move
18:48:43 <LotR> DanC: the #swig channel :)
18:49:05 <LotR> (assuming the meeting is to be here)
18:49:41 * DanC presumes the #swig channel is available, though that's and interesting issue...
18:49:49 <sbp> Arnia: well WN tried to disambiguate the senses to provide them with URIs
18:49:59 <DanC> and the required/critical resources are...?
18:50:07 <sbp> seems that timbl actually proposed the 26th. not sure if you meant this hypothetically or not though
18:50:35 <DanC> ok, I can stipulate that timbl proposed 26th
18:50:36 <sbp> as you say, it's hard to formalise. but timbl since he called the meeting, at the least
18:50:50 <DanC> the meeting isn't called yet, in my protocol
18:51:02 <DanC> the meeting call should go to public-cwm-talk
18:51:11 <sbp> metacall then. anyway, but that'd rule him out as a required resource
18:51:17 <LotR> DanC: is your protocol compatible with 2446?
18:51:36 <DanC> .g 2446
18:51:45 <phenny> 2446: http://www.ietf.org/rfc/rfc2446.txt
18:51:46 * DanC has no ideer
18:52:12 <LotR> that's the scheduling rfc :)
18:52:13 <DanC> I don't mean it as hypothetical, if this training excercise works
18:52:36 <sbp> required resources: I don't think it'd be much worth it if you, dajobe, and yosi weren't there
18:52:49 * DanC didn't think timbl proposed a time down to the minute
18:53:37 <sbp> but then I'm not sure if that's better as an OR or an AND. I don't think you allowed OR anyway
18:53:42 <DanC> right... and which of those critical resources has not yet responded re 2005-01-26T17:00:00Z?
18:53:45 <LotR> in iCalendar, times are always down to the second..
18:54:05 <sbp> you responded, dajobe IIRC responded, hence... yosi!
18:54:37 <DanC> my protocol is for meetings of 30 to 120 minutes, where being one second late is not cause for apology.
18:54:52 <sbp> you responded with a negative though, so per '<DanC> critical resources owe either "yes i can do that time" or "here's my counterproposal"', you owe a counterproposal
18:54:59 <DanC> bingo!
18:55:17 * sbp wipes sweat from brow
18:55:58 <DanC> now if you're being generous, my "either an hour earlier or later helps" was 2 counter-proposals. per this protocol, as chair-nominee, you might nudge me to order them.
18:56:20 <sbp> (this reminds me a bit of Lewis Carroll's logic problems)
18:56:58 <sbp> whoops, I read your reply in context of mine as being a day either side:
18:56:58 <sbp> [18:32] <sbp> so maybe a day either side... though I risk putting everyone out then
18:56:59 <sbp> [18:32] <DanC> right, and an hour earlier or later does help me
18:57:17 <DanC> I prefer an hour later... 2005-01-26T18:00:00Z
18:57:35 <sbp> hour's interesting. timbl already noted that he could do 16Z onwards, though that may've just been for Mondays
18:58:06 * sbp looks back at the scheduling algorithm to find what now...
18:58:51 <timbl> By the way, I have a few more helpful hints. One side will make it get earlier ...
18:58:52 <sbp> it appears to stop at this point
18:59:13 <timbl> ..and the other side will make it grow later.
19:00:14 <DanC> it may stop at this point, sbp, but either all critical resource are available at that time, in which case the proposer (timbl? sbp?) SHOULD declare the meeting time, or a critical resource owes a counterproposal
19:00:33 <LotR> sbp: surely the organizer would want to send out an updated proposal?
19:00:42 <DanC> oops; I'm the proposer now.
19:01:00 <DanC> (this is easier when there's a recognized chair, who plays the role of proposer)
19:01:29 <DanC> LotR, my protocol works before all parties agree who's organizing the meeting
19:01:31 <tlr> tlr is now known as tlr-off
19:01:41 <tlr-off> tlr-off is now known as tlr
19:01:50 <sbp> the problem is finding out whether everybody's available for the counterproposal
19:02:04 <sbp> since yosi and dajobe are in and out
19:02:15 <DanC> timbl? dajobe, yosi, 2005-01-26T18:00:00Z ok or not?
19:02:29 * timbl looks
19:02:44 <sbp> normally you'd say that given its closeness to the original proposal it should be fine, but that hour's difference made it for you
19:02:47 <LotR> DanC: by organizer, I meant the ical property, which has little to do with who is chairing the meeting
19:02:51 <dajobe> DanC: ok
19:03:03 <DanC> s/plays the role of proposer/shepherds all proposals/
19:03:29 <timbl> Yes, 2005-01-26T18:00:00Z for 90mins is good.
19:03:32 <DanC> yes, but event the ical organizer role isn't clear for this meeting
19:03:36 <sbp> we should've just held the meeting now
19:03:55 <DanC> well, we did hold a meeting. on chair training. ;-)
19:04:03 <timbl> Yes, 2005-01-26T16:00:00Z for 4 hrs is free.
19:04:31 <LotR> DanC: wouldn't that be the person who first mentioned the idea of having a meeting?
19:04:45 <DanC> so nobody can act unless that person's around?
19:05:04 <LotR> oh, I see
19:05:20 <timbl> I think sbp is the organizer, or should i say, I would like sbp to organize the meeting.
19:05:26 * DanC too
19:05:34 <timbl> sbp did a good job before.
19:06:05 <DanC> timbl and danc have delegated 'proposer' rights to sbp
19:06:18 <LotR> well, there's delegated-from and delegated-to property parameters. not sure how exactly one would use them here
19:06:40 * sbp delegates them back to timbl and DanC in one swift pelota-like movement...
19:06:46 <DanC> sbp, I think dajobe is free for 2005-01-26T18:00:00Z for 90mins , but I can't quite remember/confirm; can you?
19:06:51 <timbl> (ie he arranged a time and place su=ccessfully, chaired it to some comclusions, and then implemented the results in python. Can't beatt that!) ;-)
19:07:26 <DanC> oh. you can decline the nomination, sbp. I'd be mildly disappointed, but I'd understand. but I thought you were interested.
19:07:47 <sbp> it was pretty much arranged in the meeting the first time; I'm surprised we didn't schedule a third meeting in the second actually
19:08:03 <sbp> no I'll accept. I just wanted to mention pelota really
19:08:07 <DanC> you learned chairing lesson #2343 then ;-)
19:08:17 <jsled> .w pelota
19:08:20 <phenny> pelota 1. a Basque or Spanish game played in a court with a ball and a wickerwork racket
19:08:42 <DanC> lesson #2343 = "don't conclude meeting N without arranging meeting N+1". That's actually one of Roberts Rules, I gather (via ndw)
19:09:13 <sbp> though that meeting was officially community chaired, I believe, from the announcement
19:09:42 <sbp> and you chaired the first, so you must've delegated that responsibility out, or lost it through neglect
19:10:00 <DanC> I nominated you, I think.
19:10:30 <DanC> <dajobe> sbp managed to pick the only 2 hours that week I had a meeting
19:10:50 <DanC> on that basis, I conclude dajobe's available for 90min starting 2005-01-26T18:00:00Z
19:10:57 <dajobe> yeah, I am
19:10:58 <sbp> if only I were as good on lottery numbers...
19:11:09 <DanC> shall we take the risk that yosi's free and be done, sbp?
19:11:13 <timbl> Ah! dajobe.free = ~meeting.time.
19:11:27 <sbp> seems a good heuristic
19:11:58 <sbp> DanC: not sure about that risk since before you got here timbl and I discussed a few points that we needed to ask yosi about "in the meeting"
19:11:59 * DanC would be please if sbp would declare the meeting time
19:12:25 <sbp> on the other hand, he's a minion and that's in working hours, so...
19:12:34 <DanC> in fact, I'd be even more pleased if you'd call the meeting by sending mail to public-cwm-talk
19:12:43 <LotR> sounds like the channel needs a bot that keeps track of people's free/busy data :)
19:12:43 <sbp> [18:20] <sbp> 2005-01-26T17:00Z in stone then
19:12:48 <sbp> s/17/18/
19:13:07 <sbp> and I'll subsequently attend to that matter
19:13:10 * DanC updates http://calendar.sidekick.dngr.com/event?id=3284&date=2005-01-26 to 1800Z
19:13:37 <DanC> hmm... IOU one wiki topic on that protocol
19:13:39 <DanC> name?
19:13:49 <sbp> HowToBefuddleSbp
19:14:06 <sbp> SchedulingAlgorithm
19:14:07 <DanC> seen http://esw.w3.org/topic/MidwestWeeklyAgenda , sbp?
19:14:15 <sbp> nope. checking
19:14:35 <DanC> SchedulingAlgorithm is too generic... MeetingFromNothing is kinda generic too...
19:14:56 <DanC> TeleconferenceTiddlywinks! i named it long ago.
19:14:57 <sbp> why is it called the "midwest agenda pattern"?
19:15:11 <DanC> didn't I explain? darn...
19:15:23 <LotR> http://www.w3.org/2000/10/swap/doc/cwm.html still uses #rdfig
19:15:23 <dc_swig> C: http://www.w3.org/2000/10/swap/doc/cwm.html from LotR
19:15:30 <LotR> ugh
19:15:31 <DanC> ... because I live in the midwest, as does Paul Grosso, who heavily influenced my style
19:15:59 <sbp> I see. superset of GrossoConnolly
19:16:10 <yosi_> sbp, re preprocessing, what I ran into is that it's not what is done with n3 right now
19:16:31 <DanC> yes... it's a style I offer to the community, which is welcome to tune it, within the bounds of what would be recognized as good taste here in the midwest
19:16:40 <sbp> heh, heh
19:16:47 <sbp> yosi_: pardon?
19:17:02 <yosi_> as in, what happens if a keyword is sitting in the middle of an explicit URI?
19:17:14 <sbp> ah, right. you need to implement a parser in the preprocessor
19:17:21 <sbp> ever look at N3S?
19:17:35 <sbp> .g Notation3 N3S
19:17:38 <phenny> Notation3 N3S: http://infomesh.net/2002/n3s/
19:18:19 <sbp> a preprocessor that introduced some current N3 features before they were implemented in CWM, e.g. =>, prefix defaulting, etc., and some that never made it too
19:18:28 <sbp> generally, it's just a tokeniser
19:18:32 <yosi_> I've glanced at it. Nothing more
19:18:36 <sbp> then it munges the token list
19:18:52 * timbl fixes link from #rdfig to #swig in cwm.html
19:19:08 <sbp> thankfully it's a lot easier to write a tokeniser than a parser, so you can base the preprocessor on the former I think
19:19:22 <sbp> all you need is a good up-to-date list of tokens, which I guess you can get from n3.n3
19:19:53 <sbp> DanC: thanks for the coaching, btw
19:19:53 * timbl thanks Lotr
19:20:16 <DanC> most welcome, sbp
19:21:09 * DanC considers BLURB: Teleconference Tiddlywinks, a chair training episode
19:21:49 <sbp> go for it
19:22:19 <sbp> plenty of references that we've collected
19:22:21 <yosi_> sbp, the problem I ran into with building a stuping tokenizer is that @ has two meanings
19:22:26 <sbp> would be nice to have them in one place
19:22:34 <timbl> So, DanC, have you found an ontology in which you can describe the game, and in which you can also describe last call comment handling?
19:22:40 <yosi_> s/stuping/stupid/
19:22:52 <DanC> no... working on it, timbl
19:23:07 <timbl> (@ occurs in URIs and also introduces language codes)
19:23:45 <sbp> ah, that's a separate regexp. I see
19:24:09 <sbp> that's unfortunate. you could amalgamate it and then pass it off to the processor to deal with
19:24:35 <sbp> alternatively, you can preserve whitespace as a token
19:25:02 <sbp> then check for its presence between <URI> and @token
19:25:18 <sbp> which is likely preferable
19:25:48 <sbp> I've been trying to work out a way to read N3 as buffered input
19:26:08 <sbp> it's tricky-to-impossible because Python won't return tentative regexp match statuses
19:26:18 <sbp> and nor will any regexp library that I'm aware of
19:26:36 <sbp> on the other hand, only triple quoted strings can cross line boundaries
19:26:49 <sbp> so you can readline and special case triple quotes
19:27:17 <sbp> but thereagain you may have extraordinarily long lines, so it's not perfect. you really want to be able to read in bufsize bytes
19:28:02 * sbp isn't sure why he digressed into that
19:29:55 <sbp> oh right: I was thinking that you'd be able to efficiently handle the backwards and forwards looking on the token list using itertools, but then I thought that that's silly since it'd only be efficient if you had buffered read of Notation3 files in the first place
19:29:57 <timbl> "only triple quoted strings can cross line boundaries" --- I'd note the original intent of <> around URLs was to allow you to break them onto several lines
19:30:11 <sbp> ah, interesting
19:30:38 <timbl> So one probably *should* ignore all whitespace between the <>
19:30:58 <timbl> Though currently cwm outputs URLs with spaces in between <>
19:32:05 <sbp> incidentally, I found out the other day that a space is a lexically valid character in xsd:anyURI
19:33:19 <sbp> should you want to make infoset versions of N3 parse trees... :-)
19:34:18 <sbp> URI = element n3:URI { "<", xsd:anyURI, ">" }
19:35:23 <sbp> interesting how close RELAX NG Compact (and CSS, and XQuery) are to N3
19:39:26 <sbp> wandering off; will send email to cwm-talk shortlyish
19:39:58 <yosi_> I just checked in to cwm cvs initial support for n3 sets
19:40:30 <yosi_> currently ($ ... ) makes a set. This will change when somebody tells me otherwise
19:40:55 <timbl> Do I have to cvs update -d anything to get new bits of tree, like test/sets ?
19:41:04 <timbl> ($ .... $)
19:41:07 <yosi_> not that there is anything that can be done with them at the moment
19:41:16 <yosi_> timbl, that would be harder
19:41:16 <timbl> Any tests?
19:41:21 <yosi_> no tests yet
19:41:25 <timbl> harder?
19:41:38 <yosi_> it was hard enough getting existing tests to run
19:41:59 <yosi_> adding parser support for ($ took modifying 3 lines
19:42:17 <timbl> But ($ .... $) would be more
19:42:19 <timbl> ?
19:42:40 <yosi_> requiring a closing $ would require more, yes
19:42:52 <yosi_> it's stuck onto the list code
19:43:08 <yosi_> if there's a $ as the first character of the list, this is really a set
19:43:10 <timbl> Ah , also Ithink commas are separate.
19:43:23 <timbl> Ah , also Ithink commas are good idea too.
19:43:49 <yosi_> there's not much to test right now
19:43:53 <timbl> No, hang on ... current best syntax is {$ 1, 2, 3 $}
19:44:54 <yosi_> when we have a consensus, I'll change it. For now, I went with what was easiest to implement
19:45:22 <yosi_> it doesn't output that anyway. It does [owl:oneOf (...)]
19:46:17 <yosi_> hmmm.... {$$} is the empty set, then?
19:46:33 <timbl> yes . http://www.w3.org/2000/10/swap/test/sets.n3
19:47:33 <yosi_> {$ 1 $} <#linksEqualLists> ( 1 $} .
19:47:56 <yosi_> that looks wrong
19:47:58 <timbl> oops
19:48:09 * timbl fixes
19:49:18 * timbl adds sets of lists
19:50:20 <yosi_> :twoFormulae = ($ {:a a :mother} {:b a :father} $}.
19:50:52 <timbl> oops again...
19:51:13 * timbl wonders ho wthey got though global change ... sigh
19:54:13 <sh1mmer> evening
21:08:24 <bblfish> can this be rdf?
21:08:26 <bblfish> <generator uri="https://bloged.dev.java.net/" version="0.5">BlogEd</generator>
21:09:02 <crschmidt> not rdf:xml: can't have un-namespaced attributes. Unless I'm wrong, which I might be.
21:09:29 <bblfish> no that's only deprecated, as I understand deprecated means that you can still use it...
21:11:07 <crschmidt> rapper: Error - URI file:///home/crschmidt/foo.rdf:3 - Using an attribute 'uri' without a namespace is forbidden.rapper: Error - URI file:///home/crschmidt/foo.rdf:3 - Using an attribute 'version' without a namespace is forbidden.
21:11:10 <eaon> bblfish: as far as i'm aware, not this can't be valid rdf/xml
21:11:10 <dviner> wouldn't you _want_ namespaced attributes ?
21:11:26 <crschmidt> dviner: he's trying to convert Atom to RDF with minimal effort, I believe
21:11:44 <bblfish> crshmidt: that's it
21:11:59 <dviner> ahh... couldn't you wrap it in a rdf:RDF element, then assign xmlns="http://my.atom.com" ?
21:12:18 <crschmidt> dviner: that's what i did
21:12:19 <dviner> then all elements w/out namespace should be assumed to be in the xmlns namespace.
21:12:21 <bblfish> the namespace thing is not a problem, I could have added those to the example
21:12:22 <crschmidt> rapper gives the error above
21:12:37 <kpreid> bblfish: even if it were namespaced, the only place you can have text is inside a literal-property element, but the attribute would represent a property of the literal, which isn't allowed in RDF/XML
21:12:40 <dviner> oh... i must have missed the full xml then.
21:12:44 <bblfish> <atom:generator atom:uri="https://bloged.dev.java.net/" atom:version="0.5">BlogEd</generator>
21:13:02 <crschmidt> oh, yeah
21:13:06 <crschmidt> also what kpreid said
21:14:32 <bblfish> that's the reference: http://groups-beta.google.com/group/atom-owl/browse_thread/thread/5e3d71951c1b0957/da8a69c6f9d1f85c?_done=%2Fgroup%2Fatom-owl%2F%3F&_doneTitle=Back+to+topics&_doneTitle=Back&&d#da8a69c6f9d1f85c
21:14:49 * bblfish sigh! can't they make shorter links
21:15:35 <bblfish> the question is really can one have anything that is of the form
21:15:37 <bblfish> <name att=value att2=value>text</name>
21:15:38 <bblfish> in rdf/xml
21:15:41 <crschmidt> no
21:16:07 <dviner> according to the w3C validator service, this is ok:
21:16:08 <dviner> <?xml version="1.0"?>
21:16:08 <dviner> <rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
21:16:08 <dviner> xmlns="http://www.xmlns.com/atom">
21:16:08 <dviner> <generator
21:16:08 <dviner> uri="https://bloged.dev.java.net/"
21:16:10 <dviner> version="0.5">BlogEd</generator>
21:16:12 <dviner> </rdf:RDF>
21:16:26 <dviner> with this warning: Attribute: uri. Unqualified use is deprecated. Assuming namespace: http://www.xmlns.com/atom
21:16:45 <bblfish> oh that's what I meant
21:16:55 <bblfish> I should have checked there first...
21:18:21 <kpreid> huh. that would be text inside a description
21:18:28 <kpreid> I have no idea what that means.
21:20:29 <Arnia> Does it have a meaning?
21:20:32 <dajobe> the xml namespaces spec doesn't allow that warning's assumption. it's one particular interpretation, and not portable
21:20:33 <dviner> kpreid: i'm not sure i understand your last comment... what don't you understand? the error message? or my sample?
21:20:34 <bblfish> that makes a graph on the validator.
21:21:03 <Arnia> bblfish: Where does the literal appear in the graph?
21:21:06 <bblfish> I think the problem I am having may be more to do with context
21:21:26 <kpreid> dviner: what I don't understand is why the validator accepts that.
21:21:29 <dajobe> don't take the validator generating something when there's an ambiguity meaning that all rdf tools must generate the same thing
21:21:29 <bblfish> yes... I have to check what encloses that code...
21:21:42 <kpreid> Or rather, my current hypothesis is that the validator should be rejecting it instead.
21:21:49 <bblfish> let me look at the atom syntax spec....
21:22:59 <dviner> kpreid: you hypothesize the validator should reject because of the namespace warning? i.e., it should be a fatal error? or because the "blogEd" has no apparent semantic meaning?
21:24:15 <kpreid> the latter
21:24:17 <bblfish> <feed>
21:24:19 <bblfish> <head>
21:24:20 <bblfish> <title>...</title>
21:24:22 <bblfish> <atom:generator atom:uri="https://bloged.dev.java.net/" atom:version="0.5">BlogEd</generator>
21:24:23 <bblfish> </head>
21:24:25 <bblfish> <entry>
21:24:26 <bblfish> ...
21:24:28 <bblfish> </entry>
21:24:29 <bblfish> </feed>
21:25:07 <Arnia> kpreid: it doesn't generate any triples afaict
21:25:26 <bblfish> the graph is
21:25:27 <bblfish> _f ----a----> <Feed>
21:25:29 <bblfish> |-----head--> _h
21:25:30 <bblfish> |-----title---->"...."
21:25:32 <bblfish> |----generator->
21:25:33 <bblfish> ....
21:26:21 <bblfish> ok, so I should have put atom: all over the place...
21:27:17 <bblfish> this should be easy to test out on the validator...
21:27:33 <Arnia> Surely that's only generated with liberal use of parseType?
21:28:33 <bblfish> yes:
21:28:34 <bblfish> <?xml version="1.0" encoding="utf-8"?>
21:28:36 <bblfish> <atom:feed atom:version="draft-ietf-atompub-format-03: do not deploy"
21:28:37 <bblfish> xmlns:atom="http://purl.org/atom/ns#draft-ietf-atompub-format-03"
21:28:39 <bblfish> xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#">
21:28:40 <bblfish> <atom:head rdf:parseType="Resource">
21:28:42 <bblfish> <atom:title>Example Feed</atom:title>
21:28:43 <bblfish> <atom:link atom:href="http://example.org/"/>
21:28:45 <bblfish> <atom:updated>2003-12-13T18:30:02Z</atom:updated>
21:28:46 <bblfish> <atom:author rdf:parseType="Resource">
21:28:48 <bblfish> <atom:name>John Doe</atom:name>
21:28:50 <bblfish> </atom:author>
21:28:51 <bblfish> </atom:head>
21:28:53 <bblfish> <atom:entry rdf:parseType="Resource">
21:28:54 <bblfish> <atom:title>Atom-Powered Robots Run Amok</atom:title>
21:28:56 <bblfish> <atom:link atom:href="http://example.org/2003/12/13/atom03"/>
21:28:57 <bblfish> <atom:id>vemmi://example.org/2003/32397</atom:id>
21:28:59 <bblfish> <atom:updated>2003-12-13T18:30:02Z</atom:updated>
21:29:00 <bblfish> </atom:entry>
21:29:02 <bblfish> </atom:feed>
21:29:17 <bblfish> the idea is that rdf:parseType could be in the DTD...
21:30:06 <dviner> so, atom:href is the equivalent of rdf:about ?
21:30:48 <bblfish> I can't remember if that is necessary in this case.
21:31:01 <eaon> it is
21:31:12 <eaon> otherwise it's a literal
21:31:25 <bblfish> ahh.
21:31:35 <eaon> should be rdf:resource
21:31:54 <dviner> why rdf:resource and not rdf:about ?
21:31:58 <bblfish> can one remam in OWL atom:href to rdf:about?
21:32:09 <bblfish> s/remam/remap/
21:32:10 <eaon> dviner: link is a property, not a class
21:32:24 <Arnia> bblfish: No. Its a property of the serialisation, not the formal model
21:32:28 <eaon> bblfish: i'm not sure, but parsers won't get it
21:32:48 <eaon> right
21:32:56 <dviner> eaon: isn't the atom:entry really describing the resource pointed to by atom:href ?
21:32:56 <bblfish> I am not too bothered about parsers. I want to be able to say: it is rdf
21:33:36 <bblfish> I don't think so...
21:33:37 <eaon> dviner: it is, but i can't imagine a way to express that with that kind of xml
21:33:47 <Arnia> bblfish: uh? I don't get what you mean by that. The parsers will need to parse the XML and turn it to RDF
21:33:59 <Arnia> (RDF the data model)
21:34:21 <bblfish> yes, but perhaps future parsers would be ok, no?
21:35:19 <Arnia> rdf:about and rdf:resource are two special attributes used for performing the mapping from the document tree to a bag of triples... they're not arcs in the graph
21:35:31 <dviner> eaon: so, if the entire atom:entry describes a resource whose URI is specified in atom:href, wouldn't the proper RDF/XML representation list rdf:about="@value of atom:href@" ?
21:36:20 <eaon> dviner: yep
21:36:25 <dviner> eaon: one might read it as "the following statements are 'about' the thing referred to as @atom:href@"
21:36:36 <Arnia> dviner: rdf:about can only be used on tags that represent nodes... rdf:resource is used on predicate tags
21:36:45 <bblfish> dviner: forget the atom:href that is just a pointer to a web page or something in the context
21:37:06 <dviner> hmm... this is where i get a bit lost...
21:37:20 <Arnia> rdf:about and rdf:resource mean different things
21:37:39 <dviner> i guess i'm not clear on the difference.
21:38:08 <eaon> dviner: the rdf:about refers to the subject, the rdf:resource to the object
21:38:32 <dviner> ok... that's what i thought....
21:38:43 <eaon> _:Michael foaf:made <http://niij.org/> .
21:38:46 <dviner> so, maybe it's my understanding of atom that is fuzzy then.
21:39:00 <Arnia> rdf:about describes the root of a subgraph. rdf:resource points between subgraphs
21:39:02 <eaon> <http://niij.org/> foaf:maker _:Michael .
21:39:39 <bblfish> assuming that future parsers will allow us to map rdf:about and stuff, then I still have the problem with this:
21:39:41 <bblfish> (concentrating on generator)
21:39:42 <bblfish> <?xml version="1.0" encoding="utf-8"?>
21:39:44 <bblfish> <atom:feed atom:version="draft-ietf-atompub-format-03: do not deploy"
21:39:45 <bblfish> xmlns:atom="http://purl.org/atom/ns#draft-ietf-atompub-format-03"
21:39:47 <bblfish> xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#">
21:39:49 <bblfish> <atom:head rdf:parseType="Resource">
21:39:50 <bblfish> <atom:title>Example Feed</atom:title>
21:39:51 <bblfish> <atom:generator atom:uri="https://bloged.dev.java.net/" atom:version="0.5">BlogEd</generator>
21:39:53 <bblfish> <atom:link atom:href="http://example.org/"/>
21:39:54 <bblfish> <atom:updated>2003-12-13T18:30:02Z</atom:updated>
21:39:56 <bblfish> <atom:author rdf:parseType="Resource">
21:39:57 <bblfish> <atom:name>John Doe</atom:name>
21:39:59 <bblfish> </atom:author>
21:40:01 <bblfish> </atom:head>
21:40:02 <bblfish> </atom:feed>
21:40:20 * bblfish forget about the link problem above... sorry I should have removed that
21:41:21 <dviner> so, there's a "thing" (feed) which exists whose title is "Example Feed" ?
21:41:57 <dviner> what is the URI of the "thing" (feed)?
21:42:07 <Arnia> Could be a blank node
21:42:19 <Arnia> Not necessarily a bad modelling decision
21:42:43 <dviner> ahh... hadn't thought of that.
21:42:44 <bblfish> yes
21:43:01 <Arnia> But I find it unlikely that 'future parsers' will 'map' the basic RDF/XML serialisation attributes
21:43:17 <Arnia> The alternative is making atom:href an IFP
21:43:19 <bblfish> well, they are just uri like any others, no?
21:44:03 <bblfish> :-)
21:44:09 <Arnia> bblfish: Its just not likely to happen. rdf:about and rdf:resource are properties of the serialisation not the model
21:44:20 <Arnia> How would you ever express the mapping?
21:44:31 <bblfish> oh. yes...
21:44:36 <Arnia> The only sensible way I can conceive of is XSLT, but you have that already
21:45:01 <bblfish> I was trying to see if one could avoid xslt...
21:45:11 <bblfish> but perhaps there is no way.
21:45:29 <bblfish> especially considering the generator element above. It looses the content I think.
21:45:43 <Arnia> Is this a change to the Atom format?
21:45:56 <bblfish> well it very nearly is Atom...
21:45:59 <Arnia> If so, couldn't you get atom to use rdf:about/rdf:resource? ;)
21:46:27 <bblfish> yes, you could... but I was trying to be as flexible as possible.
21:46:41 <bblfish> I want to see where the limits are
21:47:24 <Arnia> Well... there are two sorts of limits, the limits of the model and the limits of the serialisation. The latter are more stringent than the former
21:48:44 <bblfish> yes. Apparently a DTD can make the rdf:parseType="Resource" a default
21:49:13 <Arnia> Really? That surprises me... how many RDF/XML parsers read DTDs?
21:49:49 <bblfish> again, I don't care too much about that. As long as they could, then I can argue that the above is rdf
21:49:58 <bblfish> but let me get you the article...
21:50:22 <Arnia> Don't you actually want something that can be used?
21:51:03 <dajobe> <bblfish> yes. Apparently a DTD can make the rdf:parseType="Resource" a default
21:51:07 <dajobe> don't rely on that *at all*
21:51:13 <bblfish> here http://www.intertwingly.net/blog/1565.html
21:51:52 <bblfish> I want to argue that atom is rdf, so that I can say: all extensions to atom should also be in rdf
21:52:03 <bblfish> that makes it very easy
21:53:17 <bblfish> it does not matter that current parsers don't parse atom rdf. But if I can say that atom is rdf, then I can provide a simple ontology (OWL)
21:53:19 <bblfish> for atom and have it placed at the namespace location of atom
21:53:42 <dajobe> look, I wrote atom support into my raptor parser. atom in, rdf *triples* out. not rdf/xml
21:53:56 <dajobe> so to me, all atom is rdf or near enough
21:54:19 <bblfish> cool.
21:54:42 <Arnia> Many things can be used as a serialisation of a set of statements in the RDF model
21:54:44 <bblfish> but it would be nice to have it noticed formall by the group. Cause then we can argue that atom extensions should be rdf
21:54:45 <dajobe> atom <> rdf/xml approaches broken. crossing layers
21:55:06 <dajobe> I saw the ultimatum
21:55:21 <Arnia> I'm a great believer in using serialisations appropriate to a task, rather than one serialisation fits all
21:55:25 <eaon> 5 days to go right?
21:55:34 <bblfish> yes 4 days.
21:55:39 <bblfish> sorry 5
21:56:13 <dajobe> the tricky parts are the atom elements with multiple attributes on them
21:56:14 <bblfish> yes, you are right. but there is a problem if people write extensions willy nilly that is not rdf.
21:56:27 <bblfish> then it is much less extensible the RSS1.0 I think
21:56:30 <dajobe> you eithe rneed to map them to a datatyped literal, dropping one. or to multiple predicates
21:56:59 <dajobe> only in the worst case should you introduce a blank node, it disrupts things
21:57:42 <bblfish> dajobe: yes that is
21:57:43 <bblfish> <atom:generator atom:uri="https://bloged.dev.java.net/" atom:version="0.5">BlogEd</atom:generator>
21:57:45 <bblfish> above
21:58:02 <bblfish> I think that is the only one.
21:58:21 <dajobe> invent a datatype for this
21:58:23 <dajobe> then use
21:58:34 <bblfish> how do I do this?
21:58:49 <dajobe> <tom:generator rdf:datatype="some-uri">https://bloged.dev.java.net/ 0.5 BlogEd</atom:generator>
21:59:08 <dajobe> and use some-uri to micro-parse if you want it back later (unlikely)
21:59:09 <jsled> <dajobe> only in the worst case should you introduce a blank node, it disrupts things
21:59:15 <jsled> what does it disrupt?
21:59:24 <dajobe> it disrupts the xml-way of thinking (broken, yes but that's how it goes)
22:00:12 <bblfish> well the blank nodes make atom nearly be rdf... so for my purpose that's ok. I am not here to get the atom group to change their ways too drastically..
22:00:15 <dajobe> the natural rdf-way would be to add new blank nodes
22:00:21 <jsled> Are you suggesting that if designing an RDF schema to be mapped into RDF/XML, then one should not use blank nodes?
22:00:34 <dajobe> but when you show the resulting rdf/xml to atom-xml people, they'll just say bloat or something like that
22:00:51 <dajobe> No, I was very clear I was talking about xml-heads
22:01:33 <bblfish> dajobe: so you are saying that I can't fit <atom:generator atom:uri="https://bloged.dev.java.net/" atom:version="0.5">BlogEd</atom:generator> into rdf as is?
22:02:00 <dajobe> if you start from scratch modelling in rdf, you'll use blank nodes naturally. this isn't the case retro-fitting atom, evolved as an xml file
22:02:01 <bblfish> I need to get the group to change something in other words (not much mind you.)
22:02:11 <Arnia> Personally I think the answer is to keep the XML schema as-is and either transform to RDF/XML via GRDDL or wait for a serialisation that can consume XML like that and turn it into a set of RDF statements
22:02:13 <dajobe> this is in *rdf/xml*
22:02:25 <dajobe> you cannot have property attributes on property elements which have a literal contnet
22:02:42 <dajobe> you can when it's empty, but...
22:02:49 * jsled nods
22:02:53 <jsled> attributes suck.
22:02:57 <dajobe> <atom:generator atom:uri="https://bloged.dev.java.net/" atom:version="0.5" rdf:value="BlogEd"/>
22:03:01 <dajobe> ^- is allowed
22:03:17 <bblfish> yes, I could get the group to change to that I think
22:03:17 <dajobe> that form introduces a secret blank node
22:03:52 <bblfish> I agree that it would be nice to have an rdf serialisation rdf/xml2 that would be more flexible.
22:03:54 <dajobe> try that snipped above in some rdf/xml and stick it through the validator/some rdf parser, see if you like the triples
22:04:03 <bblfish> ah.
22:04:08 <dajobe> yawn, bored of talking about rdf/xml2. done it for years. move on
22:04:14 <dviner> (sidebar) this seems a common theme/thread here... RDF/XML serialization versus RDF as data model. perhaps we should record some good introductions/tutorials/etc for people who are accustomed to xml, but want to get into rdf? it seems this is a common stumbling block when going from "non-rdf-xml-land" -> "rdf"
22:04:17 * dajobe wanders off
22:04:44 <bblfish> dajobe: thanks
22:04:47 <Arnia> Hmm... hadn't seen this before -- http://w3future.com/xr/
22:06:33 <Arnia>http://w3future.com/xr/
22:06:34 <dc_swig> D: http://w3future.com/xr/ from Arnia
22:06:53 <Arnia> D:|RDF extraction from XML
22:06:53 <dc_swig> Titled item D.
22:07:36 <Arnia> D:An XML format for mapping specific XML formats to a simplified form of RDF/XML
22:07:36 <dc_swig> Added comment D1.
22:07:45 <dviner> haha "It is clear that few people find RDF easy to understand"
22:08:08 * eaon would like to replace that with RDF/XML
22:08:20 <crschmidt> no.
22:08:23 <Arnia> D: The demonstration system uses an XSLT transform to turn the XR into XSLT and then applies it with a perl script
22:08:23 <dc_swig> Added comment D2.
22:08:24 <eaon> rdf itself is pretty easy, imo...
22:08:27 <crschmidt> RDF is hard to understand, not just RDF/XML.
22:08:37 <Arnia> RDF is really easy... very simple data model
22:08:38 <crschmidt> *I* don't understand RDF, and I've been working with it for a long time.
22:08:45 <bblfish> no rdf is easy to understand. rdf/xml is hard.
22:09:01 <crschmidt> For you guys, maybe.
22:09:05 <Arnia> Its just a directed labelled graph... pretty simple data structure
22:09:15 <crschmidt> I failed data structures for a reason.
22:09:19 <dviner> haha.
22:09:30 <dviner> crschmidt: i never took data structures :)
22:09:41 <crschmidt> You can say all you want about RDF being easy, but to some people, it's really not.
22:10:01 <crschmidt> No matter whether it's a pretty picture with labelled edges pointing to nodes, or n3, or RDF/XML serialization.
22:10:05 <bblfish> depends on the teacher
22:10:22 <bblfish> I think the problem with rdf is that it is soooooo simple
22:10:40 <bblfish> it is soooo simple. people don't understand how it can be helpful
22:10:52 <Arnia> I think RDF's data model is a lot easier to understand than infoset or hash tables or red-black trees or whatever
22:11:16 <Arnia> RDF is so simple, its difficult to imagine it as being a concept at all :)
22:11:37 <crschmidt> Again, there is a reason I flunked data structures.
22:11:53 <bblfish> yes, when I first saw it, in 2000 or so, I thought ahh now what?
22:12:13 <bblfish> the problem is it only makes sense when you have libraries of stuff written out.
22:12:51 <bblfish> it also helps to have tools, to see how it works.
22:13:21 <Arnia> bblfish: 'graph walking' query languages like reopath help
22:13:42 <Arnia> bblfish: Remind you that you're dealing with a graph that can be traversed like any other
22:14:19 <bblfish> yes, one needs to understand "traverse" and understand that tools can be built that really do traverse it.
22:14:40 <Arnia> crschmidt: So how do you perceive of a set of RDF statements?
22:14:46 <Arnia> s/perceive/conceive
22:15:01 <crschmidt> Arnia: in vim, with xml-mode on ;)
22:15:12 <bblfish> I see a circle going around me and an arrow pointing to some other thing for example.
22:15:19 <dviner> hahaha... this rocks. thank god for irc chat logs.
22:15:44 <dviner> fwiw - i tend to conceive of them as crschmidt does.
22:16:18 <crschmidt> Arnia: I don't conceive of triples as triples. When reading a serialization, I can usually imagine specific triples, but even then, I see them as ntriples: not a graph.
22:17:07 <Arnia> crschmidt: Try and see them as a graph... the whole thing makes sense then. This does explain the card-index approach you have to RDF some
22:17:39 <crschmidt> Arnia: I don't know what a graph even looks like.
22:17:54 <bblfish> ok here:
22:18:07 <bblfish> imagine an circle around you crschmidt
22:18:50 <bblfish> then imagine an arrow starting from that circle and going out of the door and all the way to you best friend
22:19:04 <bblfish> there is also a circle around your best friend
22:19:08 * crschmidt sniffs.
22:19:11 <crschmidt> I don't have any friends.
22:19:20 <bblfish> ;-)
22:19:36 <bblfish> on the arrow there is a label foaf:knows
22:19:36 <crschmidt> all this graph business is making me depressed.
22:19:48 <eaon> ^addturtle _:Michael rel:friendOf _:Chris .
22:19:49 <julie> Model size increased by 1 to 2109739 via turtle statements.
22:19:49 <julie> Adding Turtle failed ('The namespace prefix in ":" was not declared.').
22:19:52 <eaon> are you happy now? :)
22:20:09 <crschmidt> Not really. I'd still like to know why that error comes up
22:20:13 <crschmidt> I should look into that
22:20:20 <bblfish> :-)
22:20:24 <bblfish> ok.
22:20:38 <bblfish> I think that would be fun to make a video like that to explain rdf
22:20:44 <crschmidt> I understand the idea of a graph.
22:20:46 <bblfish> I'll get my brothers to try that
22:20:54 <crschmidt> However, when I conceptualize, I conceptualize in text, not image.s
22:21:02 <crschmidt> And Ascii-art graphs never turn out as well.
22:21:32 <bblfish> that's because you are an early adopter.
22:22:09 <bblfish> when a nice viewer is built into OSX then people will tend to describe it the way I am speaking. I think the fenfire have some good visualisation tools
22:22:30 <[GNU]> crschmidt still arroung?!
22:22:43 <[GNU]> or some other redlandish guy/girl?!
22:22:54 <[GNU]> i am trying to do: select ?nick where (?p foaf:nick "llunved") (?p2 foaf:knows ?p) (?p2 foaf:nick ?nick) using foaf for <http://xmlns.com/foaf/0.1/>
22:22:57 <Arnia> crschmidt: Can you join the triples together in your mind though?
22:23:00 <crschmidt> I'm not redlandish
22:23:03 <[GNU]> but receive an exception
22:23:04 <md-afk> [GNU]: you still ther?
22:23:12 <[GNU]> Furl: ja man
22:23:22 <crschmidt> [GNU]: what is the exception?
22:23:23 * Arnia conceptualises most stuff in graph form anyway
22:23:24 <DanC> ok, I've got a python program merging my 2 parallel email logs and spitting out the result in N3. There's gotta be an easier way to get a lot file with both the message id and the procmail disposition.
22:23:36 <md-afk> UNG : did you try this with cwm?
22:23:47 <[GNU]> md-afk: nope
22:24:01 <crschmidt> Arnia: Like I said, it's triples: one at a time, all the way down.
22:24:02 <[GNU]> crschmidt: hmm.. bad query, need to dum the stack tracke
22:24:05 <md-afk> ok, i can say that cwm is the best thing for small files, very simple to use.
22:24:07 <[GNU]> have stolen the code from julie
22:24:59 <[GNU]> bad query: name 'shorten_uri' is not defined
22:25:01 <Arnia> Hmm... http://www.intertwingly.net/blog/2003/08/15/Inter-net#c1060946612
22:25:19 <crschmidt> [GNU]: that's not a redland error
22:25:33 <crschmidt> [GNU]: That's a problem with the code you're using. you'll need the shorten_uri function from redlandbot
22:25:49 <[GNU]> ah, ja :) now it all make perfect sense
22:25:54 <[GNU]> need more beer
22:26:07 <[GNU]> incomplete stolen code
22:26:12 <jsled> DanC: Take a look at the LOGABSTRACT [From, Subject, dest folder, length] and LOG [since the former doesn't seem to have messageid] procmail variables.
22:26:22 <crschmidt> Going home now. Goodbyeeee
22:26:27 <jsled> crschmidt: cheers.
22:26:34 <Arnia> crschmidt: Bye
22:26:35 <[GNU]> crschmidt: bye and 1000 thanks
22:26:43 <DanC> yes, been studying LOGABSTRACT and LOG
22:26:58 <DanC> I guess LOG=$messageid would kinda do it, but I want a one-line-per-message log
22:27:07 <DanC> maybe I'll learn sgrep
22:27:11 <jsled> Ah.
22:27:11 <md-afk> @prefix log: <http://www.w3.org/2000/10/swap/log#> . this log:forAll :p,:p2,:nick. {:p foaf:nick "llunved".:p2 foaf:knows :p.:p2 foaf:nick :nick.} log:implies{ :nick :knows :p.}.
22:27:20 <md-afk> md-afk is now known as PhUrl
22:27:49 <PhUrl> or log:implies{ :nick :connected-to :p.}.
22:28:11 <PhUrl> or course you need a @prefix : <#>
22:28:15 <PhUrl> or something like that,
22:28:23 <DanC> using @keywords to eliminated : and ?foo for vars, it might look nicer...
22:28:26 <PhUrl> anyway, i find that cwm is really easy to use
22:29:25 <DanC> @keywords is, a. {?p foaf:nick "llunved". ?p2 foaf:knows ?p. ?p2 foaf:nick ?nick.} => { ?nick :knows ?p.}.
22:29:41 <DanC> @keywords a. {?p foaf:nick "llunved". ?p2 foaf:knows ?p. ?p2 foaf:nick ?nick.} => { ?nick knows ?p.}.
22:30:22 <[GNU]> ahm, what?!
22:31:52 <jsled> DanC: Why do you have two parallel email logs?
22:32:25 <jsled> [just curious]
22:32:41 <DanC> one is one-line-per-message, with date/from/to/cc/message-id, and the other is procmail's log
22:32:56 <DanC> the one-line-per-message log is for use with grep
22:33:12 <DanC> (and I keep a log per week in each format
22:33:13 <DanC> )
22:33:45 <DanC> named message-log-YYYY-MMWW and procmail-log-YYYYMMWW respectively (where WW is a week number; weeks that span months get split into 2 files)
22:33:58 <dviner> DanC: are you using sendmail and procmail ?
22:34:13 <Arnia> Can someone remind me why this whole Pie/Excho/Atom/Whatever standardisation thing was started... does it actually improve stuff measurably?
22:34:15 <DanC> don't think I'm using sendmail
22:34:19 <PhUrl> DanC: is that cwm syntax? really? i am suprised
22:34:46 <DanC> which part surprises you? cwm syntax (n3) changes from time to time
22:34:48 <jsled> Arnia: yeah, I think so. At a social/corporate level, though.
22:35:07 <[GNU]> what is pie and excho ?!
22:35:12 <[GNU]> what is the context?
22:35:25 <jsled> i.e., now entities like Google/Blogger have a format they can traffic in that's not RSS2 with it's spec issues.
22:35:32 <jsled> Which is a good alternative.
22:35:38 <Arnia> s/excho/echo
22:35:39 <PhUrl> DanC: i have been living in a cave for at least 8 months. hunting and gathering, lot has changed. :)
22:35:41 <jsled> pie and echo are previous name-candidates for Atom.
22:35:50 <DanC> Arnia, no, I don't think so. I've always thought the "interop around XYZ isn't very good; let's make a new spec that does the same thing with better interop" strategy is a waste
22:35:55 <[GNU]> oic, thx
22:36:23 <DanC> just fix XYZ
22:36:32 <DanC> i.e. fix implementations of XYZ
22:36:40 <Arnia> DanC: Yeah, that was my opinion too... I just feel its another political fragmentation
22:37:21 <Arnia> Admittedly, when ego gets involved sometimes a clean slate can be a good idea
22:37:42 <DanC> but the idea that new specs will somehow fix problems with the implementations is... beyond me.
22:37:44 <jsled> easier said than done ,though, regarding fixing implementations.
22:38:08 * jsled phonecalls
22:38:16 <DanC> fixing implementations can't be harder than implementing a new spec
22:43:41 <Arnia> I'm both a little surprised that RSS 1 has been rejected so quickly, and a little worried about how these formats have fractured
22:47:50 <timbl> Sounds like a lot of Not Invented Here syndrome.
22:48:11 <DanC> rejected? RSS 1 is alive and well.
22:48:57 <dviner> my understanding is that "most people" found rss 1 *far* more complicated than rss 0.9 or 2.0. especially for the simple article-link syndication people wanted.
22:49:36 <dviner> there may be an element of N-I-H, but i think it's more "wow this seems too hard for something relatively simple"
22:50:00 <Arnia> DanC: I don't mean its 'dead'... I mean that many people have rejected it and moved to developing other formats.
22:50:15 <DanC> so it seems. sigh.
22:50:43 <DanC> I haven't seen any RSS consumers that don't consume RSS 1.0, though
22:52:41 <Arnia> DanC: which was my point... the atom effort is just going to lead to Yet Another Format that must be consumed. I don't think it will effectively obsolete anything already out there
22:53:10 <DanC> quite
22:53:19 <[GNU]> "effectivly obsolete" is a nice pharse for not invente anything
22:54:01 <Arnia> [GNU]: ?
22:54:21 <[GNU]> ah... just learning new english phrases :)
22:55:13 <timbl> NIH: It is always amazing how people can think that an idea they can imagine coding up tonight is a safer bet than a product someone else hasn't quite completely documented... or is it that actually making a new one is more fun?
22:55:13 <dc_swig> Label NIH not found.
22:55:22 <timbl> relax, dc
22:55:43 <Arnia> dviner: Perhaps the answer is to evolve RSS1 though, rather than throwing the baby out with the bathwater and inventing Yet Another Syndication Format
22:55:58 <Arnia> timbl: deep cleansing breaths?
22:56:52 <dajobe> (aside: there's a new dc bot in testing elsewhere. Not sure if it still does that.)
22:57:38 <timbl> Well, in a way with GRDDL it doesn't matter so much what people design ... but it is interesting to start circulating RSS1 feeds full of other RDF like FOAF stuff about the particpants and event data calendar stuff and location information and metadata about the photo and so on.
22:59:48 <Arnia> Yes... hmm... a thought: would it be an idea to try and get some RDF parsers into Mono, Python, GNU Classpath etc... to try and get the same adoption as XML?
23:01:33 <timbl> That is interesting.... though you'd need more than just a parser ... a stroe too, and then you would have compettion!
23:01:46 <timbl> Working on common APIs is I think really useful.
23:01:54 <Arnia> An RDF equivalent to SAX?
23:02:00 <timbl> So stores can be swapped in and out.
23:02:05 <kota_> kota_ is now known as kota
23:02:25 <dviner> Arnia: i totally agree... don't throw baby out w/ bathwater... I think most people couldn't distinguish the baby from the bathwater in RSS 1.0.. hence the "throw away everything" approach.
23:03:37 <Arnia> timbl: I might spec up a few APIish thoughts and play with them in mono and python (possibly java if I can stomach it :). Would there be interest to see the results of this do you think?
23:04:21 <timbl> There are lots of existing APIs.
23:05:03 <timbl> If the thoughts take the form of an analysis of the field and a plan for converegence, then that would be cool. If they take the form of yet another API, .... :)
23:05:52 <timbl> I wrote a shim to make the redland api available to cwm users at one point.
23:06:18 <Arnia> :) No, not yet another API... I'm interested in the influence of idiomatic bindings though (which I believe in). I'm using idiomatic binding a lot in Frege for example so that each language sees the store in the manner it sees most natural
23:07:56 <dviner> doesn't redland offer an api like this?
23:08:42 <Arnia> dviner: If you want to write a store for redland atm you have to write it in C. It would be interesting to remove that requirement
23:08:45 <dajobe> yes, several language bindings have been given improvements to use native metaphors
23:10:59 <dviner> i've not explored this too much, but http://librdf.org/docs/perl.html (and others from http://librdf.org/bindings/) make it _look_ like i could overload/subclass the appropriate language class for my own storage.
23:11:49 <dviner> have i misread it?
23:12:03 <dajobe> should work
23:13:14 <dviner> Arnia: are you think that I (random rdf developer) might want to build a store for redland, and have it immediately accessible to all language bindings with no "thunking" layer?
23:13:23 <dviner> s/think/thinking/
23:13:59 <Arnia> dviner: No, not at all. I'm thinking that people may 'trust' RDF more if they know its in the core library for a given language/platform
23:14:12 <Arnia> dviner: like XML is now
23:16:55 <dviner> hmm... i'm not sure i understand what you mean by "in the core library for a given language".
23:17:09 <crschmidt> dviner: so you don't have to install rdflib
23:17:13 <crschmidt> dviner: you just import rdf
23:17:53 <Arnia> Or in mono; using Mono.RDF
23:18:34 <dviner> i see... i guess that's a bit language dependent... it would not be possible to do that in, say, C or C++, unless ANSI-C changed.
23:19:20 <dviner> and Perl tends to not incorporate such things into the core either ... i don't have as much familiarity w/ Mono or Python to say.
23:20:52 <Arnia> dviner: Python, mono and java have the notion of a 'standard library' that is larger than many languages/platforms. The idea is that you can always depend on things in the standard library being on all systems. No dependency hell
23:23:14 <Arnia> dviner: I definitely think its helped XML's use with these languages.
23:23:58 <dviner> makes sense. i definitely agree it would be cool, and clearly wouldn't hurt adoption to have RDF in the core language distribution/libraries
23:25:22 <jsled> ah, but it requires adoption. chicken/egg.
23:25:43 <Arnia> Yeah... I'll look into working on something over the summer (I have to get Frege done first for my dissertation)
23:26:32 <Arnia> jsled: I'm hoping to ask the Mono crowd very nicely to let me place it under their extension namespace if I provide enough unit tests and documentation of all the code
23:29:45 <Arnia> Hmm... XR looks like a cool way to write GRDDL stylesheets...
23:32:50 * timbl away family time
23:33:25 <Arnia> bblfish: Have you looked at this btw? http://w3future.com/tools/atom2rdf.xml
23:33:38 <ows> night
23:33:47 <bblfish> I think I have just invented rdf/xml2
23:33:56 <Arnia> hmm?
23:34:11 <bblfish> I am just going to publish it. wait :-)
23:34:26 <bblfish> we don't need rdf:parseType we just need OWL
23:34:38 <jsled> Oh poo. _I_ was going to invent RDF/XML2 tonight...
23:34:40 <jsled> :)
23:34:57 <bblfish> too late
23:35:01 <jsled> apparently.
23:35:02 <bblfish> :->
23:35:24 <jsled> Of course, yesterday I found out that TimBL invented RDF/XML2 years ago.
23:35:50 <bblfish> just go to http://groups-beta.google.com/group/atom-owl
23:35:51 <Arnia> Who needs RDF/XML2 when you have GRDDL and N3 ;)
23:36:39 <bblfish> rdf/xml2 is just rdf
23:36:43 <bblfish> sorry
23:36:50 <bblfish> rdf/xml2 just is xml
23:36:55 <bblfish> +OWL
23:37:12 <Arnia> How does this deal with striping?
23:40:44 <bblfish_> got cut off
23:40:58 <bblfish_> Arnia: stripping?
23:41:24 <Arnia> The alternation, in RDF/XML, of node tags and predicate tags
23:41:36 <bblfish_> everything is a blank node :-)
23:41:42 <bblfish_> just look at the diagram
23:42:00 <bblfish_> well not everything
23:42:54 <Arnia> Sorry, I still don't see how you've avoided using rdf:parseType
23:43:04 <bblfish_> <feed atom:version="draft-ietf-atompub-format-03: do not deploy"
23:43:06 <bblfish_> xmlns="http://purl.org/atom/ns#draft-ietf-atompub-format-03">
23:43:08 <bblfish_> <head>
23:43:09 <bblfish_> <title>Example Feed</atom:title>
23:43:11 <bblfish_> <link href="http://example.org/"/>
23:43:12 <bblfish_> <updated>2003-12-13T18:30:02Z</atom:updated>
23:43:14 <bblfish_> <author>
23:43:15 <bblfish_> <name>John Doe</atom:name>
23:43:17 <bblfish_> </author>
23:43:18 <bblfish_> </head>
23:43:20 <bblfish_> <entry rdf:parseType="Resource">
23:43:22 <bblfish_> <title>Atom-Powered Robots Run Amok</atom:title>
23:43:23 <bblfish_> <link href="http://example.org/2003/12/13/atom03"/>
23:43:25 <bblfish_> <id>vemmi://example.org/2003/32397</atom:id>
23:43:26 <bblfish_> <versionid>vemmi://example.org/2003/32397/v1</versionid>
23:43:28 <bblfish_> <updated>2003-12-13T18:30:02Z</atom:updated>
23:43:29 <bblfish_> </entry>
23:43:31 <bblfish_> </feed>
23:43:34 <bblfish_> gives this:
23:43:35 <bblfish_> _f --feed--->_h
23:43:37 <bblfish_> |---head--->_h
23:43:39 <bblfish_> | |---title-->"example Feed"
23:43:41 <bblfish_> | |---link--->_l
23:43:43 <bblfish_> | | |--href---->"http://example.org/"
23:43:45 <bblfish_> | |--updated->"2003-12-13T18:30:02Z"
23:43:47 <bblfish_> | |--author-->a
23:43:49 <bblfish_> | |---name--->"John Doe"
23:43:51 <bblfish_> | |---mbox--->"john@doe.com"
23:43:53 <bblfish_> |--entry-->_e
23:43:55 <bblfish_> |--title-->"Atom Powered Robots Run Amok"
23:43:57 <bblfish_> |--link--->_l
23:43:59 <bblfish_> | |--href----->"http://example.org/2003/12/13/atom03"
23:44:01 <bblfish_> |--id------->"vemmi://example.org/2003/12/13/atom03"
23:44:03 <bblfish_> |-versionid->"vemmi://example.org/2003/12/13/atom03/v1"
23:44:05 <bblfish_> |--updated-->"2003-12-13T18:30:02Z"
23:44:24 <bblfish_> that is just the interpretation in rdf/xml2
23:44:27 <bblfish_> that what I say :-)
23:44:39 * Arnia feels very lost
23:44:44 <bblfish_> do we need anything more?
23:45:24 <bblfish_> it is nearly just DOM stuff
23:45:32 <eaon> resources instead of literals...
23:45:45 <eaon> uris, i mean
23:45:47 <jsled> Arnia: he's suggesting that parseType="Resource" is the default, and XML elements are rdf properties.
23:45:57 <jsled> there is no striping.
23:46:08 <bblfish_> yes, attributes are always properties on the elements
23:46:13 <bblfish_> that contain them
23:46:16 <Arnia> jsled: Ah... ok. But how does he define that so that an RDF/XML parser can consume it?
23:46:35 <bblfish_> rdf/xml parsers out of the door
23:46:39 <bblfish_> bblfish_ is now known as bblfish
23:46:58 <bblfish> we don't need them anymore :-)
23:47:12 <bblfish> we just need xml parsers
23:47:39 <jsled> what the heck is vemmi://, btw?
23:47:51 <bblfish> don't know, some unkonwn uri tupe
23:47:53 <eaon> and how is it supposed to know whats a uri and whats not?
23:48:05 <bblfish> look at the OWL ontology
23:49:10 <bblfish> if a property is functional, inverse functional, symmetric, and transitive then the thing to the right is a uri
23:49:54 <bblfish> (symmetric will do I think
23:49:55 <Arnia> bblfish: No, that isn't true
23:49:59 <eaon> nope
23:50:06 <eaon> bblfish: look at foaf:aimChatID
23:50:13 <eaon> thats an ifp but not a uri
23:50:15 <Arnia> foaf:mbox_sha1sum
23:50:20 <eaon> or that :)
23:52:08 <bblfish> well if it is symmetric then
23:52:09 <bblfish> a---->"b"
23:52:11 <bblfish> means that "b"---->a and that can only be true if "b" is a uri, no?
23:52:49 <Arnia> A literal that contains the text "http://www.google.com" is not a URI
23:52:59 <Arnia> Its a literal that contains some text
23:53:49 <bblfish> ok so lets say the predicate is said to be from uris to uris, then the parser can deduce that what is to the right is a uri, no?
23:54:10 <eaon> what about an uri datatype? :P
23:54:46 <Arnia> bblfish: The parser can't, a reasoner may be able to later once everything is in the model
23:55:06 <Arnia> I don't believe parsers exist that do OWL or RDFS reasoning in process
23:55:22 <dajobe> you can't only allow properties=attributes, since then you can't say a) they are URIs vs literals b) are datatyped c) give them different languages
23:57:30 <bblfish> could the href property not be defined as always being from uris to uris?
23:59:13 <dajobe> and what about all the other properties
23:59:17 <dajobe> and ones that take uris and literals
23:59:21 <dajobe> all in owl-land that's evil
23:59:25 <dajobe> ^although
23:59:41 <bblfish> ah a property can take uris and literals...
23:59:58 <dajobe> nn
The IRC chat here was automatically logged without editing and contains content written by the chat participants identified by their IRC nick. No other identity is recorded.
Alternate versions:
and
Text
Provided by Dave Beckett. Hosted by Useful Information Company.