Semantic Web Interest Group IRC Chat Logs for 2003-03-27

00:08:45 <eikeon> Can someone point me at one of the pages at w3 that has multiple views?

00:08:56 <eikeon> I can not think of any off hand.

00:12:04 <eikeon> Hum... unless all the ones that used to give a page listing the choices now default to the rdf/xml instead?

00:14:55 <chaalsMEL> eikeion: you mean something that is content-negotiated?

00:15:15 * eikeon is really looking for a way to have pages written using RDF and RDF stylesheets... so that I can have one URI for both the rdf/xml content and a human readable version of it.

00:15:18 <eikeon> chaalsMEL: Yes.

00:16:21 <chaalsMEL> One approach that has been suggested is to use an XSLT to do a simple "two possible views". Don't think it really works though in the long run.

00:16:31 <chaalsMEL> URI coming - hang on a moment

00:17:38 <niq> http://www.w3.org/ is content-negotiated, FWIW

00:17:39 <dc_rdfig> A: http://www.w3.org/ from niq

00:17:55 <niq> ohbugger, forgot chump

00:18:40 * niq isn't quite sure what you're talking about, but just noted Chaals mentioning a technique I use

00:21:27 <eikeon> I have many pages where I provide some aggregation of rdf... and am currently providing both a html view and an rdf/xml one via different urls -- but I want to be able to use just one. Is content negotiation the right thing in this case?

00:23:21 <niq> probably not; conventional browsers have a bad habit of throwing garbage in to Accept: headers

00:23:54 * FloK is away: I'm busy

00:24:27 <eikeon> What content types can one content negotiate for with http://www.w3.org/

00:25:26 <chaalsMEL> depends what is there. Often RDF and HTML (e.g. irc logs), or html and xhtml, and different language versions

00:25:38 <niq> serves html as application/xhtml+xml if you ask for it

00:25:53 <chaalsMEL> or do you mean the particular page itself... (what niq said in that case)

00:26:45 <niq> (for a brief period it served xhtml as default - so it didn't work at all in certain so-called browsers)

00:27:59 <eikeon> I asked since I assume http://www.w3.org/ work in browsers. And if rdf/xml was one of the version is can serve up... then I should be able to serve up both an html and rdf/xml version to, no?

00:28:25 * chaalsMEL looks for an example of that...

00:29:37 <niq> Not sure .. any reason not to use distinct page.[html|rdf]?

00:30:19 <chaalsMEL>http://www.w3.org/2003/03/05-tp-irc

00:30:19 <dc_rdfig> B: http://www.w3.org/2003/03/05-tp-irc from chaalsMEL

00:30:27 * niq uses a variant on that principle; not content-negotiation, but users select either HTML or RDF from a Form

00:30:49 <chaalsMEL> B:| an example of a URI that is available as different representations - html, RDF or plain text.

00:30:50 <dc_rdfig> Titled item B.

00:31:35 <chaalsMEL> B:Each variant has its own URI as well - for example http://www.w3.org/2003/03/05-tp-irc.html is the HTML representation

00:31:35 <dc_rdfig> Added comment B1.

00:32:00 <chaalsMEL> B:| http://www.w3.org/2003/03/05-tp-irc - an example of a URI that is available as different representations

00:32:01 <dc_rdfig> Titled item B.

00:32:36 <chaalsMEL> B: there are RDF and plain text variants available too - http://www.w3.org/2003/03/05-tp-irc.rdf and http://www.w3.org/2003/03/05-tp-irc.txt

00:32:36 <dc_rdfig> Added comment B2.

00:33:04 <niq> s/rdf/xml/ (as far as content negotiation is concerned)

00:35:33 <eikeon> chaalsMEL: Thank you... that is what I was looking for.

00:36:04 <DanC> hmm... should I chmod -w the old twiki stuff now that MoinMoin is rolling?

00:36:09 <DanC> I think I shall...

00:39:09 <eikeon> chaalsMEL: One last question... do you know if there is a way to get browsers to request and display say the plain/text version. For my case I would like to be able to link to it as a for of "view source"

00:40:12 <eikeon> Ah... think I know...

00:40:40 <eikeon> In that case I can create another uri in addition.

00:42:21 * eikeon concludes... think I will have three uris for each of my resource. One that content negotiates... and one for each of the two types.

00:43:22 <DanC> BLURB: migrated ESW wiki from twiki to MoinMoin

00:43:22 <dc_rdfig> C: migrated ESW wiki from twiki to MoinMoin from DanC

00:43:47 <DanC> C:[http://esw.w3.org/t/view/ESW/AboutThisService|twiki service] no longer writeable. writes fail silently :-{

00:43:48 <dc_rdfig> Added comment C1.

00:44:24 <DanC> C:[http://esw.w3.org/topic/AboutThisService|new service] does conflict detection, reasonable attribution, RSS RecentChanges

00:44:24 <dc_rdfig> Added comment C2.

00:59:26 <soccos|away> soccos|away is now known as soccos

01:00:33 <soccos> soccos is now known as soccos|off

01:15:54 <danbri_> The Wiki is dead! Long live the Wiki! http://esw.w3.org/t/view/ESW/RdfCalendar -> http://esw.w3.org/topic/RdfCalendar

01:16:09 <soccos|off> soccos|off is now known as soccos

01:22:32 <taaz> too many wiki options ;) i've had people hijack moinmoin wikis i setup and convert to phpwiki or zwiki.

01:23:28 <danbri_> now you tell us! ;)

01:24:32 <taaz> i like moin... written in python so it must be good ;)

01:25:20 * bitsko likes MoinMoin's extensibility

01:25:51 <bitsko> Moin's a little on the heavy side, tho. could do for a lot of refactoring

01:27:52 <taaz> i've always wanted to refactor the save/revision-view/diff code out to be modular so you could plug in CVS instead of the lots-of-files revision history method

01:28:17 <taaz> well... i think it still works like that...

01:29:09 <taaz> i'm suprised you all don't have a custom rdf based wiki ;)

01:29:50 <zool> nn

01:31:40 <taaz> maybe someone should define "esw" on the front wiki page... for those of us who have no idea what it means ;)

01:32:46 <danbri_> either Extended, Evolving or European Semantic Web

01:32:59 <danbri_> but yes, we need to work on the welcome page stuff

01:33:02 <danbri_> not tonight though!

01:34:14 <soccos> taaz: ;-)

01:44:35 <bitsko> I'm not sure what an rdf-*based* wiki would do, but wiki as an ad-hoc store could be cool. calendar, foaf, rss:item's, software pkgs

01:56:02 <soccos> soccos is now known as soccos|away

02:51:44 <soccos|away> soccos|away is now known as soccos

02:51:55 <soccos> is edd's foaf bot running here?

02:58:18 <soccos> soccos is now known as soccos|away

04:40:29 * DanC returns from a nice run

04:41:14 <DanC> I'm ... uh... in training for a 5k race. http://www.sanctuaryofhope.org/soh/race_home.htm

04:56:56 <chaalsMEL> ?me isn't sure about the idea of a nice run, but understands doing it as training.

04:57:25 * chaalsMEL tries to get the hang of keyboard swapping :(

04:58:09 <DanC> well, as winter turns to spring, and you can walk around outside at 10pm in shorts, it's nice.

04:58:28 <DanC> it was a gorgeous day here in OPKS, but I didn't really get outside until I went running tonight.

04:58:38 <chaalsMEL> Yep, this looks nice...

04:59:05 * chaalsMEL thinks about climbing into the roof to play with fibreglass insulation batts :(

04:59:44 <DanC> ew

05:17:16 <chaalsMEL> and it's a lovely day outside :(

05:22:46 * chaalsMEL wonders again about pointers to bits of PDF documents.

05:22:58 * chaalsMEL heads into the roof to ponder it.

05:23:02 <chaalsMEL> chaalsMEL is now known as chaalsOot

06:09:02 <DanC>http://bitconjurer.org/BitTorrent/

06:09:02 <dc_rdfig> D: http://bitconjurer.org/BitTorrent/ from DanC

06:09:07 <DanC> D:|BitTorrent

06:09:07 <dc_rdfig> Titled item D.

06:09:19 <DanC> D:interesting bulk file distribution mechanism. Anybody tried it?

06:09:19 <dc_rdfig> Added comment D1.

06:09:27 <deltab> yes

06:10:16 <taaz> taaz is now known as taazzzz

06:17:47 <DanC> er... and...?

06:19:07 <DanC> TrustBot, help?

06:19:07 <TrustBot> I'm sorry. Your input didn't make sense to me. Please type "TrustBot, help" for instructions.

06:19:11 <DanC> TrustBot, help

06:19:11 <TrustBot> You can talk to me on an IRC channel or through a private message.

06:19:12 <TrustBot> Please say: "TrustBot, me@something.example.com to stranger@something.example.com" to determine how much you should trust stranger@something.example.com

06:19:13 <TrustBot> "TrustBot, maxpath me@something.example.com to stranger@something.example.com" will give you the value of the maximum weight path from me@something.example.com to stranger@something.example.com

06:19:14 <TrustBot> "TrustBot, minpath me@something.example.com to stranger@something.example.com" will give you the value of the minimum weight path from me@something.example.com to stranger@something.example.com

06:19:15 <TrustBot> "TrustBot, paths me@something.example.com to stranger@something.example.com" will show you the longest and shortest paths from from me@something.example.com to stranger@something.example.com. It will show the source and the sink, along with the number and weight of the intermediate edges.

06:19:17 <TrustBot> For example, an output of "me@something.example.com -> 8.0 -> 3.0 -> stranger@something.example.com" would indicate that me@something.example.com trusted someone at level 8.0 who in turn trusted stranger@something.example.com at level 3.0. To respect privacy, the sources of the intermediate edges are not shown.

06:19:21 <TrustBot> If you want to know *why* I gave you a trust rating, please say "TrustBot, explain me@something.example.com to stranger@something.example.com"

06:19:24 <TrustBot> You can also say "TrustBot, who" to get a list of all of the people I know.

06:19:26 <TrustBot> If you've added files to the database, say "TrustBot, reload"

06:19:28 <TrustBot> If you would like to contribute to the graph, please visit http://www.mindswap.org/~golbeck/web/trust.shtml - Thanks for your support of TrustBot.

06:19:46 <golbeck> :)

07:30:05 * DanC researches WikiNames a bit... http://c2.com/cgi/wiki?WikiNamePluralProblem

07:38:32 * FloK is away: I'm busy

10:20:51 <golbeck> golbeck is now known as golbeck_zzz

10:57:15 <danbri> danbri is now known as danbri_lap

13:57:40 <libby> hey andys

13:58:21 <chaalsMEL> hey andy, libby

13:59:04 <libby> hey chaalsMEL, didnt see you there!

13:59:20 * danbri_ waves

13:59:24 <danbri_> danbri_ is now known as danbri

13:59:44 <danbri> danbri has changed the topic to: RDF/SemWeb 24x7 rdfquery-flavoured hack'n'chat http://rdfig.xmlhack.com/

14:01:52 <dajobe> why?

14:02:07 <dajobe> and also not-rdfquery flavoured too

14:02:24 * chaalsMEL wants dajobe to repeat his query in a different syntax

14:02:54 <danbri>http://lists.w3.org/Archives/Public/www-rdf-rules/2003Mar/0035.html

14:02:55 <dc_rdfig> E: http://lists.w3.org/Archives/Public/www-rdf-rules/2003Mar/0035.html from danbri

14:03:11 <danbri> E:|AGENDA for RDF query testcases IRC meeting, Thursday, March 27, 2003, 15:00 UTC

14:03:12 <dc_rdfig> Titled item E.

14:03:58 <dajobe> dajobe has changed the topic to: RDF/SemWeb 24x7 hack'n'chat http://rdfig.xmlhack.com/ | rdf query meeting 15:00 UTC thursday

14:04:15 <danbri> ta

14:04:33 <dajobe> how do I/anyone find out about these in advance?

14:04:40 <danbri> Join the mailing list

14:04:44 <dajobe> no thanks

14:04:53 <danbri> Why not?

14:05:06 <danbri> You were talking about doing rdf query for Redland...

14:05:08 <dajobe> er, less email is what I'm aiming for

14:05:38 <dajobe> I just was asking how the irc-world got any indication of what was coming up?

14:05:57 * chaalsMEL wondesr if we still have an rss feed from the swad-e events page

14:06:01 <dajobe> given being here in time&cyber-place requries a bit of advanced notice

14:06:08 <libby> well, last meet we siad when the next one was and that was chumped

14:06:08 <dajobe> vs email, doesn't

14:06:29 <libby> so the irc world, reading the chump might find out that way

14:06:41 <libby> chaals, good point. we could link these meets in here

14:06:43 <danbri> I would like to get a regular fortnightly slot for query collaboration, same for calendar

14:07:15 <danbri> the cal meetings are available in ical, aren't they...?

14:07:28 <dajobe> can we put the regular meeting times in the topic or is that too early?

14:07:32 <libby> yep ical

14:07:35 <chaalsMEL> the good point was wondering if we can get the RSS feed off the swad-e events page (and further, if the calendar folks have come up with an explanation for how to do recurring events yet :)

14:07:56 <libby> we have alayout for recuuring events; not tested them yet really

14:08:33 <danbri> dajobe, too early for query

14:08:37 <libby> yes, there's a swad-e rss events channel.

14:15:08 <dajobe> how about a wiki page for upcoming events, publically writable

14:15:25 <dajobe> and link to that in the topic, then the wiki page can link to the announcements, agendas etc.

14:18:54 <libby> we're experimenting with that in the calendar one. maybe see how it goes there.

14:18:55 <danbri> yes, that would be good. hopefully we'll get fixed timeslots, which will make things less complex

14:19:07 <danbri> classic prob of wanting info in multiple places, though

14:19:15 <libby> yeah

14:19:47 <danbri> an #rdfig regular would want it in one place for the channel; a cal regular would maybe prefer it on the cal page, as not so interested in other rdf stuff

14:21:15 <libby> true and many thgs to keep updates would be a pain

14:22:44 <dajobe> I don't mind where

14:22:51 <dajobe> as long as it is in 1 place that I can find

14:24:56 <libby> dave it is in one place you can find. as with query we always send times and dates of meetins, and agendas and summaries to www-rdf-rules, the rdf query and rules mailing list. which is archived.

14:25:49 <danbri> <dajobe> I don't mind where

14:25:49 <danbri> <dajobe> as long as it is in 1 place that I can find

14:26:09 * danbri wonders if we're talking past each other

14:26:20 <dajobe> I meant 1 URL (rss feed?), not having to go to a mailing list, find the right month, seartch the messages etc.

14:26:59 <danbri> yes it would be nice to have one place for any given topic/query/forum, but that right now means building aggregators that don't exist, or else doing a bunch of extra admin

14:27:15 <danbri> a wiki page for rdfig meetings would be a start, but there'll always be other info elsewhere you'll also want

14:28:50 <dajobe> not sure I'm getting across. If someone reading irc only wanted to know about upcoming events, schedule one themselves, find the links, how would they find the right mailing lists and messages.

14:28:57 <dajobe> nevermind

14:30:27 <danbri> yes, that situation should get improved. the wiki is part of getting there...

14:33:33 <libby> hey alberto!

14:35:48 * AndyS waves (finally!)

14:35:50 * danbri adds link to W3C QA Activity to http://esw.w3.org/topic/RDFQueryTestCases

14:36:04 <danbri> hi

14:36:13 <alberto> waves to danbri, andy and libby (again:)

14:36:51 <danbri> alberto, i would've sent my comments to www-rdf-rules rather than offlist, but wanted to give you a chance to announe the doc...

14:37:13 <alberto> yes, I am gonna do it right now - thanks

14:37:47 <libby> so, what do we think about testcases, e.g. how to oragnise them.....

14:38:01 <danbri>http://www.w3.org/2003/03/rdfqr-tests/rdf-query-testcases.html

14:38:02 <dc_rdfig> F: http://www.w3.org/2003/03/rdfqr-tests/rdf-query-testcases.html from danbri

14:38:17 <danbri> F:|RDF query test cases - a discussion document

14:38:17 <dc_rdfig> Titled item F.

14:38:35 <danbri> F:"This document describes an RDF vocabulary to express machine-processable RDF Query Test cases to support interoperability testing among several different RDF query implementations, the creation of a test cases repository and the design of the machinary for mapping between different query syntaxes."

14:38:35 <dc_rdfig> Added comment F1.

14:38:40 <alberto> is writing to www-rdf-rules...

14:38:49 <danbri> cool

14:38:54 <libby> andys, alberto, me, danbri all have some testcases

14:39:00 <libby> jeen too, I think

14:39:12 <libby> jos...

14:39:18 <danbri> my tests currently only test the query parser, but can extend easily

14:39:18 <libby> there must be loads more

14:39:32 <danbri> am not sure if alberto's format entirely makes mine redundant

14:39:35 <libby> I'm wondreing how we should put them on the site

14:40:03 <danbri> My format: <t:Query t:name="examples/squish/test0c" t:format="squish" t:clauses="7" t:variables="7" t:wf="true" t:comment="ok"/>

14:40:07 <AndyS> Let's get people to package them independently from their (our) downloads.

14:40:18 <danbri> ...how much of that can I do in the new format?

14:40:24 <libby> for calendar ones, we have ones we've generated in one dir; others, I'm wondering whether to put in separate directoies

14:40:31 <libby> andys, I agree

14:40:47 <danbri> AndyS, I agree. Having a few in the w3.org filetree might be useful, but I think there'll naturally be bundles of queries+data managed elsewhere

14:41:06 <danbri>http://rdfweb.org/2003/02/a-z/intro.html

14:41:06 <dc_rdfig> G: http://rdfweb.org/2003/02/a-z/intro.html from danbri

14:41:15 <danbri> G:|FOAF Test Data: A Fictional A-Z

14:41:15 <dc_rdfig> Titled item G.

14:41:32 <danbri> G:I want to package some queries and a testcase manifest with these sample files

14:41:32 <dc_rdfig> Added comment G1.

14:41:39 <AndyS> Libby is suggesting stronger than just describing the tests ad hoc with "describing triple-pattern queries" ...

14:42:08 <AndyS> ... Danbri's format relies on understanding literlas like "squish"

14:42:11 * danbri not quite following

14:42:31 <danbri> Yes, that was a design error. A querylanguageSpec property that took a URI would be more decentralist

14:43:10 <AndyS> Even a URI is just ref'ing it - not enabling one test suite be fed into another system.

14:43:30 <AndyS> (Unless Libby puts her translation service up as a webservice!)

14:44:13 <libby> andys, if you'er talkign about proposed agensda item for next time, I think we should discuss it, but that doesnt mean we shopudl implement it....

14:44:45 <libby> the important bits I think are yours and albertos - a neutral format for testing...

14:45:02 * danbri agrees; declaring which QL was used is one step; being able to translate between them, and represent the lossiness when that happens, is a natural next step

14:45:20 <libby> RDF Query tests meet 1. (H) Proposed resultset vocabulary and example

14:45:25 <danbri> ntriples provides a neutral middleground, but starts to hurt immediately when we want to explore richer interop

14:45:28 <libby> sorry, ignore that!

14:45:34 <libby> BLURB: RDF Query tests meet 1. (H) Proposed resultset vocabulary and example

14:45:35 <dc_rdfig> H: RDF Query tests meet 1. (H) Proposed resultset vocabulary and example from libby

14:46:01 <AndyS> Implementing is good - it is a test of ideas. An executable format is possible (c.f. DQL, Edutella, and my pref, cwm)

14:46:10 <libby> H: see [agenda|http://lists.w3.org/Archives/Public/www-rdf-rules/2003Mar/0035.html]

14:46:10 <dc_rdfig> Added comment H1.

14:46:38 <AndyS> Let's wait for the meeting ... !

14:46:43 * danbri grins

14:46:50 <libby> BLURB: RDF Query tests meet agenda item 2. (I) Proposed manifest vocabulary

14:46:50 <dc_rdfig> I: RDF Query tests meet agenda item 2. (I) Proposed manifest vocabulary from libby

14:47:01 <libby> yeah sorry, just trying to cump the agenda items

14:47:33 <libby> BLURB: RDF Query tests meet agenda item 3. (J) RDF query events at www2003?

14:47:33 <dc_rdfig> J: RDF Query tests meet agenda item 3. (J) RDF query events at www2003? from libby

14:47:40 <libby> idoru: see [agenda|http://lists.w3.org/Archives/Public/www-rdf-rules/2003Mar/0035.html]

14:47:46 <libby> I:see [agenda|http://lists.w3.org/Archives/Public/www-rdf-rules/2003Mar/0035.html]

14:47:46 <dc_rdfig> Added comment I1.

14:47:55 <libby> J:see [agenda|http://lists.w3.org/Archives/Public/www-rdf-rules/2003Mar/0035.html]

14:47:55 <dc_rdfig> Added comment J1.

14:48:22 <libby> BLURB: RDF Query tests meet agenda item 4. (K) regularity of meetings, next meeting time and date

14:48:22 <dc_rdfig> K: RDF Query tests meet agenda item 4. (K) regularity of meetings, next meeting time and date from libby

14:48:29 <libby> K:see [agenda|http://lists.w3.org/Archives/Public/www-rdf-rules/2003Mar/0035.html]

14:48:29 <dc_rdfig> Added comment K1.

14:48:59 * danbri considers BLURB:A goal to achieve by (?)WWW2003: 4 query engines tested using test cases manifest format + resultset representation format + ntriple-represented queries

14:49:19 * libby would like that very much

14:52:00 <AndyS> Which 4?

14:52:38 <danbri> whoever gets there first ;)

14:52:57 <danbri> yours, libby's, alberto's and mine would be natural candidates...

14:53:38 <AndyS> Suppose that is done - what effect do you hope it will have? Just a WG?

14:53:45 <danbri> i'd like to cross somesuch hurdle before spending time making the test machinery much more general (eg. supporting constraint types etc)

14:53:49 <danbri> 'just'?:)

14:54:30 <danbri> It would provide some sanity check that the test case machinery was (at some level) usable, and that our engines performed (or didn't perform) in certain ways.

14:54:49 <danbri> It would provide input into WG charter drafting discussions, too

14:55:11 <AndyS> Yes "just" - a WG is a 2y cycle. I think the users prove the usability and your paper at QL98 shows the theory.

14:55:32 <danbri> And maybe more importantly, give us an implemented basis for discussion of richer testing machinery that addressed query facilities that have been implemented but aren't tested by simple ntriples approach

14:55:52 <AndyS> Mine's done (except the n-triple queries)! :-)

14:56:06 <Jeen> hello all

14:56:11 <danbri> Maybe I should go on the record (again -- said this at Tech Plenary) as not being super enthused at chartering a query WG.

14:56:48 <danbri> Getting the test stuff done (for some sense of 'done') would be most of what I'd want from a WG. Actually doing a full W3C WG would add on, as you say, ¬¬¬~2 years, for (imho) debatable benefits.

14:57:06 <AndyS> I saw that - could you say why? Is the testcases your preferred alternative?

14:57:39 <alberto> is back

14:57:52 <danbri> A WG would be great, if folk have the energy to do loads and loads of work

14:57:53 <alberto> sorry I was writing/doing other things

14:57:59 <AndyS> The current proposed WG charter is wider - remote access, result formatting(?) - and a wider set of query systems (e.g. DQL)

14:58:40 <danbri> Test cases provide a cheaper / lower effort way of getting some basic demonstrated interop amongst rdf tools, and stimulating debate about QL features

14:58:48 <danbri> 'current proposed WG charter'?

14:58:51 <Jeen> there's a proposed charter available?

14:58:58 <danbri> Oh, you mean timbl's notes from the TP session

14:59:15 <libby> hey jeen :)

14:59:22 <libby> glad you could make it

14:59:23 <Jeen> hi libby

14:59:48 <libby> ok I make it 3

15:00:03 <libby> --------RDF query testcases meeting starts ------

15:00:13 <libby> thanks for coming everyone who is here for that

15:00:33 <libby> hope the rest of you aren't too annoyed by it....

15:01:08 <libby> the conversation above was interesting.....

15:01:09 <alberto> is trying to catch up

15:01:30 <danbri> E:

15:01:30 <dc_rdfig>http://lists.w3.org/Archives/Public/www-rdf-rules/2003Mar/0035.html

15:01:31 <dc_rdfig> AGENDA for RDF query testcases IRC meeting, Thursday, March 27, 2003, 15:00 UTC

15:01:36 <alberto> the www2003 demo idea sounds cool - for the WG we can work it

15:01:51 <libby> I shoul say that I have to run shortly after 4, apologies for that (although we finish at 4, it's nice to stay later)

15:01:54 <danbri> E:[http://www.w3.org/People/DanBri/ Dan Brickley] attending

15:01:54 <dc_rdfig> Added comment E1.

15:02:07 * danbri encourges attendees to sign up, typing E:... as I did

15:02:16 <danbri> (attendees of this meeting I mean)

15:02:38 <libby> E:[http://ilrt.org/people/libby|libby Miller] attending

15:02:39 <dc_rdfig> Added comment E2.

15:02:48 <libby> while peopel do that, let's start

15:02:58 <libby> 1. Proposed resultset vocabulary and example

15:03:12 <libby> agenda item 1. Proposed resultset vocabulary and example

15:03:13 <AndyS> E: Andy Seaborne

15:03:13 <dc_rdfig> Added comment E3.

15:03:25 <Jeen> E:[http://sesame.aidministrator.nl/ Jeen Broeksta] attending

15:03:25 <dc_rdfig> Added comment E4.

15:03:36 <libby> Andy's done a great job here.

15:03:46 <libby> - http://www.w3.org/2003/03/rdfqr-tests/recording-query-results.html#result_set_vocabulary

15:04:04 <danbri> +1, nice work Andy

15:04:04 <libby> I suggest we spend some time discussing it

15:04:26 <AndyS> Apologies that NS URL does not point anywhere yet - server now exists but isn't set up

15:04:34 <danbri> It's traditional to leap straight into nitpicking ;-)

15:04:46 <AndyS> Go for it Danbri

15:04:50 <Jeen> +1 here

15:04:52 <libby> heh

15:04:52 <danbri> AndyS, do you need any special facilities on server, or just an rdf/xml schema

15:05:05 <alberto> e: Alberto Reggiori

15:05:07 <alberto> E: Alberto Reggiori

15:05:07 <dc_rdfig> Added comment E5.

15:05:19 <Jeen> one nitpick: is the size attribute optional? I think it should be.

15:05:26 <danbri> nitpics: I was curious about the ResultTable==ResultSet synonyms... is that a legacy issue

15:05:39 <AndyS> The NS is on jena.hpl.hp.com - not W3C - and I have to set Apache up to serve different formats. Just time ...

15:05:53 <AndyS> Re: ResultTable==ResultSet ...

15:06:17 <AndyS> No - it is two ways of thinking about it. One more friendly to rules, one more SQL-ish

15:06:21 <danbri> From an RDFS perspective, it's all optional. From a document format design perspective (eg. if we wrote an XML schema for this) I'd suggestion optional

15:06:36 <danbri> which is which?!

15:06:44 * danbri never sure what's rule-friendly

15:07:01 <AndyS> I would be happy with just the "solutions" way : ResultSet

15:07:31 <AndyS> It encodes why a match happened in language that is as good for Euler/cwm as a query

15:07:46 <Jeen> danbri: well from RDFS it might be considered optional, but if a processor is expected to process such a resultset, it's good to have that clear. But it's a minor thing :)

15:08:00 <danbri> I think the value of having just one way outweighs concerns about trying to please everyone with word choice. The rdf:comment could perhaps cover some of this...

15:08:07 <AndyS> What do people think? Are the owl:sameAs equivalents more trouble than they are worth?

15:08:08 <danbri> <AndyS> I would be happy with just the "solutions" way : ResultSet

15:08:09 <danbri> +1

15:08:13 <libby> I think it's a good point jeen...

15:08:14 <danbri> Yes, more trouble!

15:08:22 <libby> +1 andys

15:08:23 <danbri> Jeen, I agree

15:08:44 <alberto> agree

15:08:45 * AndyS wonders how many reasoners will do just own:sameAs?

15:09:02 * AndyS mistyled : s/own/owl/

15:09:07 <danbri> Andy, what are the redundancies at the moment? The 'classes' list confused me a bit

15:09:17 <libby> I propose we drop rs:ResultTable, noting that it is equalivalent to rs:ResultSet

15:09:20 <danbri> rs:ResultRow

15:09:20 <danbri> same as rs:ResultRow ? synonym for table view of results

15:09:21 <AndyS> That's an edit I like - remove it!

15:09:27 <danbri> 2nded

15:09:36 <libby> any objections?

15:09:43 <Jeen> nope, fine with me

15:09:48 <alberto> fine

15:09:49 <danbri> how to record resolutions?

15:09:52 <danbri> BLURB:Resolutions

15:09:52 <dc_rdfig> L: Resolutions from danbri

15:09:54 <AndyS> The classes are needed as declarations.

15:10:01 <libby> H:RESOLVED: drop rs:ResultTable, noting that it is equalivalent to rs:ResultSet

15:10:02 <dc_rdfig> Added comment H2.

15:10:04 <AndyS> Aren't they oh RDFS editor?

15:10:08 <libby> H:

15:10:08 <dc_rdfig> blurb

15:10:09 <dc_rdfig> RDF Query tests meet 1. (H) Proposed resultset vocabulary and example

15:10:11 <dc_rdfig> (1:libby) see [agenda|http://lists.w3.org/Archives/Public/www-rdf-rules/2003Mar/0035.html]

15:10:11 <dc_rdfig> (2:libby) RESOLVED: drop rs:ResultTable, noting that it is equalivalent to rs:ResultSet

15:10:20 <libby> that ok?

15:10:24 <libby> L:RESOLVED: drop rs:ResultTable, noting that it is equalivalent to rs:ResultSet

15:10:24 <dc_rdfig> Added comment L1.

15:10:25 <danbri> yup

15:10:30 <danbri> other redundancies?

15:11:03 <AndyS> Anything in the N3 with "=" : two items

15:11:43 <AndyS> Lucky its today - Jena2 preview2 code freeze is tomorrow :-)

15:11:47 <danbri> :ResultRow = :ResultSolution ;

15:11:47 <danbri> rdfs:comment "Alternative name for :ResultSolution" ;

15:12:01 <danbri> How about fixing on just one of those?

15:12:13 <danbri> Does either fit with ResultSet better?

15:12:14 <libby> I kinda like row, but not that bothered. one wonly wouldf be good

15:12:20 <AndyS> That can go as well - leave ResultSolution (which is what my code actually uses)

15:12:23 * danbri prefers ResultSolution

15:12:36 <danbri> ('row' sounds a bit too physicalist...)

15:12:55 <libby> but solution seesm like a cell rather than a row to me...

15:13:03 <libby> but as I say, no strong pref really

15:13:06 <danbri> hi arjohn

15:13:09 <AndyS> Table/row go together: Set/Solution go together Set/Solution is better IMHO

15:13:28 <danbri> Solution sounds good to me, sorry lib!

15:13:43 <AndyS> Table/row is a bit implementation (and my implementation isn't a table).

15:13:51 <libby> PROPOSED: keep ResultSolution, remove ResultRow

15:13:52 <dc_rdfig> Label PROPOSED not found.

15:13:55 <danbri> proposed: drop rs:ResultRow, noting that it is equivalent to rs:ResultSolution

15:13:57 <arjohn> Hello all. Just checking out whats happening here.

15:14:03 <danbri> hey, parallel typing.

15:14:20 <libby> hey arjohn, there's a meeting here till 4 a out rdf query. welcometo participate!

15:14:36 <libby> I think we can take it as seconded.

15:14:46 <libby> any comments?

15:14:53 <AndyS> I'll make the changes ASAP.

15:15:00 <Jeen> for those of you who don't know, Arjohn is also a developer on Sesame.

15:15:13 <libby> excellent :)

15:15:15 <alberto> hi Arjohn

15:15:23 <libby> any objections to danbri's proposal?

15:15:24 <danbri> pleased to meet you

15:15:47 <danbri> (tumbleweed)

15:15:50 <Jeen> heh

15:15:59 <libby> H:RESOLVED: drop rs:ResultRow, noting that it is equivalent to rs:ResultSolution

15:16:00 <dc_rdfig> Added comment H3.

15:16:08 <libby> thanks andys :)

15:16:16 <danbri> Thanks Andy, yup

15:16:17 * AndyS wonders why no one complains about the "dc:identifier : ; #:-)"

15:16:24 <libby> any other comments so far?

15:16:28 * danbri takes another look

15:16:35 <danbri> heh

15:17:21 <AndyS> Note the "undef" declaration - even conjunctive queries return undefs (variable in SELECT but no use made of it)

15:17:33 <alberto> dcq:URI perhaps ? ;-)

15:17:36 <danbri> AndyS, re namespace etc., you're v welcome to use a http://www.w3.org/2003/03/rdfqr-tests/...something namespace URI for this. I guess I'd encourage that actually, though I'm sure your own server/services are good too...

15:18:07 <alberto> tried to use that "namespace" for the other document/schemas

15:18:11 <danbri> ...one reason is that it would make migration into Working Group efforts simpler, since if a WG was launched it would almost certainly use www.w3.org URIs rather than those of a Member org...

15:18:45 <AndyS> but wouldn't it be in the QL-WG name tree?

15:19:30 <danbri> not necessarily... eg the RDF test cases are in http://www.w3.org/2000/10/rdf-tests/rdfcore/

15:19:50 <danbri> ...which Art and I worked on prior to RDFCore getting underway

15:19:51 <AndyS> I suspect that my vocabulary isn't sufficiently general as I woudl expect more than squish-like QL's to be in scope. e.g. DQL.

15:20:26 <danbri> I think as long as we write 'status of this document' disclaimers making it clear that we're (at this stage) only testing a subset of RDF query systems, that would be fine...

15:20:36 <AndyS> Maybe it can do DQL - but I didn't have that in mind at the time. Need a DQL person.

15:20:44 <alberto> yes I agree danbri

15:21:05 <danbri> I asked Jeen about this at a previous meeting re RQL. It's important to pick a scoped problem, and I think we're doing that... and to be explicit about that scoping

15:21:20 <libby> I agree very much re scoping

15:21:24 <AndyS> Hmm - it would need "value is a variable" but its possible given URIs are the variables for DQL.

15:21:43 <libby> andys, do you have any objections to changing the namespace?

15:21:45 <danbri> ...ie testing squish-like QLs is a good start on the broader problem set

15:21:47 <libby> DQL asside?

15:22:00 <Jeen> as far as result sets are concerned the current proposal still covers RQL fine.

15:22:13 <libby> excellent jeen, that's good

15:22:17 <danbri> I think we could probably do DQL and XUL with this format or a slight modification, but noting if we couldn't would also be an important and useful finding

15:22:21 <danbri> thanks Jeen, good to know

15:22:41 <AndyS> Sort of - I would hope there would be much publishing of RDFS and if is all W3C - and people who can update CVS - it is a bit closed

15:22:48 <AndyS> Nothing technical against it.

15:22:51 <Jeen> actually we will probably implement this result set format in Sesame as soon as it's more or less fixed

15:22:56 <libby> I dont see why this should not change in response to implementation experience.

15:23:13 * danbri suggests Morbus ask in #foaf for now

15:23:40 <AndyS> Let's wait until it is stable : final publishing can go to a W3 namespace

15:23:46 <danbri> re CVS access, 3 of us here (Libby, AndyS, DanBri) have access, and that could be widened

15:23:50 <AndyS> (hopefully soon!)

15:23:56 <libby> well, andys already ahs an implementation. does anyone else volunteer to try and implement this in their testing?

15:24:12 <libby> I do

15:24:15 <libby> anyone else?

15:24:16 <danbri> I'm all set to switch over to use these vocabs for testing rubyrdf, made progress since last week

15:24:22 <DanC> er... widened at considerable cost, no danbri?

15:24:27 <AndyS> Danbri - I didn't just mean QL schemas - but a world of vocabs about all sorts fo things. Publish stuff!

15:24:42 <Jeen> libby: are we talking about the result set, or the full test case thingy?

15:24:49 <libby> the resultset for now

15:24:57 <danbri> AndyS, I'm only talking about the schemas to support RDF query testing.

15:25:05 <alberto> I will start integration of manifest+results soon

15:25:07 <alberto> good question libby :)

15:25:15 <AndyS> I wil package the RDQL/Jena stuff separately from the Jena2 download.

15:25:20 <alberto> which include manifest + results

15:25:33 <libby> H:ACTION libby implement the resultset format for next meet.

15:25:34 <dc_rdfig> Added comment H4.

15:25:48 <libby> anyone like to action themselves? could be a later date...

15:25:49 <danbri> ie. namespaces that will be in common use to test RDF databases, engines etc. For other stuff, yes folks should use sourceforge etc., no need to be bottlenecked thru desire to have w3.org uris

15:25:49 <AndyS> but it is my RDF manifest at the moment.

15:26:00 * danbri nods to DanC, acks there is some cost

15:26:05 <alberto> I can not promise it for next meeting but i will try to implement it asap :)

15:26:10 <danbri> Agreed, don't want to steal it!

15:26:12 <libby> ok :)

15:26:49 <danbri> I will try to implement before next meeting, can't promise

15:26:56 <libby> great :)

15:26:58 * Jeen echoes danbri

15:27:16 <libby> The implementation experience is the most useful

15:27:21 <libby> can we have actions without dates?

15:27:27 <libby> I'll just add a comment

15:27:53 <libby> H:alberto, danbri and jeen plan to implement resultset vocab soon

15:27:54 <dc_rdfig> Added comment H5.

15:28:12 <libby> so we've got about 5 mins left on this topic - are we done?

15:28:17 <alberto> ok, but with which manifest?

15:28:33 * danbri notes andys took on his other suggestion, adding an rdf:nodeID example

15:28:42 <danbri> I implemented rdf:nodeID in the RubyRDF parser in anticipation :)

15:28:49 <libby> shall we move on to disussion of manifest format?

15:28:52 <danbri> TimBL did Cwm/SWAP too

15:28:56 <AndyS> New version of result set (no result table/ result row)

15:29:00 * danbri nods

15:29:06 <danbri> cool, thanks Andy!

15:29:10 * dajobe [nodeID related how?]

15:29:26 <libby> query meet agenda 2. Proposed manifest vocabulary

15:29:30 * AndyS did cmw as well - for testing!

15:29:39 * AndyS s/cmw/cwm/

15:29:59 * dajobe sees

15:30:03 <alberto> ok libby

15:30:04 <danbri> dajobe, rdf:nodeID needed when the same resource appears at various places in the table of result bindings, yet doesn't have a URI. We can represent that using rdf:nodeID plus AndyS's vocab

15:30:11 <libby> thanks alberto

15:30:27 <libby> vocabulary is here: http://lists.w3.org/Archives/Public/www-rdf-rules/2003Mar/0013.html

15:30:27 <AndyS> Already can be done in the vocab - just needs an example

15:30:44 * danbri finds Alberto's later announcement, http://lists.w3.org/Archives/Public/www-rdf-rules/2003Mar/0036.html

15:31:02 <libby> and alberyo also has a discussion document: http://www.w3.org/2003/03/rdfqr-tests/rdf-query-testcases.html

15:31:34 * danbri wonders if alberto wants to say a few words to introduce this

15:31:34 <libby> I:see [testcases discussion document|http://www.w3.org/2003/03/rdfqr-tests/rdf-query-testcases.html]

15:31:34 <dc_rdfig> Added comment I2.

15:32:04 <libby> I:[and www-rdf-rules announcement|http://lists.w3.org/Archives/Public/www-rdf-rules/2003Mar/0036.html]

15:32:04 <dc_rdfig> Added comment I3.

15:32:06 <alberto> yes sorry..

15:32:37 <alberto> the manifest format is based on Libby and DanBri work includign some suggestions of Andy

15:32:57 <libby> thanks alberto

15:33:03 <AndyS> Minor: some properties are essential to make a testing set, some just informative. Could note which is which.

15:33:11 <libby> you did a bunch of work on this too right?

15:33:20 <DanC> .time UTC

15:33:20 <datum> 2003-03-27T15:33:20

15:34:53 <danbri> I just noticed the use of rdf:parseType="Collection", which I don't think the rubyrdf parser implements yet. I guess I can work around it since you have mf:num properties.

15:35:03 <alberto> the idea is to provide a very basic set of properties and classes to describe a test case for regression tests for various RDF toolkits

15:35:06 <alberto> so, ideally a tool maker should pick up the manifest (like the one of RDF Test Cases), parse it and generate the rdf queries in his/her local native query language

15:35:20 <alberto> yes danbri that's Andy's suggestion to keep ordereded tests

15:35:30 <alberto> but i kept mf:num

15:35:32 <danbri> I:"While modeling the RDF vocabulary we tried to capture the very general case when a given RDF query has several different input sources, multiple query graph patterns and possibly several different output result sets."

15:35:33 <dc_rdfig> Added comment I4.

15:35:55 <danbri> could you expand on what "has several different input sources" means?

15:35:58 <alberto> I think if the parser does not do Collections can still figure out the numbering, but it could be redundant

15:36:02 <libby> "several different input sources" i.e. formats or several graphs?

15:36:12 <alberto> yes that's I should cump my text :-)

15:36:43 <danbri> yes, I wasn't sure of that. Several formats, or just input data split over multiple files?

15:36:51 <alberto> the manifest allows to specify different input source, query expressions and outputs

15:36:52 <danbri> I'm anticipating both scenarios...

15:37:08 <libby> me too

15:37:09 <danbri> an input source being 'source data' to be queried?

15:37:25 <alberto> each on can be in RDF/XMl, N-Triples or N3 (inspired to RDF Test Cases - thanks dajobe)

15:37:33 <Jeen> sorry if this was discussed before, but why mf:num and not simply rdf:ID?

15:37:33 <alberto> yes danbri

15:37:41 <alberto> input source = RDF input

15:37:47 <danbri> Jeen, for ordering I think

15:37:58 <alberto> query expression = triple-patterns (QBE)

15:38:09 <alberto> output = result set (Andy's doc)

15:38:37 <AndyS> I'm not sure about inline data because it is either quoted, reified or else asserted.

15:38:45 <danbri> So the 'several' works a bit differently for data files versus query files. For data files, we believe them all to encode identical RDF graphs. For query files, they are similar queries, but because expressive power of QLs differ, do we guarantee they're the 'same'?

15:39:03 <alberto> Jeen, mf:num is coming from "legacy" stuff and it could be skipped if using (automatic) ordering via rdf:parseType="Collection"

15:39:09 <AndyS> Also, isn't the syntax type a mime-type negotiation thing? Just a URI here?

15:39:41 <danbri> AndyS, in principle we could do it that way. But there is a big cost associated with getting real mime-type registrations for each (sometimes drafty) QL proposal.

15:39:51 <danbri> oh sorry, thought you meant query syntax

15:39:53 <danbri> not rdf syntax

15:40:04 <AndyS> Not query syntax - data

15:40:12 <danbri> I don't think we have a mimetype for ntriples

15:40:18 <dajobe> hehe

15:40:25 <alberto> danbri, allowing different inputs, query patterns and outputs we might have the problem of graph equality and stuff like that

15:40:25 <danbri> also if bundled as a tar.gz and accessed locally, no http conneg

15:40:30 <AndyS> URLs with .nt ? :-)

15:40:38 * danbri grins

15:40:39 <dajobe> "The Internet Media Type / MIME Type of N-Triples is text/plain and the character encoding is 7-bit US-ASCII." -- rdf test cases

15:41:02 <AndyS> So programs can't tell :-(

15:41:15 <AndyS> Same for N3 isn't it?

15:41:21 <alberto> that's why there is qt:inputDocument with rdf;type="&tq;NT-Document"

15:41:24 <dajobe> for rdf test cases, the manifest tells you what the urls at the other end should be

15:41:28 <dajobe> N3 - dunno

15:41:36 <dajobe> that's a research lang, ask the developers

15:42:11 <AndyS> Guess we have to keep the data typing - I hoped it could be converted by Apache as needed.

15:42:18 <alberto> you hard code tq:NT-Document, tq:RDF-XML-Document and tq:N3-Document - all the rest is octed-stream (data)

15:42:25 <danbri> yes, rdf;type="&tq;NT-Document" typing seems cheaper than content neg approach at moment

15:42:49 <dajobe> it also works with static files

15:43:19 <alberto> and the software can croak if it does not know how to parse the input query test case :)

15:43:20 <AndyS> but it binds it to the manifest. Can't change it (cache it) in other formats. But there is only RDF/XML :-)

15:43:55 <AndyS> (N-triples is a WG lang)

15:44:00 <danbri> Do you folks feel a need to represent lossiness when translating from Squish/RDQL/etc into the ntriple query representation?

15:44:29 <danbri> ie. there are tools like libby's (@@url) which can convert some formats into others (ntriples into most; squish/rdql into ntriples)...

15:44:38 <danbri> ...but sometimes we lose expressivity

15:44:54 <libby> or, we shosse testcases thatmean we dont lose expressivity

15:45:05 <alberto> the problem of having a full RDF/XML, N* format for the manifest/results test cases is that it requires to have a "working" RDF/XML parser ahead

15:45:08 <alberto> I would suggest "open" rdf:type thingies for dialects

15:45:08 <danbri> I'd like _ultimately_ to represent that lossiness

15:45:10 <libby> -lowest common denominator type

15:45:16 <AndyS> What gets lost? (other than value constraints and the precise namepsace declarations and the projects)

15:45:35 <alberto> I mean if you want to encode a query in SquishQL or RDQL with constraints (AND part) you have to say that is "special"

15:45:37 <AndyS> But certainsly focus on the graph matching part and just do testing for that.

15:45:44 <dajobe> the rdf core test cases approach is the that manifest file is also a regular XML doc, in that you can parse it with perl or regexes if you like

15:45:57 <danbri> value constraints, any maybind/mustbind stuff (xul, dql, algae)

15:45:59 <dajobe> regular=I mean, same format, no rdf/xml abbreviation variants

15:46:00 <libby> well no clear way of doing the variables you actually want

15:46:07 <libby> may/must is a big issue

15:46:11 <alberto> we can instead focus on a kind of "transformation format", which simply does graph matching

15:46:21 <AndyS> Its teh RDF part - the constrains depend on value calculations (e.g. arithmetic on xsd:integer)

15:46:31 <dajobe> would rdf datatypes help mark some things as special?

15:46:34 <danbri> also issue of which variables are requested for the resultset, rather than just there as link-nodes. Not sure how important that is yet.

15:46:42 <alberto> without fanciness like constraints - i.e. RDF in ----> RDF out

15:47:12 <AndyS> +1 RDF in , RDF out

15:47:23 <danbri> +1

15:47:34 <alberto> people will invent their own format anyway :-)

15:47:35 <AndyS> If want projecrt variables, then the query format isn't a template for N-triples.

15:47:47 <AndyS> because need to annotate the query .

15:47:51 * dajobe could slip ?vars into ntriples, it is a test language after all ...

15:47:54 <alberto> we can not stop them :-) that's why XSLT works

15:47:56 * danbri didn't want to get into talking about the query formats, but slipping into it by virtue of talking about representation of query formats in the manifest

15:48:00 <alberto> tree-to-tree

15:48:30 <AndyS> So is it proposed : N-triples+variables (?name)?

15:48:34 <danbri> Is it possible to use this format with query-by-example style?

15:48:43 <libby> this is a topic for next week. have we finished with the manifests topic?

15:48:49 <alberto> we should provide tests about querying RDF as a kind of "tranformation" IMO (sub-graph matching)

15:48:51 <danbri> ...sorry i mean in rdf/xml, with (as yet unspecified) decorations on the graph

15:48:56 <AndyS> Alternative is (non-pure-RDF) N3 formulae - I coudl try that for dicussion if people are interested.

15:48:58 <DanC> er... ?name only works inside {} in N3. Pls don't use that syntax at the top level.

15:49:00 <danbri> ...would the manifest format itself need to change for that?

15:49:12 <alberto> I thinkwe should drop '?' in RDQl/SqushQL anyway :-)

15:49:22 * danbri doesn't see urgent need to extend ntriples

15:49:28 <AndyS> If RDf/XMl (c.f. DQL variables) then issues on variables for predicates.

15:49:47 <danbri> hmm we could adopt a convention that (a la prolog) capitalisation matters

15:49:50 <alberto> we are having technical problems to pass RDQL queries with '?' in vars thourgh ODBC

15:50:00 <danbri> eg _:person foaf:age _:A

15:50:10 <alberto> variable names maps to _:varname

15:50:11 <danbri> ...would have A as variable

15:50:12 <AndyS> Surely need to extend N-triples because _:a are graph values WRT a query?

15:50:22 <dajobe> my suggestion was <foo> <bar> "variable"^^<someURIhere>

15:50:32 <alberto> what's the format supposed to be for rdf:nodeID ?

15:50:47 <danbri> its a slightly perverted _use_ of ntriples, not yet sure ntriples needs syntactic extension

15:50:55 <dajobe> lol

15:51:12 * danbri wonders where we are in the agenda

15:51:14 <alberto> not XMLQNAME, so we can use more or less fanciness to name bNodes ?

15:51:14 <AndyS> Need to separate variables fro RDF to values (bNodes, existential variables) and query variables (universals)

15:51:27 <AndyS> Could predeclare c.f. cwm.

15:51:31 <alberto> back to the manifest....

15:51:32 <libby> guys, we have 10 minutes left and 2 agenda items to go

15:51:51 <alberto> I will try to add more examples to the document (or links them)

15:52:02 <AndyS> Good point - what else on manifest?

15:52:04 <danbri> re manifest, I wanted to say that I much prefer the non-inlined version

15:52:15 <danbri> ...it keeps it a nice simple table of contents format

15:52:21 <libby> re in-line, me too

15:52:21 <AndyS> +1 - keep it simple, agree it, extend later

15:52:35 <alberto> one is gonna be about "inlining" input source, query exp and output i.e. self-contained manifest/test-case

15:52:57 <libby> alberto, why do you like in-line particularly?

15:53:24 <alberto> I wanted originally for the HTML FORM interface to it

15:54:00 <libby> explain a bit more?

15:54:26 <danbri> so someone could upload a complete thing in a single transaction?

15:54:49 <libby> a complete manifest?

15:54:56 <danbri> +everything it needs

15:55:05 <alberto> ie. if you submit some text via a Web interface for a test case you need to create a URL for it (that's easy) but most people what to have one file with everything in it....

15:55:08 <alberto> no just that...

15:55:09 <danbri> (at which point imho it stops being a manifest, and becomes a packaging format)

15:55:22 <alberto> like in the RDF Query/Rules survey of me and Andy you can submit text by typing it into the TEXTAREA which goes as CDATA into the "manifest"/testcase

15:55:26 <libby> ok well look, times coming on

15:55:32 <libby> right

15:55:36 <AndyS> Coudl the server allocate a URLK for an uploaded file? Or split out part of the form?

15:55:43 <alberto> but 90% of the manifest uses will be with 'file:///' style URLs

15:56:03 <alberto> we can drop it if you want

15:56:21 <danbri> So we can observe that RDF itself allows all sorts of stuff to be inlined within a given RDF/XML document, but that one common use of the Manifest format we advocate is as a 'table of contents' xml document with no inlining.

15:56:31 <AndyS> Libby - do you want to squeeze in agenda 5?

15:56:40 <danbri> Again this distinction between what the basic RDF vocabulary imposes (not much) versus document format design choices.

15:56:44 <libby> +1 danbri

15:56:51 <alberto> it is ok danbri

15:57:01 <AndyS> s/5/4/!

15:57:09 <libby> do we have implementation volunteers?

15:57:14 <libby> for the manifest?

15:57:16 <alberto> so, we can drop the inlining idea - who is against it but me? :-)

15:57:18 * libby volunteers

15:57:19 <danbri> Yes! I do...

15:57:42 <libby> alberto, I think implementation is they key - ease of imleemntation

15:57:42 <AndyS> Will see if I can make changes

15:57:46 <alberto> I will implement the manifest together with the results set if possible...

15:57:47 <AndyS> c.f. http://cvs.sourceforge.net/cgi-bin/viewcvs.cgi/*checkout*/jena/jena2/vocabularies/manifest-vocab.n3

15:57:47 <danbri> +1 on dropping inlining for manifest doc format (but can mention that some may find it a useful technique to use)

15:58:02 <libby> you done it andys? :)))

15:58:06 <danbri> I plan to implement both manifest format and result set format

15:58:07 <libby> nice one

15:58:10 <libby> me too

15:58:16 <AndyS> My old style - not Albteros new one

15:58:29 <libby> I:danbri and libby plan to implement manifest format

15:58:30 <dc_rdfig> Added comment I5.

15:58:30 <AndyS> Proves concept but not current proposal.

15:58:36 <libby> right, cool

15:58:54 <alberto> +1 danbri, and they can problably use RDF itself (parseTypes/CDATA) to make fancier things

15:59:00 <AndyS> (It is in Jena2's JUnit main test package)

15:59:00 <danbri> yup

15:59:05 <libby> ok, skip 3 for now (postpone), quick discussion about 4

15:59:11 <libby> 4. regularity of meetings, next meeting time and date

15:59:19 <alberto> I will start my "my" new one (manifest)

15:59:27 <AndyS> Time/date OK for me but what about West Coast USA?

15:59:40 <danbri> I would like a regular timeslot for RDF query testing chat, for a short period (< 6 months) and some agreed goal(s)

15:59:45 <libby> I:alberto will start my "my" new one (manifest)

15:59:45 <dc_rdfig> Added comment I6.

16:00:00 <libby> I have a feeling this is not a great time

16:00:01 <danbri> 1 or 2 hours later would be fine by me. Better even.

16:00:13 <libby> shall we try 16:00 UTC?

16:00:33 <AndyS> does it miss anyone who has expressed interest? Note starts to impact Europe the other way and we do have Jeen/Arjon

16:00:37 <libby> that makes it 5-6 in continental europe

16:00:40 <Jeen> that is not great for me to be honest

16:00:42 <danbri> 2nd'd

16:00:45 <libby> yep, and jos sometimes

16:00:57 <libby> darn

16:00:58 <danbri> England's in Europe too ;-) (but yes, 1hr more American)

16:01:25 <libby> I think it's important jeen and arjon and jos can make it

16:01:33 <danbri> 15:30 UTC? for 1hr

16:01:38 <AndyS> (If I said "mainland Europe" you would have picked that up as well!)

16:01:42 <alberto> ok for me

16:01:45 <Jeen> deal

16:01:52 <danbri> (I sometimes have a prior meeting that finishes 15:30 UTC)

16:01:53 <libby> great!

16:02:07 * DanC combs webont records for outstanding dissent...

16:02:09 <Jeen> in two weeks?

16:02:16 <arjohn> 15:30 is fine with me, just can't promise I'll stay till the end

16:02:19 <AndyS> 15:00 or 15:30 is OK but prefer to move with Euprope time change

16:02:29 <libby> is everyone ok with every 2 weeks

16:02:32 <AndyS> 15:00 looks better for people

16:02:45 <danbri> Thru end May 2003, I could do 2-weekly.

16:03:02 <AndyS> Thurs is fine for me.

16:03:16 <libby> ok, I propose 3.30, thursdays, every 2 weeks until end may 2003

16:03:18 <alberto> fine by me too

16:03:20 <libby> for one hour

16:03:27 <Jeen> oh drat, i forgot about daylight savings time, that's next week right?

16:03:31 <danbri> Yeah

16:03:44 <AndyS> Luckily we skip the week US and Euorpe are different.

16:03:51 * danbri incapable of thinking clearly about timezones. decide something and I'll agree!

16:03:53 <libby> phew!

16:03:54 <danbri> ok

16:04:05 <libby> any objections?

16:04:15 <libby> ....

16:04:18 <libby> ....

16:04:20 <Jeen> I'm too confused to object, but I'll just try to make it :)

16:04:21 <libby> ...

16:04:24 <dajobe> (that's 16:30 UTC next time)

16:04:25 <libby> heh

16:04:33 <libby> ah yes.

16:04:35 <libby> ok

16:05:25 <danbri> resolved?

16:05:39 <alberto> what time then?

16:05:39 <libby> K:RESOLVED: to meet at 16:30 UTC every 2 weeks until end may 2003

16:05:39 <dc_rdfig> Added comment K2.

16:05:48 <danbri> i wanted to mention possibility of a WWW2003 BOF, for those of us who will be attending the Web conference...

16:05:52 <libby> hehe, let me get a link

16:06:00 <libby> ---ajourned

16:06:01 <AndyS> I'm still here ...

16:06:20 <libby> I will find a link so that everyone can tell, and will send around a reminder about the time and date

16:06:35 <danbri> I sent the organisers an inquiry, prelim request for a slot on the thursday at lunchtime. Basically it would just get us use of a room that we could wander back to with our (buffet) lunch, and have a working lunch.

16:06:50 <danbri> No response yet. I'll mail www-rdf-rules soon as I hear anything...

16:07:00 <AndyS> BOF: Seems like a good thing: is there a theme in mind danbri?

16:07:01 <dc_rdfig> Label BOF not found.

16:07:43 <danbri> theme would be RDF Query testcases (under vaguely RDFIG/SWAD-Europe umbrella, ie. I'd feel some duty to have a decent meeting record)

16:07:46 <alberto> I have not plans yet for www2003 but I will try to arrange for it

16:08:08 <danbri> alberto, that'd be great! note that earlybird (ie. cheap) registration ends soon (tommorrow?)

16:08:32 <libby> 31st I think

16:08:45 * Jeen will have to miss www2003 :/

16:08:45 <danbri> AndyS, were you planning to attend?

16:08:47 <danbri> Anyone else?

16:08:51 <danbri> Jeen :(

16:09:41 <alberto> I will let you know if I am gonna be there - cheers

16:09:58 <danbri> great, thanks

16:10:00 * danbri heads off

16:10:09 <libby> K2:RESOLVED to meet 14:30 UTC every 2 weeks, thursdays until end May 2003

16:10:10 <dc_rdfig> Replaced comment K2.

16:10:12 <AndyS> I hope to be at WWW2003 but the HP travel situation is "confused" :-(

16:10:32 <danbri> hope it works out, AndyS. It'd be useful to get folks together to work on this stuff...

16:10:38 <libby> K:[http://www.timeanddate.com/worldclock/fixedtime.html?day=10&month=4&year=2003&hour=14&min=30&sec=0&p1=0|14:30 UTC]

16:10:38 <dc_rdfig> Added comment K3.

16:11:06 <eikeon> Is UTC and Europe/London timezone still in sync after dst? aka is this right: http://eikster.com/2003/events/?start=2003-04-10

16:11:12 <libby> gotta to run. thanks all

16:11:37 <libby> not sure eikster any hepl appreciated. we were going for 3.30pm in london, starting 2 eweeks time

16:11:45 <AndyS> Change in Europe is this weekend.

16:12:08 <AndyS> What about the US? Week after? Usually is end MArch/ first April weeks.

16:12:19 <AndyS> Thanks Libby

16:13:42 <AndyS> Danbri - BOF - do you have a focus in mind? Might be an ide to have both general and a theme (like common representation of *simple* subset).

16:15:39 <danbri> gota run. would want to focus on hooking up testcases to real data and engines, and whether a common representation can be shown to have been implemented across several of us

16:15:43 <danbri> bye!

16:15:47 * danbri -> ilrt

16:15:55 <AndyS> Bye Danbri

16:15:55 <alberto> waves

16:15:58 <alberto> ciao

16:17:34 <Jeen> alright, I'm off too. it was a pleasure.

16:17:53 <alberto> bye Jeen

16:25:58 <dajobe> dajobe has changed the topic to: RDF/SemWeb 24x7 hack'n'chat http://rdfig.xmlhack.com/

16:28:47 <taazzzz> taazzzz is now known as taaz

16:32:19 <larsbot> anyone know of a good way to check an OWL/RDFS schema for reasonable consistency?

16:35:35 <dajobe> I think there should be OWL validators out there, DAML ones definitely, rdfs ones yes

16:36:03 <dajobe> in fact you just missed the sesame people who do RDFS validation

16:36:11 <dajobe> "validation"

16:36:28 <larsbot> I'll have a look at Sesame, then. thanks! :)

16:36:48 <dajobe> I think there are online demos for it. Maybe also for VRP / RDFSuite

16:37:32 <ericP> i think there's a DL validator that does OWL, or at least DAML

16:37:43 <ericP> can't remember the name, think it's two letters

16:37:56 <dajobe> how about the horrocks thing

16:37:58 <dajobe> FaCT?

16:38:01 <chaalsMEL> chaalsMEL is now known as chaalZzz

16:38:16 <ericP> yes, those were the two letters i was thinking of

16:38:47 <ericP>http://www.cs.man.ac.uk/~horrocks/FaCT/

16:38:47 <dc_rdfig> M: http://www.cs.man.ac.uk/~horrocks/FaCT/ from ericP

16:38:55 <larsbot> even better :)

16:39:13 <dajobe> but that's probably OWL-DL only

16:39:22 <dajobe> and don't ask me to explain what that means

16:39:42 <dajobe> maybe by xml europe, I'll have worked it out

16:40:16 <larsbot> I'll use the empirical method

16:40:24 <dajobe> dangerous

16:40:32 <dajobe> the w3c's rdf validator allows all sorts of things through

16:40:40 <dajobe> I bet it'll pass any XTM file

16:41:13 <larsbot> found http://owl.bbn.com, which appears to be pretty good

16:41:16 <larsbot> it's found one problem already

16:41:46 <dajobe>http://owl.bbn.com/

16:41:47 <dc_rdfig> N: http://owl.bbn.com/ from dajobe

16:42:03 <dajobe> N:|BBN OWL site including an [http://owl.bbn.com/validator/|OWL Validator]

16:42:03 <dc_rdfig> Titled item N.

16:43:31 <larsbot> uh, actually, I'm beginning to wonder whether the problem is me or the validator

16:44:23 <dajobe> maybe email them and ask

16:44:29 <larsbot> this one seems better: http://phoebus.cs.man.ac.uk:9999/OWL/Validator

16:45:20 <taaz> taaz is now known as taaz-away

16:46:11 <dajobe>http://phoebus.cs.man.ac.uk:9999/OWL/Validator

16:46:11 <dc_rdfig> O: http://phoebus.cs.man.ac.uk:9999/OWL/Validator from dajobe

16:46:30 <dajobe> O:|OWL Ontology Validator from Bechhofer & Voltz

16:46:30 <dc_rdfig> Titled item O.

16:46:58 <dajobe> O:nice to see the three OWL versions(?)

16:46:59 <dc_rdfig> Added comment O1.

16:47:11 <dajobe> O:some example links would be nice

16:47:11 <dc_rdfig> Added comment O2.

16:48:33 <dajobe> O:let's try with the [http://www.w3.org/TR/owl-guide/wine.owl|Wine ontology]

16:48:33 <dc_rdfig> Added comment O3.

16:48:35 <dajobe> O:oh dear

16:48:36 <dc_rdfig> Added comment O4.

16:49:19 * em suggests turning recursion download off

16:49:30 * em thinks he may be too late...

16:50:26 <larsbot> N:Appears to be problems with this one. The foaf schema makes it very unhappy.

16:50:26 <dc_rdfig> Added comment N1.

16:50:27 <dajobe> O4:it's OWL Full, and not any of the others - but fails to import an example.org ontology

16:50:28 <dc_rdfig> Replaced comment O4.

16:51:27 <DanC> O:one of a growing list (well, 2 now) of [http://www.w3.org/2001/sw/WebOnt/#Tools|OWL Tools]

16:51:27 <dc_rdfig> Added comment O5.

16:55:08 <larsbot> anyway, now I have a seemingly-ok schema for my RDF-to-TM mapping vocabulary at http://psi.ontopia.net/rtm/

16:55:48 <dajobe> argh, what a chump mess

16:56:08 <dajobe> N:|BBN OWL site

16:56:08 <dc_rdfig> Titled item N.

16:56:18 <dajobe> N:incluing an [http://owl.bbn.com/validator/|OWL Validator]

16:56:19 <dc_rdfig> Added comment N2.

16:56:22 <dajobe> N2:including an [http://owl.bbn.com/validator/|OWL Validator]

16:56:22 <dc_rdfig> Replaced comment N2.

17:07:28 <AndyS> L: [http://www.w3.org/2003/03/rdfqr-tests/recording-query-results.html|Vocabulary changes done]

17:07:28 <dc_rdfig> Added comment L2.

17:08:20 <AndyS> H: [http://www.w3.org/2003/03/rdfqr-tests/recording-query-results.html|Vocabulary changes done]

17:08:20 <dc_rdfig> Added comment H6.

18:51:34 <soccos|away> soccos|away is now known as soccos

19:09:31 <soccos> soccos is now known as soccos|away

19:24:25 * danbri works on hooking parser to rdfcore test cases

20:14:27 <eikeon> Anyone know off hand what google sends for an Accept header when spidering?

20:19:50 <danbri> grep -i google /var/log/apache2/rdfweb.org-combined_log | more

20:19:50 <danbri>

20:19:58 <danbri> -> "Googlebot/2.1 (+http://www.googlebot.com/bot.html)"

20:20:46 <danbri> grep googlebot /var/log/apache2/rdfweb.org-combined_log | grep -v 2.1

20:20:52 <danbri> ...finds nothing

20:21:03 <deltab> grep -i

20:21:09 <danbri> Oh I misread the question, i think

20:21:18 <deltab> oh

20:21:22 <danbri> anyway, that was interesting for me :)

20:22:38 <danbri> try http://www.google.com/search?hl=en&ie=UTF-8&oe=UTF-8&q=googlebot+accept+header&btnG=Google+Search

20:22:51 <danbri> ...and then look at the _cached_ results from some of the header-echo scripts it finds

20:22:59 <eikeon> Yeah... do not think Accept headers normally get logged.

20:23:03 <danbri> eg. (:USER-AGENT "Googlebot/2.1 (+http://www.googlebot.com/bot.html)")

20:23:03 <danbri> (:FROM "googlebot(at)googlebot.com")

20:23:03 <danbri> (:ACCEPT "text/html,text/plain")

20:23:03 <danbri> (:HOST "www-ksl-svc.stanford.edu:5915")

20:23:04 <eikeon> Cool... will try that.

20:23:38 <danbri> GET /dumprequest.html HTTP/1.0

20:23:38 <danbri> Host: rudolf.org.uk

20:23:38 <danbri> User-Agent: Googlebot/2.1 (+http://www.googlebot.com/bot.html)

20:23:38 <danbri> Accept: text/html,text/plain

20:23:38 <danbri> From: googlebot(at)googlebot.com

20:24:51 <deltab> I guess there are other crawlers for PDF, images, etc.

20:26:52 * danbri would imagine it'd show as the same bot

20:38:34 <eikeon> Guess I will find out soon enough... as I now have documents that are returned as rdf/xml or text/html depending on the Accept header -- hopefully rdf/xml spiders are sending Accept headers with application/rdf+xml in them.

20:43:28 <dajobe> google indexes rdf/xml files

20:43:34 <dajobe> mime-typed that is

20:53:11 <danbri>http://www.w3.org/2001/12/rubyrdf/pack/tests/coretests.rb

20:53:11 <dc_rdfig> P: http://www.w3.org/2001/12/rubyrdf/pack/tests/coretests.rb from danbri

20:53:42 <danbri> P:|Testing RubyRDF parser(s) with RDFCore test cases

20:53:42 <dc_rdfig> Titled item P.

20:54:16 <danbri> P:Here are [http://www.w3.org/2001/12/rubyrdf/pack/tests/ts_rdftests-out.txt|results of first tests].

20:54:16 <dc_rdfig> Added comment P1.

20:54:29 <danbri> P:Assuming this is wired up OK, wonder where to start with the fixups...

20:54:30 <dc_rdfig> Added comment P2.

20:54:43 <DanC> pass/fail stats, danbri?

20:55:19 <danbri> Hmm I should've counted the successes!

20:55:45 <danbri> P:grep Mistaken ts_rdftests-out.txt finds 37 'mistakes'

20:55:45 <dc_rdfig> Added comment P3.

20:56:47 <danbri> P:I'd get better stats out of the testing machinery if I didn't treat this as one huge test with a lot of assertions (see [http://rdfig.xmlhack.com/2003/03/26/2003-03-26.html#1048691883.680525|recent puzzling over testing techniques]).

20:56:47 <dc_rdfig> Added comment P4.

20:57:43 <danbri> P:This is solely in terms of accepting/rejecting legal or illegal files. Haven't got to number of triples yet, let alone whether the triples were right.

20:57:43 <dc_rdfig> Added comment P5.

20:57:49 * DanC updates ical2rdf.pl w.r.t. a few new properties...

20:58:13 <DanC> P:note number of triples isn't relevant. graph matching is all that matters.

20:58:13 <dc_rdfig> Added comment P6.

20:58:48 <danbri> P:I wanted a sense of whether this was going to be a very expensive exercise (in which case I'll concentrate on calling Redland's parser) or whether with a little effort this is a usable parser.

20:58:48 <dc_rdfig> Added comment P7.

20:59:25 * danbri nods re graph matching; triple counting is a cheapscate stopgap

20:59:31 <DanC> ah

20:59:53 <dajobe> it wasn't very accurate in my experience

21:00:04 <dajobe> after i did proper triple counting, there were lots of things to fix

21:00:10 <dajobe> er, graph matching

21:00:23 <danbri> Do you use ntc or the Jena one? ntc wasn't it...?

21:00:33 <dajobe> ntc is jang's

21:00:38 <danbri> yup

21:00:59 * dajobe lol at using Jena to test things. I want it done a bit faster than that :)

21:04:39 * danbri checks out rdf-tests dir

21:06:48 <danbri> yes dave, your code is very fast ;)

21:07:31 <dajobe> it's not that, I want 'make check' to run fast

21:07:56 <DanC> hm... I wonder about scraping an RDF schema from RFC2445... maybe with sgrep...

21:08:55 <DanC> or maybe elisp...

21:22:03 * DanC making progress in perl...

21:22:05 <DanC> property name: ACTION in section 4.8.6.1 Action

21:22:05 <DanC> ... type TEXT

21:22:05 <DanC> property name: REPEAT in section 4.8.6.2 Repeat Count

21:22:05 <DanC> ... type INTEGER

21:24:26 <libby> <DanC> hm... I wonder about scraping an RDF schema from RFC2445... maybe with sgrep..

21:24:29 <libby> - inetrestign plan

21:24:56 <libby> J:postponed

21:24:56 <dc_rdfig> Added comment J2.

21:25:40 <DanC> [ a r:Property; s:label "Any"

21:25:41 <DanC> s:isDefinedBy [ s:label "4.8.8.1 Non-standard Properties"

21:25:41 <DanC> :valueTypeName "TEXT"

21:25:41 <DanC> ].

21:25:52 <DanC> [ a r:Property; s:label "LAST-MODIFIED"

21:25:52 <DanC> s:isDefinedBy [ s:label "4.8.7.3 Last Modified"

21:25:52 <DanC> :valueTypeName "DATE"

21:25:52 <DanC> ].

21:26:42 <DanC> [ a r:Property; s:label "ORGANIZER";

21:26:43 <DanC> s:isDefinedBy [ s:label "4.8.4.3 Organizer" ];

21:26:43 <DanC> :valueTypeName "CAL";

21:26:43 <DanC> ].

21:26:57 <libby> heh, neat :)

21:28:44 <danbri> :)

21:30:17 <DanC> bug:

21:30:18 <DanC> [ a r:Property; s:label "DTEND";

21:30:19 <DanC> s:isDefinedBy [ s:label "4.8.2.2 Date/Time End" ];

21:30:19 <DanC> :valueTypeName "The";

21:30:19 <DanC> ].

21:30:31 <libby> darn

21:30:36 <DanC> Value Type: The default value type is DATE-TIME. The value type can

21:30:36 <DanC> be set to a DATE value type.

21:35:40 <DanC> fixed...

21:35:41 <DanC> ical:dtend a r:Property; s:label "DTEND";

21:35:41 <DanC> s:isDefinedBy [ s:label "4.8.2.2 Date/Time End" ];

21:35:41 <DanC> ].

21:36:25 <DanC> using interpretation properties to relate ical properties to their value types...

21:36:27 <DanC> ical:description a r:Property; s:label "DESCRIPTION";

21:36:27 <DanC> s:isDefinedBy [ s:label "4.8.1.5 Description" ];

21:36:27 <DanC> s:subPropertyOf valueType:text;

21:36:27 <DanC> ].

21:36:59 <DanC> hmm... maybe that's not right...

21:38:05 <bitsko> DanC: re BitTorrent, yes.

21:38:24 <DanC> and... what do you think of it, bitsko?

21:39:19 * danbri commits new http://www.w3.org/2001/12/rubyrdf/pack/tests/ts_rdftests-out.txt

21:39:40 <danbri> ...fixed up nodeID appearing where there's an about or ID bug

21:39:45 <danbri> (thanks to the rdfcore tests)

21:39:55 <bitsko> it works just fine for "ordinary" downloads (ie. no faster than just reading off the site itself), and has been seen to get amazing download speeds with simultanous downloads, with very little load on the original server

21:39:59 * danbri suitably smug, goes to pub

21:40:38 <DanC> "simultaneous downloads"?

21:40:53 <bitsko> the biggest problem is mostly social, but the tools don't do much to help. downloads would be faster if clients stayed connected longer

21:41:39 <DanC> has anybody looked at connecting it to, say, the debian package system? I'd be pretty happy to share the packages I have cached

21:42:08 <DanC> I guess Mandrake 9.x isn't far off

21:42:14 <bitsko> BitTorrent is designed to push downloading "to the edge", and it does it by sending fragments out to clients, then expecting clients themselves to share the fragments among themselves. more clients, faster downloads

21:42:46 <DanC> seems pretty ideal for digital libraries, e.g. RFC mirrors.

21:44:01 <bitsko> exactly.

21:44:22 <bitsko> the one tool currently available kinda makes it difficult to just set up a repository to share, though

21:44:41 <DanC> in fact, it would work well for any ftp mirroring situation, no? i.e. in stead of publishing a list of ftp sites and having humans play hunt-the-fast-one, you could just publish the original address and have the sites that were serving as "mirrors" just act as bittorrent clients.

21:44:53 <bitsko> yup

21:45:39 * DanC noodles on using it in the simile project, on W3C tech reports

21:45:57 <bitsko> there's also scale-down possible too, for allowing small sites (weblogs, wikis, small orgs) to handle large peaks in demand (/.)

21:46:26 <bitsko> also too for the aggregator issue

21:46:47 <bitsko> at that level, tho, it needs to be built into the browsing and aggregator tools

21:46:49 <DanC> you need ubiquitous deployment to handle the slashdot effect, yes?

21:47:40 <DanC> tricky to get this sorta stuff deployed... it competes with akamai, and its marketing/sales budget is, well, a little smaller ;-)

21:47:41 <bitsko> yes, but "downloads fastest in Mozilla" is a pretty good marketing meme ;)

21:48:40 <DanC> mozilla groks?

21:48:48 <bitsko> one of the biggest difference is that BT is self-organizing and free whereas Akamai is pre-cooked and $$$$

21:49:00 <bitsko> no, that's a suggestion

21:49:49 <DanC> yes, BT is self-organizing, but there's a chicken/egg problem. i.e. if I'm suffering from the slashdot effect, either as client or server, it doesn't help (much) for me to install BT in my system. It only helps if a large part of the slashdot audience plays.

21:50:06 * bitsko notes that Akamai supports XSLT and Java at the edge these days, in case anybody doesn't follow that closely

21:50:32 <DanC> "XSLT and Java at the edge"? what does that mean?

21:51:42 <DanC> self-organizing systems are *much* more robust once deployed, but getting them over the hump isn't guaranteed to ever happen.

21:52:03 <bitsko> taking XSLT first: if your site is XSLT based, you can send out your transforms and stylesheets, and Akamai will render at the edge, caching dynamic portions or not (as it supports with ordinary HTML too)

21:52:15 * DanC considers the fact that W3C is just the sort of endorsement machine that can get this sort of thing over the hump...

21:53:00 <DanC> render at the edge: wow! what a cool use-case for XSLT! (mumble 'not a general purpose transformation language' mumble)

21:53:01 <bitsko> with Java, it's mostly a "push" deployment, where one can push Java versions of code to run at the edge, and just talk back to the central site

21:54:44 <bitsko> the current BT tools would handle any current mirroring situation with little effort, and with little effort on downloader's parts as well.

21:55:19 <DanC> it's clearly cost-effective at scale...

21:55:24 <bitsko> all that would take would be for a RedHat, kernel.org, or Debian to start offering the service. others will quickly follow

21:55:47 <DanC> you say the server-side tools are kinda icky as yet?

21:55:58 <bitsko> the tools need to mature some before it can be used for smaller scale uses or for smaller archives

21:56:32 <DanC> I find myself downloading from sourceforge as often as anyplace, these days. Perhaps they have the right mix of capability and motivation.

21:57:08 <bitsko> no, the tools currently seem to favor either a single static file or a static directory structure. it's not a drop-in replacement for, say, an ftp or http server

21:57:10 <DanC> sourceforge and debian, for me.

21:57:38 <bitsko> BT wouldn't be able to handle SF well, because of the lack of dynamic updates

21:57:56 <bitsko> I believe it's a tool problem, not a protocol problem

21:58:20 <DanC> I was gonna say... (a) no updates is a *big* limitation, and (b) I didn't see it in the protocol.

21:59:02 <DanC> just a matter of recomputing the checksums and keeping track of who's got what version, no?

21:59:14 <bitsko> afaict

21:59:48 <DanC> SF downloads... they're pretty static, no?

22:00:02 <DanC> maybe not latest.tgz

22:00:18 <bitsko> for single files (or dirs? haven't found or tried a dir), BT client is nice, just pass it the URL of the .torrent and download.

22:00:45 <bitsko> SF downloads are fairly static, but the environment as a whole is very dynamic

22:00:53 <DanC> have they considered an HTTP header to point from the original content to the .torrent?

22:01:14 <bitsko> one would want to run "a" BT server and mirrors for SF, not one for each project

22:02:01 <bitsko> they recommend not putting the original file on the net at all. only the .torrent is a public resource

22:02:06 <DanC> so you could do GET the first, say, 6kbytes... if you see a BT: xyz.torrent , you stop there and go into BT mode

22:02:26 <DanC> huh? the original file has to get out there somehow, no?

22:02:45 <bitsko> out-of-band

22:03:21 <DanC> hmm... I thought this was a supplement to normal HTTP.

22:04:09 <bitsko> the way to boot-strap a torrent is to 1) run the init utility to create the .torrent, giving it the URL of the tracker, 2) start a seed server with an already-local copy of the file

22:05:22 <bitsko> it should be, and maybe could be

22:12:51 <bitsko> since I don't know much about the various p2p tools out there, I don't know if BT is doing anything different.

22:13:38 <DanC> I think it is. other tools seem to use just one peer for the bulk transfer most of the time.

22:31:05 <golbeck_zzz> golbeck_zzz is now known as golbeck

23:19:30 <DanC> eek! GET with side effects... http://esw.w3.org/topic/BreakingURIOpacity?action=subscribe

23:20:33 <danbri> uhoh

23:22:41 <danbri> re bittorrent, i noticed that kazaa (sp?) P2P apps use HTTP byte range requests to do parallel downloads from multiple srcs

23:28:03 <deltab> as I understand, most p2p systems transfer whole files from peer to peer

23:29:54 <bitsko> Kazaa is based on Gnutella, no?

23:31:13 <danbri> I think its their own protocol, or was

23:34:15 <deltab> bitsko: people from redhat and kernel.org have expressed interest

23:35:06 <bitsko> what do they need to get started?

23:35:47 <deltab> the source files, somewhere to host the tracker, and distribution of the torrent (metadata) file

23:36:16 <deltab> there are plans to distribute next week's redhat 9

23:37:10 <deltab> a trial run for that is handling the three ISOs of mandrake 9.1: the tracker's at http://sucs.org:6969/

23:46:30 <libby>http://lists.w3.org/Archives/Public/public-esw/2003Mar/0007.html

23:46:31 <dc_rdfig> Q: http://lists.w3.org/Archives/Public/public-esw/2003Mar/0007.html from libby

23:46:55 <libby> Q:|Ideas for SWAD-Europe publicity plea.

23:46:56 <dc_rdfig> Titled item Q.

23:49:00 <libby> Q:+[saucy postcard idea|http://dannyayers.com/images/postcard.jpg] delighful saucy postcard idea from Danny Ayres ;)

23:49:00 <dc_rdfig> Added comment Q1.

23:50:03 <libby> Q:we're also wondering about a map of semnatic web researchers in Europe.

23:50:03 <dc_rdfig> Added comment Q2.

23:54:42 <taaz-away> taaz-away is now known as taaz