Semantic Web Interest Group IRC Chat Logs for 2003-04-24

00:03:37 <Morbus> hey Aaron.

00:48:54 <danbri>http://www.guardian.co.uk/online/news/0,12597,942024,00.html

00:48:54 <dc_rdfig> A: http://www.guardian.co.uk/online/news/0,12597,942024,00.html from danbri

00:49:25 <danbri> A:|Battle of the blog builders, Ben Hammersley (Guardian Online, 2003-04-23).

00:49:25 <dc_rdfig> Titled item A.

00:49:35 <danbri> A:"Ben Hammersley reports on his exclusive glimpse of a new weblog program set to take the web by storm".

00:49:36 <dc_rdfig> Added comment A1.

00:49:50 <danbri> A:"The company behind one of the weblog world's most popular tools is preparing to launch a new service which will attack market leaders Blogger.com head on."

00:49:51 <dc_rdfig> Added comment A2.

00:50:25 <Morbus> for those who use XML::Parser.

00:50:29 <Morbus> - http://www.perlmonks.org/index.pl?node_id=251630

00:50:46 <Morbus> I can confirm that was an issue in my own code, for some users, and can also confirm that the fix worked (300% drop in utilization)

00:52:25 <collord> recommendations for which RDF toolkit ot use for python?

00:53:01 <Morbus> redland, I think.

00:58:33 <collord> Morbus: do yo have any experience with 4RDF?

00:58:43 <Morbus> nope.

01:46:27 <nym> hey

01:46:52 <nym> what other good blogging apps are there for publishing like movable type but OSS?

03:06:06 <eikeon> collord: You might want to try rdflib: http://rdflib.net/

03:58:25 <tav|offline> tav|offline is now known as tav

08:59:40 <q|zzz> q|zzz is now known as qmacro

09:55:29 <dajobe> hmm, where is the owl ontology in OWL (in rdf/xml that is) defined

09:55:52 <dajobe> I can't see it in the OWL Semantics and Abstract Syntax document which "is the final and formally stated normative definition of the language"

09:58:43 <dajobe> hmm, it's in the owl reference which is "a structured informal description "

10:07:52 <sbp`-> sbp`- is now known as sbp`

11:41:46 <wkearney99> wkearney99 is now known as wkearney_away

12:33:37 <aharth> what's the best way to use xml:lang in RDF? use rdf:Alt or not?

12:34:27 <danbri> rdf:Alt is rarely the best way to do anything...

12:35:37 <aharth> ok, then just two tags, ie <dc:title xml:lang="de">Deutsch</dc:title> and <dc:title xml:lang="en">English</dc:title>?

12:37:36 <aharth> in http://www.w3.org/TR/vcard-rdf they use rdf:Alt for multilingual stuff, but that's cumbersome to type...

12:39:17 <danbri> just repeat the property is often ok

12:43:43 <ChanServ> [#rdfig] This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome

12:48:50 <dmwaters> [global notice] Hi all. We've had some problems with one of our us hubs Things should be fine for the time being. Any further information will be given on wallops, /mode yournick +w to enable them. Thank you for using freenode!

13:01:58 <dajobe_> dajobe_ is now known as dajobe

13:04:26 <DanC> .time

13:04:27 <datum> Thu, 24 Apr 2003 13:04:26 GMT

13:04:52 <danbri> DanC, is mit offline?

13:05:00 <DanC> seems to be

13:08:36 <DanC> oddly, I can get to http://esw.w3.org/topic/PoorMansHypertext

13:08:51 <DanC> ah; tux is back

13:13:22 <ericP> we've been asked for an estimate of the number of RDF query BOF attendees. any idea how to estimate?

13:15:42 <dajobe> run a query?

13:17:06 * danbri just replied, guestimated 7-15

13:17:14 <danbri> depends on what the other BOFs are, I epxect

13:27:51 <MarkB>http://www.brainoff.com/geoblog/

13:27:52 <dc_rdfig> B: http://www.brainoff.com/geoblog/ from MarkB

13:28:04 <MarkB> B:|Geoblog

13:28:04 <dc_rdfig> Titled item B.

13:28:39 <MarkB> B:you want geo info? give 'em eye candy baby, yah

13:28:39 <dc_rdfig> Added comment B1.

13:30:34 <DanC> B:what plugin or whatever am I missing? I can't see anything

13:30:34 <dc_rdfig> Added comment B2.

13:30:44 <danbri> flash

13:30:56 <MarkB> yup

13:31:23 <danbri> I see a peeled globe with day/night overlayed, and blog locations drawn on, with alternating recent headlines showing. quite nice really.

13:32:08 <MarkB> almost makes one wanna geo enable their blog, just to be "on the map" so to speak. I did, though in geotag format.

13:33:18 <MarkB> is there a convention for RDF-in-HTML-META-tags? Ntriples maybe?

13:33:32 <MarkB> nevermind this compound document crappola 8-)

13:33:41 <dajobe> well I was going to answer

13:33:57 <dajobe> but your last line put me off

13:34:02 <MarkB> 8-)

13:34:42 <MarkB> it's gotta be simpler than that. if an author can just say name="geo.location" content="x;y", versus figuring out RDF-in-HTML, it's a no-brainer

13:35:23 <dajobe> a no-brainer like rss2?

13:35:41 <DanC> RDF-in-HTML: cf http://esw.w3.org/topic/EmbeddingHTMLinRDF

13:35:44 <MarkB> i'm afraid so 8-(

13:36:07 <DanC> ntriples in HTML.... that's sick, twisted, and... gee... one of the more straightforward options available.

13:36:31 <dajobe> lots of < & > escaping issues and more I just can't bear to think about

13:36:33 <DanC> oops...

13:37:04 <MarkB> good point dajobe

13:37:19 * dajobe returns to writing "A retrospective on the development of the RDF/XML Revised Syntax"

13:37:23 <DanC> rather http://esw.w3.org/topic/EmbeddingRDFinHTML

13:37:49 <DanC> .google geourl

13:37:50 <datum> geourl: http://geourl.org/

13:38:09 <DanC> ^quite reasonable HTML meta conventions for geo-blogging and such there

13:38:28 <DanC> and the server answers queries with RDF/XML

13:40:24 <DanC> B:see also [geourl|http://geourl.org/]

13:40:25 <dc_rdfig> Added comment B3.

13:40:43 <DanC> B:discussion raised the [http://esw.w3.org/topic/EmbeddingRDFinHTML|EmbeddingRDFinHTML] question

13:40:43 <dc_rdfig> Added comment B4.

13:45:04 <nmg> what should the profile attr of the head element read for the example that geourl give?

13:46:15 <nmg> eg. they have metadata elements from DC as well as their own (ICBM) vocabulary

13:55:21 <DanC> good question, nmg

14:18:22 <Wack> inter-continental ballistic missile vocab? :]

14:18:41 <libby> .time UTC

14:18:41 <datum> 2003-04-24T14:18:40

14:19:02 <Wack> ah, geourl, nm :]

14:19:43 <libby> BLURB: rdf query tests meeting today, 1430 UTC

14:19:44 <dc_rdfig> C: rdf query tests meeting today, 1430 UTC from libby

14:19:52 <libby> C:[http://lists.w3.org/Archives/Public/www-rdf-rules/2003Apr/0033.html|agenda]

14:19:52 <dc_rdfig> Added comment C1.

14:20:24 <libby> would someone mind changing the channel title for me in 5 minutes?

14:21:09 <libby> BLURB:date of next RDF query tests meeting (agenda item D)

14:21:09 <dc_rdfig> D: date of next RDF query tests meeting (agenda item D) from libby

14:21:56 <libby> BLURB: previous RDF query testcases meetings actions and plans (agenda item E)

14:21:57 <dc_rdfig> E: previous RDF query testcases meetings actions and plans (agenda item E) from libby

14:22:02 <libby> hi alberto

14:22:16 <libby>http://esw.w3.org/topic/TMQLBOF

14:22:16 <dc_rdfig> F: http://esw.w3.org/topic/TMQLBOF from libby

14:22:20 <alberto> hello all!

14:22:45 <libby> F:|Topic maps QL and RDF QL BOF, XMLEurope 2003 (agenda item F)

14:22:45 <dc_rdfig> Titled item F.

14:23:05 <libby> BLURB: #undef resource in Andy's ResultSet vocabulary (agenda item G)

14:23:05 <dc_rdfig> G: #undef resource in Andy's ResultSet vocabulary (agenda item G) from libby

14:23:26 <libby> BLURB: optionals, provenance and transformation into RDF queries (agenda item H)

14:23:27 <dc_rdfig> H: optionals, provenance and transformation into RDF queries (agenda item H) from libby

14:23:39 <libby> how you doing alberto?

14:24:12 <alberto> fine, just having a bit of ADSL connection problems today :(

14:25:06 <alberto> libby, saw your other email about geo FOAF groups - interesting

14:25:49 <AndyS> Hi there, Alberto!

14:25:59 <alberto> hello Andy!

14:26:18 <alberto> did not replied yet to your email , sorry, too much busy with other deadlines these days....

14:26:25 <libby> thanks alberto

14:26:29 <libby> hey andys

14:28:11 <libby> re groups alberto, if you don;t think you'll have time, tell me your nearest airport and I'll do it....

14:28:18 <libby> or start it anyway

14:28:18 <alberto> Libby we might experimet GEO RDF display with some WMS server we are developing to display maps of people around the world [see here non-RDF-yet experiments|http://www.asemantics.net/showcase/wms.html]

14:28:34 <libby> WMS server?

14:28:46 <alberto> Web Mapping Server ie. OGC stuff

14:28:56 <alberto> dirk has been working on that

14:29:23 <AndyS> Re: email - that's OK - I hope it isn't a short conversation

14:29:45 <libby> neat alberto!

14:29:49 <libby> re ttp://www.asemantics.net/showcase/wms.html

14:30:28 <alberto>http://www.asemantics.net/showcase/zoom.html

14:30:28 <dc_rdfig> I: http://www.asemantics.net/showcase/zoom.html from alberto

14:30:34 <libby> .time UTC

14:30:34 <datum> 2003-04-24T14:30:34

14:30:45 <libby> ------rdf query testcases meeting start-----

14:30:58 <libby> hi all who are here for the meet

14:31:08 <alberto> I: WMS server demo display maps of Apache/FreeBSD developers/commiters and others

14:31:09 <dc_rdfig> Added comment I1.

14:31:11 <libby> everyone else, we are scheduled to meet for an hour

14:31:37 <DanC> DanC has changed the topic to: RDF Query test meeting 1430Z http://rdfig.xmlhack.com/

14:31:40 <alberto> I: need to be RDF-ied one of these days - the back end is pure perl/C

14:31:41 <dc_rdfig> Added comment I2.

14:31:46 <libby> could I encourage attendee to add themselves to C?

14:32:06 <libby> C:attending [Libby Miller|http://ilrt.org/people/libby/]

14:32:06 <dc_rdfig> Added comment C2.

14:32:09 * DanC isn't attending, unfortunately

14:32:43 <libby> alberto, andys, add yourselves to C? anyone else?

14:32:53 <eikeon> C:attending [eikeon|http://eikeon.com/]

14:32:53 <dc_rdfig> Added comment C3.

14:32:57 <libby> I think danbri is unwell

14:33:02 <libby> ooh, hey eikeon

14:33:06 <AndyS> C: [Andy Seaborne|http://www.hpl.hp.com/semweb/people.htm#Andy%20Seaborne]

14:33:06 <dc_rdfig> Added comment C4.

14:34:19 <AndyS> Looks like the virtual Alberto is unwell as well :-(

14:34:30 <libby> ok, first agenda item - date of next RDF query tests meeting (agenda item D)

14:34:34 <libby> yeah :(

14:35:06 <alberto> sorry my IRC client keeps crashing today :(

14:35:19 <libby> we had this problem yesterday for the calendar meet - XMLEurope is on next scheduled meeting time, 8th May

14:35:34 <AndyS> 15 May is OK for me as far as I know

14:35:37 <libby> - I'll send me regrets if we decide to go ahead with the meeting

14:35:40 <alberto> C:Alberto Reggiori

14:35:40 <dc_rdfig> Added comment C5.

14:35:54 <alberto> fine by me too libby for the 15th

14:35:54 <libby> 15th ok for me too - alberto?

14:35:57 <libby> cool

14:36:02 <libby> eikeon?

14:36:17 <eikeon> 15th works

14:36:19 <libby> any objections to 15th May 2003. 1430 UTC next meet?

14:37:30 <libby> D:RESOLVED: to meet [http://www.timeanddate.com/worldclock/fixedtime.html?day=15&month=5&year=2003&hour=14&min=30&sec=0&p1=0|2003-05-15, 1430UTC]

14:37:30 <dc_rdfig> Added comment D1.

14:38:15 <libby> at the same time I was wondering at what stage we can declare victory - we agreed to meet until www2003, so 15th will be the last meet before then

14:38:59 <alberto> yes, the last chance to try something ;-)

14:39:13 <AndyS> Given the last agenda item :-) victory may be grander and later - seriously - a checkpoint of manifest/results-set/(tests?) first is good idea

14:39:28 <libby> I would say when we have perhaps 3 or 4 toolkits using the same tests, that would be something to aim for anyway

14:39:42 <libby> expand a bit andys?

14:40:03 <libby> re www2003, would be nice to have something to show then yep :)

14:40:09 <AndyS> The complete tests themselves might be later (c.f. constraints) I think a short term goal by WWW2003 is good

14:40:35 <libby> any ideas what it should be andys?

14:40:40 <alberto> I agree libby, just simple testing of existing languages

14:40:45 <AndyS> Also - Alberto's msg goes beyond just tests.

14:40:59 <libby> indeed

14:41:05 <alberto> then if we would get graph-equality and constraints too it would be cooler of course

14:41:19 <libby> we could continue the meetings, but they may have a different focus

14:41:27 <alberto> Andys, yes, I would really like to talk more about that last agenda item :-)

14:41:31 <libby> heh

14:41:39 <alberto> in www2003 too

14:42:14 <libby> so today I've been trying to use the jena tests with inkling, and it's taken longer than I thought (e.g. converting queries to squish)

14:42:21 <AndyS> So - www2003: status and report backs (demos - but may not be that "visual" :-)

14:42:23 <libby> but I think doable shortly

14:42:39 <AndyS> Anything particular? I had a converter script once but I've lost it

14:42:48 <libby> yep, not very visual, unless you count text scrolling past :)

14:43:10 <libby> hm, I'm experimenting with perl, and I'm not too good with perl, but getting there now.

14:43:25 <libby> I might also have a go with conversion to N3, or simpler tests anyway

14:43:25 <alberto> I will work to get my tests in the manifest format + running the perl 'make test' on that, perhaps not going to check too much ;-)

14:43:45 <alberto> agree with andy, less visual perhaps

14:44:20 <libby> I think the real test is interoperability of testcases, but interim goal, maybe getting say 3 of us using the format. 4 maybe

14:44:35 <alberto> it would be cool to get all the tests we have in some "native" syntax and write a "meta-generator" to convert/map between input/outoput i.e. RDF/XMl, N3, NTriples

14:44:39 <libby> formats rather - manifest and resultset

14:44:50 <libby> +1 alberto....

14:45:30 <libby> I'm thinking maybe try ntriples as the 'native' syntax

14:45:32 * ericP watches with 1/2 eye

14:45:44 <libby> I can convert betweebn ntriples and lots of query syntaxes

14:45:48 <libby> hey ericp

14:45:51 <alberto> the problem is of those Document-RDQL and Document-Squish at the moment - we still need a parser for each syntax

14:45:54 <alberto> hi ericP

14:46:34 <libby> I'm half way there converying squish and RDQL to ntriples so I'm going to action myself to finish that off and circulate

14:46:59 <alberto> I got a native NTriples parser at least :)

14:47:19 <alberto> and it does not croak on bArcs either

14:47:22 <libby> C:ACTION libby convert squish and RDQL jena tests to ntriples and circulate by date of next meet

14:47:22 <dc_rdfig> Added comment C6.

14:47:35 <libby> neat alberto, me too, though I've not fully tested it

14:48:09 <libby> the issue I founf with Ntriples as query is specifying which vraiables you actually want back - that's needed really

14:48:10 <AndyS> Libby - you don't need to do all my tests - many are not about the graph match part.

14:48:21 <alberto> might try to get thoese tests and use them for rdfstore testing i.e. mapping to native dialiect/syntax query

14:48:36 <libby> thanks andys, yep I'll separateout the very simple ones

14:49:02 <libby> loads of test andys :))

14:49:28 <AndyS> Grown over time - there are some duplicates me thinks

14:49:37 <alberto> libby, I would convert "blindly" the triple-patterns from SquishQL to n-triples and then use the manifest.rdf to get from Andy's result set the required bound vars

14:49:40 <libby> shall we take alberto's suggestion as a goal for WWW2003? "get all the tests we have in some "native" syntax and write a "meta-generator" to convert/map between input/outoput i.e. RDF/XMl, N3, NTriples"

14:50:11 <libby> right alberto, that's an idea. better than hacking with the ntriples syntax anyway

14:50:48 <alberto> yes, ntriples is nice and simple :)

14:51:18 <AndyS> +1

14:51:31 <AndyS> +! to suggestion of native syntax , I mean)

14:51:31 <libby> PROPOSED: (p.ntriples) get all the tests we have in some "native" syntax and write a "meta-generator" to convert/map between input/outoput i.e. RDF/XMl, N3, NTriples

14:51:31 <dc_rdfig> Label PROPOSED not found.

14:51:45 <libby> any seconders?

14:51:57 <alberto> I can write a quick.pl script to convert RDQL triple-patterns to n-triples using s/?/_:/

14:51:58 * AndyS thinks the bot needs an upgrade!

14:52:04 <AndyS> Seconded

14:52:12 <alberto> ok

14:52:15 <libby> that'd be excellent andys

14:52:52 <libby> so, just clarifung p.ntriples: is that ntriples as the query, or for all input, output etc docs?

14:53:09 * libby was thinking the query

14:53:28 <libby> since the rest is doable using existing tools anyway...

14:53:37 <AndyS> IMHO go for the query case

14:53:46 <alberto> all of them would be better, but RDF/XML as input and output are fine by me

14:54:03 <libby> ok, let me ammend that proposal

14:54:26 <alberto> perhaps more attractive to have input and output in RDF/XML and query in ntriples to show the "flexibility" of the manifest and approach

14:55:35 <libby> PROPOSED: (p.ntriples.1 ) get all our simple conjunctive queries from tests in ntriples format, and write a meta-generator to convert between different syntaxes for query

14:55:42 <dc_rdfig> Label PROPOSED not found.

14:55:43 <libby> that sound ok? any seconders?

14:55:56 <AndyS> That sound good - seconded

14:55:59 <libby> thanks

14:56:06 <libby> any objections?

14:56:11 <alberto> very goody libby :)

14:56:24 <libby> I'm going to set the date as 15th may for review.

14:56:25 <libby> :)

14:56:49 <libby> caz: RESOLVED: (p.ntriples.1 ) get all our simple conjunctive queries from tests in ntriples format, and write a meta-generator to convert between different syntaxes for query, by 15th May meeting

14:56:52 <libby> darn

14:56:57 <libby> C:RESOLVED: (p.ntriples.1 ) get all our simple conjunctive queries from tests in ntriples format, and write a meta-generator to convert between different syntaxes for query, by 15th May meeting

14:56:57 <dc_rdfig> Added comment C7.

14:57:18 <libby> ok, we shoudl move on, although I think we've covered some of the other items anyway

14:57:41 <libby> agenda item E: previous RDF query testcases meetings actions and plans

14:58:17 <libby> ACTION danbri take a look at cant.py for comparing graphs

14:58:17 <libby> ACTION danbri mix in his namespace for syntax of query with alberto's

14:58:17 <libby> format

14:58:17 <libby> ...in rubyrdf test cases, and circulate pointer

14:58:28 <libby> danbri's not here, so skip

14:58:42 <libby> danbri and libby plan to implement manifest format

14:58:47 <libby> - I';ve done that

14:59:01 <libby> alberto will start my "my" new one (manifest)

14:59:17 <libby> - alberto, I've forgetton, did you do that for your tests?

14:59:27 <libby> (it wasn;t an action anyway)

14:59:52 <libby> we also have

14:59:54 <libby> ACTION libby implement the resultset format for next meet.

14:59:54 <libby> alberto, danbri and jeen plan to implement resultset vocab soon

15:00:04 <alberto> nope :-( sorry

15:00:14 <AndyS> I use a simpler form of manifest but its "in the spirit of" the new one.

15:00:17 <libby> jeen and co did their bit; I have too, though not checked with grpah match yet

15:00:54 <libby> alberto did a version of the jena tests in that manifest though right?

15:00:57 * eikeon interested in cant.py... will keep an eye out for any discussion on it.

15:01:02 <alberto> I will get the manifest with the new action about ntriples conversion for the 15ht

15:01:21 <libby> yep eikeon - intersting, but timbl and danbri not here today ...

15:01:27 <libby> cool alberto

15:01:45 <libby> ok, well are we ok leaving that agenda item?

15:01:52 <alberto> sorry I do get, though right?

15:02:12 <libby> ? alberto?

15:02:17 <libby> what do you mean?

15:02:38 <alberto> referred to your Jena related question

15:03:15 <alberto> myself and Andy have to work on getting those tests "running"

15:03:21 <libby> right, sure

15:03:29 <libby> I'm having a go too - very useful

15:03:59 <libby> ok, let's move to Topic maps QL and RDF QL BOF, XMLEurope 2003 (agenda item F)

15:04:33 <libby> just to say that there is one, organised by Lars, and I'll be there

15:04:35 <alberto> yes, saw that chumped some days ago...interesting

15:04:46 <libby> will anyone else be at XMLEurope?

15:05:16 * Jeen is back

15:05:16 * DanC has been noodling on http://esw.w3.org/topic/TopicMaps

15:05:26 <libby> hey jeen

15:05:34 <Jeen> sorry I'm late everyone

15:05:39 <libby> jeen, you at XMLEurope?

15:05:50 <alberto> nope :-(

15:05:51 <DanC> xml:lang in RDF should be killed in favor of some use of InterpretationProperties to model topicmap scopes, I think.

15:05:51 <libby> no prob, jeeb, could you add your name to C:?

15:05:52 <alberto> waves to Jeen

15:05:58 <Jeen> nope...

15:06:08 <libby> ah well

15:06:12 <Jeen> I'm a bit overbooked :(

15:06:40 <libby> F:ACTION libby report back to www-rdf-rules about this meeting

15:06:40 <dc_rdfig> Added comment F1.

15:06:59 <DanC> does xml:lang come up in RDF query discussions? I saw one use case in the ... er.. repository... it's the *only* one with *no* implementations, last I looked.

15:07:10 <Jeen> heh

15:07:17 <libby> jeen, see http://rdfig.xmlhack.com for resolutions and actions so far this meet

15:07:18 * DanC is considering asking RDF Core to reopen its decision on xml:lang

15:07:19 <Jeen> we did an implementation of it in SeRQL :)

15:07:26 <Jeen> cheers libby

15:08:03 <DanC> Jeen, you implemented xml:lang query support? details, please? pointers?

15:08:24 * libby has never discussed xml:lang with anyone, ever ;)

15:08:50 <Jeen> DanC, it's in this new QL proposal of ours called SeRQL. It's part of Sesame. http://sesame.aidministrator.nl/ runs a demo server

15:08:59 <alberto> DanC, I do not generally agree with zapping xml:lang, but yes having it into properties would be "queryable" :)

15:09:14 <Epesh> gah, wkearney_away is, like, away :)

15:09:46 <libby> ok, 2 more agenda items, 20 minutes

15:10:04 * eikeon does some queries that look at xml:lang -- may have some thoughts on issue.

15:10:05 <AndyS> Does SeRQL allow a query to split up the literal? i.e. get all @fr text?

15:10:24 <libby> maybe we could add xml:lang to a future agenda? or would people like to talk about it now

15:10:24 <libby> ?

15:10:26 <alberto> good one Anndy...

15:10:44 <alberto> libby, let's procede....

15:10:44 <AndyS> (Sorry Libby - went before I saw your process point)

15:10:50 * DanC would be happy to see it as a future agenda item, didn't mean to disrupt the meeting

15:11:00 <Jeen> AndyS , it has functions that allow you to get the different facets of a literal.

15:11:42 <AndyS> OK - so its external to the RDF retreival as we discussed a while ago

15:11:43 <Jeen> Have a look at http://sesame.aidministrator.nl/sesame/serql/serql-examples-museum.txt , example query no. 5.

15:11:55 <AndyS> Ta - back to the agenda . . .

15:12:11 <libby> cool

15:12:13 <libby> interesting q

15:12:16 <libby> #undef resource in Andy's ResultSet vocabulary (agenda item G)

15:12:40 <libby> I put this in because there's been discussion; I don;lt know if anyone would liek to discuss it further...

15:12:54 <Jeen> hi arjohn

15:12:57 <libby> hey arjohn

15:13:06 <arjohn> Finally made it:-)

15:13:11 <libby> :)

15:13:13 <AndyS> It isn't mandatory - good software is goiong to have to cope with the absense of undefs and the presence.

15:13:13 <alberto> hello

15:13:34 <AndyS> Arjohn - perfect timing - see #undef agenda item!

15:13:47 <ericP> has anybody here gotten sesame working for them?

15:13:56 <Jeen> ericP: I did :)

15:14:20 <libby> G:[arjohn's message to www-rdf-rules|http://lists.w3.org/Archives/Public/www-rdf-rules/2003Apr/0021.html]

15:14:20 <dc_rdfig> Added comment G1.

15:14:34 <libby> perhaps everyone is happy with the result of tgaht discussion on the list?

15:14:36 <Jeen> ericP: if you have problems with getting it up and running feel free to contact me about it.

15:14:45 <ericP> excellent - tx!

15:15:19 <AndyS> I have had it running to write the Jena-Sesame query engine.

15:15:21 <ericP> i'll start from scratch after this meeting

15:15:21 <arjohn> Andy's answer to my e-mail answered my questions on the #undef thing

15:15:29 <libby> ok, thanks arjohn

15:15:45 * AndyS puts the clock on EricP ....

15:15:53 <alberto> I think the 'undef' thingie has been clarified

15:15:59 <libby> G:resolved to arjohn's satisfaction

15:15:59 <dc_rdfig> Added comment G2.

15:16:02 <arjohn> Just hope we (he) made the right decission and that we don't have to change our minds in the future :-)

15:16:18 <libby> heh, I'm sure we'll find out ;)

15:16:23 <AndyS> Going to have to cope with both on reading to be robust software

15:16:32 * ericP is feeling the pressure now.

15:16:34 <alberto> arjohn, coding will tell you that :-)

15:16:54 <libby> ok, well let's move on to 'free-form discussion' then....

15:16:55 <libby> optionals, provenance and transformation into RDF queries (agenda item H)

15:17:02 <arjohn> yep, you never know what comes up when coding stuff...

15:17:36 <alberto> that's a good one libby ;)

15:17:51 <libby> heh

15:17:58 <alberto> well, I just dumped some ideas I have been struggling over in the past months

15:18:08 <alberto> 1. optionals

15:18:13 <libby> H:[alberto's message (see thread)|http://lists.w3.org/Archives/Public/www-rdf-rules/2003Apr/0030.html]

15:18:13 <dc_rdfig> Added comment H1.

15:18:33 <AndyS> Given these new directions, it is really a new QL? Would be good to stabilze current QLs for users while new one(s) done.

15:19:19 <arjohn> We had some ideas on adding optional paths to SeRQL, but the planned notation (using a '?') is said to be problematic for JDBC/ODBC compatibility.

15:19:24 <alberto> yes Andy, it will be a "new" query language, I agree on that

15:19:33 <libby> optionals is dead useful

15:19:46 * larsbot has found that, too

15:19:58 <Jeen> I'll second that

15:20:02 <alberto> arjhon, same here for ODBC/JDBC '?' problems

15:20:14 <AndyS> The ?/JDBC came up elsewhere. Soln could be an *alternative* (not replacement) variable indicator char

15:20:17 <alberto> Andy knows too ;)

15:21:02 <alberto> the general problem we have is the '?' sign clashes with ODBC/JDBC drivers supporting standard SQL

15:21:04 <Jeen> I'm curious: why would using ? be a problem?

15:21:34 <AndyS> Its the client-side substitution character

15:21:40 <alberto> in DBI and JDBC (I think) '?' means something else - anyway, we can chat about that another moment

15:22:09 <AndyS> i.e. query is a template with ? in - then is "set(2,something)" to replace it.

15:22:27 <alberto> do you agree that 'optionals' are important for an RDF query then? yes/no?

15:22:32 <Jeen> yes

15:22:36 <arjohn> yes!!!

15:22:37 <libby> yes

15:22:41 <alberto> even if not in the current implementations

15:22:46 <AndyS> Yes

15:23:04 <alberto> good! :-) that is a new requirement then

15:23:05 <AndyS> Especially in RDF->RDF

15:23:13 <libby> I think it's very useful; I aso think that QLs can be useful without it

15:23:24 <libby> but, if starting from scratch, yep

15:23:38 <AndyS> At this point, it is (transformation) rules.

15:23:38 <libby> are you building a set iof requirements alberto?

15:23:54 <libby> does anyone else have requirements list, e.g. serql maybe?

15:23:58 <libby> rdql?

15:24:13 <libby> be good to link such a think in with usecases

15:24:19 <libby> s/think/thing/

15:24:39 <AndyS> I have an informal set of ideas - I split between RDFand beyond-RDF features (e.g. provenance and quads)

15:24:45 <Jeen> you mean requirement for additional features? Yeah, we have a list of those. It's not very prioritized though.

15:24:51 <libby> we could perhaps use the wiki to create such a thing, at least as a temporary repository of ideas

15:24:57 <AndyS> Shoudl we use the Wiki for collecting them?

15:25:10 <libby> yeah, that'd be good andys

15:25:26 <libby> want to start one andys?

15:25:48 <arjohn> Example use of optional paths in SeRQL (not yet implemented): select * from {Person} <foo:name> {Name}, {Person} <foo:email> {Email}?

15:25:52 <arjohn> Email is optional here.

15:25:57 <alberto> the basic problem we have is that we got very sparse RDF/XML data now days describing satellites and most fields a very specific and not defined (or defineable) - then we need to do case sensitive matching using the API code i.e. optionals

15:26:01 <alberto> libby, I think the "requirements" is something me and Andy have been talking/thinking in the past days about about a "new" language

15:26:09 <AndyS> Can do but its Alberto's input.

15:26:31 <libby> well, up to you. be an intersting resource though on the wiki if you decide to do so

15:26:35 <alberto> +1 Wiki Page

15:26:38 <libby> between you

15:26:43 <libby> cool!

15:26:46 <AndyS> The optional I want is (?x dc:* ?y) - i.e. all properties from Dublic Code namespace

15:26:53 <alberto> let's call that the "new" language PipeR = Pipe RDF :-)

15:27:00 <libby> interesting andys

15:27:07 <Jeen> ah, that's a different type of thing. also interesting.

15:27:20 <ericP> re optionals, i've found it very useful in algae.

15:27:24 <libby> wiki names are tricky (for me anyway)

15:27:29 * ericP is a bit behind and apologizes

15:27:44 <AndyS> jeen - agreed - its is both unnamed vars, optionality and filtering on URIs.

15:27:49 <ericP> also want to point out that DQL optionals are different from optional triples.

15:27:52 <alberto> Andys, those are called regular-path-expressions in semi-structured data - I very much like those :-)

15:28:01 <alberto> s/Andys/Andy/

15:28:04 <Jeen> we were thinking of adding a namespace function for that kind of thing: {X} prop {Y} where namespace(prop) = dc

15:28:24 <libby> ericp, expand a bit?

15:28:35 * AndyS must go at end of the hour

15:28:48 <ericP> DQL has (don't|may|must)bind variables

15:29:00 <libby> ok, will you and alberto talk abotu a possible wiki page andyS?

15:29:08 <alberto> ok - next items : transformation (construct) and provenance/context

15:29:59 <libby> transformation?

15:30:02 <arjohn> the contruct-example in the e-mail is RDF --> XML, I would prefer RDF --> RDF

15:30:03 <ericP> person X worksAt FredUniversity.

15:30:13 <ericP> person X hasFather ?f.

15:30:17 <arjohn> (that's the way it works in SeRQL)

15:30:19 <ericP> person ?f worksAt FredUniversity.

15:30:29 <ericP> but don't bind ?f

15:30:35 <alberto> transformation = RDF --> RDF

15:30:45 <libby> riht, useful alberto

15:30:53 <alberto> see my email and andy's reply explaining that

15:30:59 <ericP> the triple is required, but the name of the obecjtin in the triple is not sent back tot he queryer

15:31:02 <ericP> .done

15:31:10 <libby> thanks ericp

15:31:24 <libby> hm, we're out of time, unformtunately

15:31:29 <alberto> the idea in there is to reduce the DBI/JDBC "coding" part i.e. express into the query what you what back

15:31:38 <arjohn> Alberto, agreed on that, but the example allows you to create arbitrary XML

15:31:59 <AndyS> So it is projection? Or does it stop multiple solutions as well?

15:32:02 <libby> ----adjourned, but don't let that stop you :) -------

15:32:15 <arjohn> i've just started;-)

15:32:19 <libby> heh

15:32:22 <alberto> arjhon, I think XQuery do that already - we should be "similar" using XML technologies as far as possible

15:33:30 <arjohn> I think creating XML output is easier if based on select-type queries than on construct-type queries

15:34:06 <arjohn> select-type queries have more regular results, construct-type queries can result in "free-form" RDF graphs

15:34:53 <arjohn> Example tech. solution: have a fixed XML-output for select-type queries and apply XSLT stylesheet to remodel to your specific needs

15:35:57 <alberto> arjohn, then idea would be to have input graphs and output graphs, like RDBMS do table-to-table

15:35:59 <alberto> (not the right english wording perhaps ;)

15:36:03 <alberto> yes kind of that

15:36:15 <alberto> libby? you still there?

15:37:00 <libby> sorta...

15:37:02 <libby> :)

15:37:27 <alberto> anything else ? I need to go....

15:37:48 <alberto> let's move some of the bits of the last item to next meeting if possible

15:37:50 <arjohn> CU next time

15:37:51 <libby> thats cool alberto

15:37:56 <libby> thanks for coming

15:38:01 <libby> :)

15:38:30 <alberto> I will talk with Andy about setting up a WiKi page for those "new" ideas about querying

15:38:48 <libby> cool, ill move items as you suggest

15:38:57 <Jeen> this would be dead-useful. I have the feeling there's a strong overlap with our own ideas and requirements.

15:39:02 <alberto> cu and thanks to all

15:39:08 <Jeen> bye alberto

15:39:23 <libby> bye!

15:41:00 <Jeen> libby, this bof meeting at XML Europe, is that instead of the one at WWW? Or just an extra meeting?

15:41:20 <libby> no, diffrent meet - as well as

15:41:35 <Jeen> ok good :)

15:42:09 <libby> yeah - different focus

15:43:05 <Jeen> ah, I see. overlap with topic maps QL

15:43:16 <Jeen> (I really should read my e-mail a bit more carefully)

15:43:20 <libby> heh

15:43:20 <arjohn> i'll leave now. Hope to be on time next meeting.

15:43:52 <Jeen> when is the next meeting btw?

15:44:03 <libby> ok, thanks arjohn , jeen, 15th may, same time

15:44:12 <Jeen> ok great

15:44:26 <Jeen> I have to be going as well actually.

15:44:38 <libby> ok, bye, thanks!

15:44:39 <Jeen> thanks, seeya later.

15:52:50 <DanC> should I change the topic? is the meeting over?

15:53:21 * DanC sees "adjourned" above

15:53:27 <libby> yes please

15:53:41 <DanC> DanC has changed the topic to: Semantic Web hack 'n' chat http://rdfig.xmlhack.com/

15:55:18 <libby> thanks

15:59:56 <Epesh> sorry I interrupted the meeting, my apologies :)

16:22:49 <libby> no worries epesh

16:31:38 * maxf waves

16:31:49 <maxf> danbri, you around?

17:03:30 <collord> rdflib really is that much easier to deal with, isn't it

17:31:22 <eikeon> Anyone know what packages need to be installed on debian to get a python2.2 working with expat supprt?

17:32:07 <eikeon> ... expat support seems to sometimes come in a different package... python-dev?

17:33:08 <dajobe> maybe python2.2-xmlbase

17:33:28 <dajobe> it has things like /usr/lib/python2.2/xml/sax/expatreader.py

17:34:23 <eikeon> Ah... that looks like it.

17:34:38 <dajobe> you can search for this on packages.debian.org

17:35:21 * eikeon not a debian user... yet.

17:41:26 <libby> there's some stuff here from when I installed cwm: http://esw.w3.org/topic/RdfCalendarSchema

17:42:14 <eikeon> libby, dajobe: Thank you.

17:57:33 <nwalsh> Hello all.

17:57:37 <dajobe> hi

17:57:57 <nwalsh> So, I've got some RDF lying around that I invented.

17:58:03 <dajobe> ah

17:58:04 <nwalsh> Notionally, it conforms to some sort of schema.

17:58:17 <nwalsh> So, I could probably figure out how to write an RDF Schema for it. I think.

17:58:27 <nwalsh> And then I'd have a schema and some instances lying around.

17:58:38 <dajobe> yeah

17:58:39 <nwalsh> And...wait for it...I want to validate that the instances conform to the schema.

17:58:48 <nwalsh> The tool I'd use is...uhm, what?

17:58:51 <dajobe> that's ok, there are things for that

17:59:04 <nwalsh> Pointers, please?

17:59:05 <dajobe> sigh. Usual question #1 what language do you care about?

17:59:27 <nwalsh> Uhm. You mean validator implementation language?

17:59:29 <dajobe> yes

17:59:41 <nwalsh> Perl, Python, Java, C++, C, you name it, I'll hack it. :-)

17:59:55 <dajobe> great, you're unusual in language-neutrality then :)

18:00:01 * nwalsh decides it's probably pointless to mention Modula-3 and silly to mention various assemblers

18:00:04 <dajobe> ok, try sesame in java

18:00:16 <Dorward> Real men do all their coding in postscript! :)

18:00:35 <dajobe> it's got a constraining (assume close world) validator, web interface, pretty easy to set up

18:00:46 * DanC perks up at the mention of Modula-3... waves to nwalsh

18:00:53 <nwalsh> Postscript. I hadn't thought of Postscript.

18:01:00 * nwalsh waves back

18:01:14 <DanC> don't forget XSLT, nwalsh.

18:01:15 <nwalsh> Dorward: you do all your RDF model programming in Postscript?

18:01:41 <arnarl-home> hi

18:02:09 <nwalsh> Is there just some command line tool that I can hand a schema and an instance and test the conformance of the instance to the schema?

18:02:14 <Dorward> nwalsh: All the RDF model programming I've done in the past has been in PS. (Of course as there is zero code that doesn't mean very much). I'm still in the `getting to grips with the format' stage.

18:02:28 <dajobe> nwalsh: I think there is a simple cwm one, ask DanC or timBL

18:02:49 <DanC> "How can I check that my RDF document conforms to an RDF schema?" -- http://esw.w3.org/topic/RDFCheckingTools

18:03:10 <nwalsh> Thank you DanC

18:03:14 <dajobe> actually there's another answer now

18:03:17 <dajobe> use an OWL validator

18:03:22 <dajobe> check the [OWL Full] box

18:03:47 <DanC> pointer?

18:04:02 <dajobe> in google, yes

18:04:15 <DanC> .google owl validator

18:04:16 <datum> owl validator: http://owl.bbn.com/validator/

18:04:21 <nwalsh> Yeah, I kind of got to this question by way of OWL, but a quick peek at OWL looked like it was building on top of RDF Schema and I don't need any of the things mentioned in the intro (cardinality, disjointness, etc.)

18:04:43 <DanC> yeah, but OWL tools exist, which is a feature that RDFS tools don't seem to have. ;-)

18:04:46 <danbri> without OWL, there isn't so much one really needs to 'check' at the RDFS level...

18:04:56 <dajobe> there's also VRP - validating RDF parser

18:04:56 <danbri> ...in that you can't contradict yourself

18:04:58 <dajobe> part of rdf suite

18:05:11 <dajobe> at http://139.91.183.30:9090/RDF/

18:05:29 * danbri skeptical of VRP; they implemented their own view of RDFS and disagree with the current form of the specs.

18:05:32 <dajobe> DanC: rdfs tools do exist, plent of them

18:05:44 * nwalsh chuckles. Forth is a language I never did learn.

18:05:59 <dajobe> well, I never learnt perl, it sort of invade me

18:07:10 <DanC> ok, it's not so much that RDFS tools don't exist as that there's not much to check at the RDF level, so not much motivation to build that sort of checker

18:07:30 <DanC>http://esw.w3.org/topic/RDFCheckingTools

18:07:30 <dc_rdfig> J: http://esw.w3.org/topic/RDFCheckingTools from DanC

18:07:46 <DanC> J:|esw:RDFCheckingTools

18:07:46 <dc_rdfig> Titled item J.

18:08:01 <DanC> J:updated 'How can I check that my RDF document conforms to an RDF schema?' with pointers to some tools

18:08:01 <dc_rdfig> Added comment J1.

18:08:01 <dajobe> J:RDF checking tools

18:08:02 <dc_rdfig> Added comment J2.

18:08:54 <DanC> norm, I've got some rules for building a schema from data...

18:09:08 <dajobe>http://sesame.aidministrator.nl/

18:09:08 <dc_rdfig> K: http://sesame.aidministrator.nl/ from dajobe

18:09:16 <DanC> i.e. http://www.w3.org/2000/10/swap/util/sniffSchema.n3

18:09:23 <dajobe> K:|Sesame - for RDF, RDFS, RDF model theory checking

18:09:23 <dc_rdfig> Titled item K.

18:09:29 <nwalsh> I'll remember that. I think I'd like to start "by hand" though so I can see if I can figure it out :-)

18:09:59 <DanC> sniffSchema.n3... wherein we see...

18:10:00 <DanC> # declare the properties I'm using...

18:10:00 <DanC> # thanks to the DAML validator for keeping me honest.

18:10:00 <DanC> # http://www.daml.org/validator/

18:10:27 <dajobe> that's what the bbn OWL validator is based on IIRC

18:10:32 <DanC> yup

18:11:14 <dajobe> hmm 10/swap/doc has gained some big files

18:12:11 <DanC> our www2003 tutorial is in swap/doc; is that what you're seeing?

18:12:20 <nwalsh> Speaking of keeping me honest, I can see I'm actually going to have to put documents at all those URIs I've been using. Baaad norm, no biscuit!

18:12:44 <nwalsh> I wonder if I can teach the validator to do RDDL indirection so I don't have to put actual RDF Schemas there...

18:13:14 <DanC> let us know how it goes; see if you don't discover why RDDL is perhaps not such a good idea...

18:13:30 <nwalsh> heh

18:14:42 * DanC has been trying to get his thoughs around the whole namespaceDocument-8 malarky straightened out in the wiki (in/near http://esw.w3.org/topic/SelfDescribingWeb), but hasn't really gotten over the hump.

18:16:29 * nwalsh goes off to read it

18:20:21 * nwalsh follows the link to AnitPatterns and is reminded again of just *how much stuff* is out there.

18:21:12 <nwalsh> I was hoping you'd got a little further DanC :-). I'm stuck on the first two of your worries. I'm not sure about the third. The third is just RDF-centric RDDL, I think.

18:23:42 * nwalsh goes back to worrying about an RDF schema for (as a first use case) his RDF configuration file format

18:24:25 <nwalsh> ...and immediately trips over a problem. Maybe this is a bad example.

18:24:45 <nwalsh> What's the right way to associate name/value pairs in RDF?

18:25:16 <danbri> turn the names into RDF properties is one way; build a structure that groups pairs of name and value is another.

18:25:23 <nwalsh> I had used the convention <ex:name>value</ex:name>, but of course that makes the set of "ex:..." properties boundless.

18:25:26 <nwalsh> Is there a better way?

18:26:04 <nwalsh> Or is there an RDF Schema way of saying "for namespace xxx, all xxx:... names are properties with literal values"?

18:26:07 <danbri> that seems ok to me... if you can guarantee the names will have sensible XML names...

18:26:44 <nwalsh> Ok, but now the RDF validator says "ex:name" is an implicit property. Maybe I just live with that?

18:27:00 <danbri> not with a 'all'... you'd have to enumerate them

18:27:14 <danbri> I'm not familiar with that msg from the validator...

18:27:47 <nwalsh> I'm running the OWL validator (Vowlidator) from bbn. It says, for example,

18:27:55 <nwalsh> INFORMATION - Implicit Property: Property http://nwalsh.com/rdf/jpegrdf/ns#tech is not explicitly defined at the specified namespace URI.

18:27:56 <nwalsh> In Unknown URI

18:28:37 <nwalsh> The unknown URI bit is just my bad. Ok, so in this case, I'd need a different schema for each configuration file. (And so would you, if you invented your own terms) But I guess that's fair.

18:29:23 <danbri> Ok right, its deref'ing the namespace. that's a handy tool.

18:31:38 * nwalsh gets brain cramp.

18:31:54 <nwalsh> Thanks all, for the pointers. I clearly need to think about this problem a bit more.

18:42:26 <DanC> re boundless names, it's kinda tricky. RDF doesn't disallow them, but you'll get less support. You can say what you mean in N3: { ?P log:uri [ str:startsWith "http://www.nwalsh.com/vocab1/name#"] } => { ?P a norm:Property. norm:Property rdfs:subClassOf rdf:Property}

18:44:20 <nwalsh> Hmmm. Interesting.

18:44:34 <nwalsh> Is there a better way of doing name/value pairs?

18:45:13 <nwalsh> What I have is a whole bunch of numbers with which I want to associate names. Parallel lists or something would work, I guess, but they'd, uhm, suck to maintain.

18:46:14 <DanC> I don't have a clear picture of what you're doing. concrete example, please?

18:46:27 <DanC> you may be running into esw:NaryRelations

18:46:56 <nwalsh> Ok. This is for the EXIF data in your digital camera. There are lots of fields in there, two are "make" and "model", let's say they're identified by the numbers 0x001 and 0x002.

18:47:49 * danbri interested; I made an exif2rdf hack at some point... (rummages)

18:47:58 <nwalsh> The user wants to have that appear in his actual RDF as <exif:Make>Nikon</exif:Make>, <exif:Model>E950</exif:Model>

18:47:59 <DanC> exif:make and exif:model are pretty clearly RDF properties. and they're not unbounded. I'd write { exif:make exifutil:index 1. exif:model exifutil:model 2 }

18:48:16 <nwalsh> But that means the configuration file needs to associate "Make" with 001 and "Model" with 002.

18:48:32 <DanC> little m exif:make, right?

18:48:53 <nwalsh> Yeah, but suppose the person wants use a language other than english for his names?

18:49:00 <danbri> (just conventiion to keep properties lowercase, Classes upper initial?)

18:49:10 <nwalsh> And the set is pretty unbounded since camera manfuacturers can add anything they want in proprietary strings

18:49:23 <DanC> er... can we stick to one use case here? are we talking about the exif vocabulary or not?

18:49:47 <nwalsh> Oh, yes. I suppose that should be "make" not "Make" but I copied the EXIF standard property names exactly so they wound up uppercase

18:49:52 <danbri> norm, re device description, see http://esw.w3.org/topic/UAProfIndex for how the mobile phone folks are doing this in UAProf, cc/pp...

18:50:55 <DanC> ok, exif:Make as a property name is excusable if it's derived from some existing source. but if the source is case-independent, you'll have to normalize, and you might as well normalize to camelCase per RDF convention (cf painful experience going thru this for icalenar)

18:51:28 <DanC> from where did you copy the EXIF standard names?

18:51:32 <nwalsh> Yes, I can normalize. Or you can. The names come from the configuration file, which brings us full circle :-)

18:52:17 <DanC> camera manfuacturers can't add to whatever document you took the exif:Make name from, can they? if they make proprietary extensions, those belong in another namespace.

18:52:35 <DanC> what are we configuring, by the way?

18:52:56 <nwalsh> EXIF: http://www.exif.org/specifications.html

18:52:56 <dc_rdfig> Label EXIF not found.

18:53:20 * DanC swats dc_rdfig

18:53:46 <nwalsh> We're configuring http://nwalsh.com/java/jpegrdf/

18:53:51 <DanC> wikibot, note this discussion in http://esw.w3.org/topic/ImageDescription

18:54:12 <DanC> logger, pointer?

18:54:22 <nwalsh> Ok, fair point. I could pick the EXIF names and fix them in a set schema.

18:54:33 <nwalsh> If Nikon adds their own fields, I could describe those in a schema too.

18:54:49 <nwalsh> And if someone wants to use French words, they just make their own declaration of isomorphism?

18:55:04 <nwalsh> s/in a schema too/in a different schema/

18:55:33 <DanC> logger_2, pointer?

18:55:33 <DanC> See http://ilrt.org/discovery/chatlogs/rdfig/2003-04-24#T18-55-33

18:56:19 <DanC> french words where? the URIs aren't supposed to be the user interface labels.

18:56:35 <DanC> exif:make rdfs:label "..."@fr.

18:57:08 <DanC> i.e. <rdf:Property about="...#make"><rdfs:label xml:lang="fr">...</></>

18:57:25 <nwalsh> Right but the word "make" is always english, I don't get to say exif:[whatever the french word for "make" is]?

18:57:47 <DanC> (which brings up issues around xml:lang and TopicMaps names ...)

18:57:50 <nwalsh> Luckily most of the actual values are numbers :-)

18:58:24 <nwalsh> And the I18N world just lives with exif:compressedBitsPerPixel even if they don't know any English at all?

18:58:51 <DanC> you could declare another property exif:modelo (spanish; I don't grok french) and say it's owl:samePropertyAs exif:model, but that would be shooting a fly with a cannon.

18:59:06 <nwalsh> Ok. Fair enough.

18:59:14 <DanC> if you don't want to be anglo-centric, you could use exif:prop0x001

18:59:26 <DanC> exif:prop0x001 rdfs:label "make".

18:59:54 <nwalsh> That'd be just too hard to use, I think.

19:00:00 <nwalsh> I want to be able to say things like:

19:00:18 <nwalsh> jpegrdf -qr dc:subject xab:dan-connolly *.jpg

19:00:31 <DanC> yes, please ;-)

19:00:35 <nwalsh> Er, but that's the wrong namespace. Uhm,

19:00:59 <nwalsh> jpegref -ql exif:datetime "2003-04-31" *.jpg

19:01:23 <nwalsh> Having to say "... -ql exif:prop0x829a "2003-04-31"..." is just a non-starter

19:01:47 * nwalsh notes immodestly that he can actually do the DanC query on his images and it works :-)

19:01:50 <DanC> if I've got the idea, then I'll be able to do this? jpegrdf -qr foaf:depicts dans-foaf:me *.jpg

19:02:16 <nwalsh> If you've put "foaf:depicts" properties in your images, yes

19:02:24 * DanC can't wait

19:02:24 <nwalsh> So you could say:

19:02:45 <nwalsh> jpegrdf -ar foaf:depicts dands-faof:me dcim100.jpg

19:03:07 <nwalsh> Which...curiously...brings us to exactly the same configuration problem. Let me explain.

19:03:28 <DanC> speaking of immodesty, would you like to present your work to a crowd in a http://esw.w3.org/topic/ScheduledTopicChat on http://esw.w3.org/topic/ImageDescription

19:03:41 <DanC> ok, back to config issues..

19:03:56 <nwalsh> You can say 'jpegrdf -ns foaf http://whatever-foaf-uri-is -ar foaf:..." but it's really tedious to have to specify the namespace prefixes on the commandline every time

19:04:58 * nwalsh notes he'd be happy to show of his new toy to anyone that wanted to listen

19:05:00 * DanC thinks timbl implemented a $HOME/.n3 for default bindings or something

19:05:12 <nwalsh> Ok, so you can specify the default bindings you want in the config file:

19:05:25 <nwalsh> <ns:dc>http://purl.org/...</ns:dc>

19:05:35 <nwalsh> <ns:rdfpiclang>http://sofia/...</ns:rdfpiclang>

19:05:36 <nwalsh> etc.

19:05:46 <nwalsh> But that has precisely the characteristics I described before.

19:05:50 <DanC> ew... why not use normal xmlns bindings?

19:06:01 <nwalsh> Uhm...

19:06:38 <DanC> i.e. jpegrdf --loadPrefixesFrom someTypicalData.rdf ...

19:06:48 <nwalsh> Uhm...

19:07:08 <nwalsh> I'm loading these files using the Jena toolkit and turning them into models, I don't think it lets me ask what the namespace bindings were...

19:07:16 <DanC> so don't do that.

19:07:27 <nwalsh> Uhm...

19:07:28 <DanC> ;)

19:07:43 <nwalsh> Curses, foiled again!

19:07:47 <DanC> namespace prefixes are an under-the-covers sort of thing.

19:08:07 <nwalsh> Right.

19:08:23 <DanC> cwm will let you ask what namespace bindings were used.

19:08:35 <DanC> it doesn't make a lot of guarantees, but it's good for heuristics

19:08:35 <nwalsh> I suppose I could load the config file and say whatever namespaces are declared on the root are available on the command line

19:09:04 <nwalsh> Yeah, I don't really care about the prefixes. And Jena doesn't even attempt to preserve them. E.g.:

19:09:05 <DanC> root->command line smells like an 80% winner to me.

19:09:29 <nwalsh> jpegrdf -s 20030421-113243.jpg

19:09:30 <nwalsh> <?xml version='1.0' encoding='ISO-8859-1'?>

19:09:30 <nwalsh> <rdf:RDF

19:09:30 <nwalsh> xmlns:rdf='http://www.w3.org/1999/02/22-rdf-syntax-ns#'

19:09:30 <nwalsh> xmlns:NS0='http://nwalsh.com/rdf/exif#'

19:09:30 <nwalsh> >

19:09:32 <nwalsh> <rdf:Description rdf:about='20030421-113243.jpg'>

19:09:34 <nwalsh> <NS0:ExposureBiasValue>0</NS0:ExposureBiasValue>

19:09:38 <nwalsh> <NS0:BitsPerPixel>8</NS0:BitsPerPixel>

19:09:40 <nwalsh> <NS0:MaxApertureValue>13/5</NS0:MaxApertureValue>

19:09:42 <nwalsh> <NS0:NikonQuality>12</NS0:NikonQuality>

19:09:45 <nwalsh> etc.

19:10:08 <nwalsh> But by the magic of namespaces, I can still ask about exif:ExposureBiasValue...uhm, if I care :-)

19:10:20 <DanC> right.

19:10:47 <nwalsh> Ok, ok, let me try one more time. (Tell me how to fix this! :-)

19:10:47 * DanC wonders if that could/should be <Nikon:Quality>12</>

19:11:06 <nwalsh> Yes, per our earlier decision about extension names, it should!

19:11:20 <DanC> </> </> </> </> </> </> </>. whee! I can do that all I want here!

19:11:24 <nwalsh> (And will be, uhm, by tomorrow :-)

19:12:01 <nwalsh> Ok, I think this is the last pairwise example in my config file. (And about the last thing I need the config file for, hmpfh! :-)

19:12:24 <nwalsh> The NikonQuality comes from a field defined as type "undefined" in EXIF proper.

19:12:29 <nwalsh> I have a decoder for it, so I say:

19:12:43 <nwalsh> <exifd:MakerNote>com.nwalsh.jpegrdf.Nikon950Decoder</exifd:MakerNote>

19:12:50 <nwalsh> which means use that class to decode the field labelled "MakerNote"

19:13:01 <nwalsh> Got a way to fix *that* :-)

19:13:24 * DanC noodles... <exifd:MakerClass rdf:resource="java:com.nwalsh.jpegrdf.Nikon950Decoder"/>... hmm... maybe not.

19:14:07 <nwalsh> That'd still be pairwise. "MakerClass" == java:...

19:15:06 <nwalsh> I might need another class to decode something else

19:15:12 <DanC> er... hang on... did we switch from NikonQuality to MakerNote somwhere?

19:15:30 * DanC needs food; struggles to think clearly...

19:15:59 <nwalsh> MakerNote contains the raw data for NikonQuaility, NikonFoo, NikonBar, etc.

19:16:14 <nwalsh> So the classfile that decodes MakerNote adds (a bunch) of them

19:16:32 <nwalsh> The association is between the class and the field that it should be used to decode

19:16:45 <danbri> are you using adobe's XMP conventions for embedding all this? or plain JPEG comment fields?

19:16:59 <nwalsh> I'm just using plain JPEG comments

19:17:11 <nwalsh> I haven't figured XMP out yet

19:17:29 <danbri> I made an extractor for it w/ nasty perl regex, see http://www.w3.org/2001/sw/Europe/200206/imagemeta/extract/extract

19:17:33 <DanC> can you go slower for me? does the string "NikonQuality" occur in the .jpg file?

19:17:37 <danbri> (first two demo links currently offline)

19:17:57 * nwalsh starts over, slowly

19:18:14 <nwalsh> The JPEG contains a set of labelled fields.

19:18:21 <nwalsh> One of those fields is labelled "MakerNote"

19:18:41 <nwalsh> The content of that field is proprietary, but the class com.nwalsh.... knows how to decode it

19:18:50 <DanC> in the .jpg file, are those labels integers or strings?

19:18:58 <nwalsh> Those labels are integers

19:19:30 <DanC> but we know from reading exif.org/foo that the integer, say 67, corresponds to "MakerNote", yes?

19:19:36 <nwalsh> That class decodes the field and adds nikon:quality, nikon:foo, etc. to the RDF model associated with the image

19:19:44 <nwalsh> DanC: yes

19:20:11 * DanC thinks he's with you now

19:20:18 <nwalsh> The model is eventually written out in the JPEG comment. At that point "nikon:quality" appears in the comment

19:20:54 <nwalsh> But there could be several fields of type "unknown" in strict EXIF terms that would require special classes to decode

19:21:12 <nwalsh> So there needs to be a pairwise association: when you say X, decode with x.class; when you see Y, decode with y.class, etc.

19:21:37 <DanC> only one such field is called MakerNote, though, right?

19:21:48 <nwalsh> Yes.

19:21:49 <DanC> and each such field is labelled by a distinct integer?

19:22:13 <nwalsh> Yes

19:22:44 <DanC> exif:makerNote exifutil:index 67; jpegrdfconfig:decoderClass "com.nwalsh...".

19:23:17 <DanC> exif:makerNote rdfsLabel "MakerNote"; exifutil:index 67; jpegrdfconfig:decoderClass "com.nwalsh...".

19:23:23 <DanC> or perhaps

19:23:28 <DanC> exif:MakerNote rdfsLabel "MakerNote"; exifutil:index 67; jpegrdfconfig:decoderClass "com.nwalsh...".

19:23:34 <nwalsh> Uhm. show me an example with two decoders, please?

19:23:50 <DanC> two decoders... for different fields?

19:24:15 * DanC meant rdfs:Label where he wrote rdfLabel

19:24:37 <nwalsh> Yes

19:25:01 <DanC> tell me about this other field? (ok to make it up)

19:25:24 <nwalsh> There's also a GPSInfo field, number 55, that I can decode with com.nwalsh...

19:26:56 <DanC> exif:GPSInfo rdfs:Label "GPSInfo"; exifutil:index 55; jpegrdfconfig:decoderClass "com.nwalsh.jpegrdf.GPSInfoDecoder". exif:MakerNote rdfsLabel "MakerNote"; exifutil:index 67; jpegrdfconfig:decoderClass "com.nwalsh.jpegrdf.Nikon950Decoder".

19:27:34 <nwalsh> So you've still got this pairwise mapping. The number of property names (GPSInfo, MakerNote, etc.) in your exif: namespace is unbounded

19:28:02 <DanC> oh, sorry... GPSInfo doesn't go in the exif: namespace. who made it up?

19:28:20 <nwalsh> Oh, I see.

19:28:49 <DanC> the exif guys made up MakerInfo, yes?

19:28:54 <nwalsh> yes

19:29:00 <bitsko> quick q for those who know the spec by heart: is rdf:RDF still optional, if the content can be known to be RDF by an application?

19:29:19 <DanC> not a quick q, sorry, bitsko

19:29:27 <DanC> i.e. what do you mean by "optional"?

19:29:53 <bitsko> this email by Dave Beckett hits it exactly: http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2001Aug/0169.html

19:30:03 <bitsko> I'm wondering if that is still the case

19:30:50 <DanC> nwalsh, does exifutil:index and jpegrdfconfig:decoderClass make sense?

19:31:04 <bitsko> the thread I'm thinking of answering is the last six comments to http://www.intertwingly.net/blog/1353.html#c1051199069

19:31:06 <nwalsh> Now I'm confused in a different way. I had expected exif:MakerNote to be a property of the image. Can properties have properties?

19:31:37 <nwalsh> I think the exifutil:index is superflous since the name and the value have to be associated already. Having the decoder be a property of the field makes sense

19:32:04 <DanC> yes, properties can have properties. (freaks some people out, of course)

19:32:39 <dfg_olin> Could someone offer a quick comparison of Squish vs. Jena? When would I want Squish over Jena?

19:33:02 <danbri> Not really comparing like with like...

19:33:24 <danbri> One is a query language (descendent of rdfdb-ql, ancestor of RDQL), the other is a Java RDF toolkit and API.

19:33:31 <DanC> bitsko, I'm still not sure what you're after. Are you really interested in a language-laywer-ese answer? or do you want to know what tools grok? or are you designing some XML format where you want to embed RDF?

19:33:47 <nwalsh> Thanks, DanC (et. al.) I have a bunch of new ideas to think about. I really need to think some more now. But I think jpegrdf 1.2 will be better for it :-)

19:33:58 * DanC can't wait

19:34:25 * nwalsh thought he'd spend tomorrow morning fiddling with rssview, but no, I guess not :-)

19:34:56 <bitsko> first, an XML format; second, what tools grok; third, not too interested in a language-lawyer answer, no. if the answer is, "no, no tools support that", that's good enough for me

19:35:29 <danbri> it's generally best to leave the wrapper rdf:RDF in, imho

19:36:00 <DanC> tools are pretty interoperable when the root element is rdf:RDF. Some tools (e.g. cwm with a particular command-line switch) will look for rdf:RDF below the root.

19:36:33 <dfg_olin> danbri: i was thinking of the jena rdql

19:36:40 <DanC> I dunno if RDF parser APIs can do "parse from here down, as if it were wrapped in rdf:RDF"

19:36:52 <Epesh> ugh

19:37:06 * DanC thinks that would be much easier to do with dv's XML reader API than with SAX.

19:37:13 <Epesh> why would you use RDF - a validatable format - if you're not going to use valid data?

19:37:37 <DanC> who says his data isn't valid? maybe it's got some other structure at the root, but somewher inside it uses RDF syntax.

19:37:38 <danbri> Its a matter of taste I think, dfg_olin. I prefer the simplest subset of rdql/squish, without the '<', '>' and whatnot that Libby and Andy added...

19:39:01 <dfg_olin> thanks. I started with jena, but think I will check out squish.

19:39:37 <Epesh> danbri: hmmm

19:40:15 <bitsko> there's two angles to where the RDF will be used, one is in HTML pages (where precedent already has rdf:RDF) and in a RESTful interface using SOAP message wrappers around an rss:item, which is an RDF resource. the question is whether the latter really needs an rdf:RDF, and I'm pushing that it should, natch

19:58:06 <Epesh> What is ETags?

19:58:16 <Epesh> Ted Leung mentions them but doesn't say what it is

19:58:28 <Morbus> kinda like a unique tag for the content being server over http.

19:58:34 <Morbus> if the etag changes, the content has changed.

19:58:45 <Morbus> its an http header, like Last-Modified.

19:59:04 <Epesh> Morbus: *ponders* Why wouldn't you use the RDF tags for that, too?

19:59:19 <Epesh> dublin core has one for that, IIRC....

19:59:26 * Epesh could be very wrong

19:59:27 <Morbus> i dunno the context in which you're asking.

19:59:35 <Morbus> i just got "here", mentally.

20:00:01 <Epesh> Morbus: I really don't have a context outside of ted leung saying RSS aggregators needed to support ETags and gzipped streams

20:00:19 <Morbus> ah. the point then, is to not download all the data.

20:00:29 <Morbus> if you stick it as part of the rdf, the aggregator has to download the file, then parse, then compare.

20:00:45 <Morbus> in the ETags case, you could just get the headers, compare etags, and then move on if they're similar.

20:00:48 <Epesh> ah, I see

20:02:00 <Epesh> Ahhh, I see how I'd use it, too

20:02:14 <Epesh> although for me, parsing an RSS feed is really pretty inexpensive

20:02:23 <Morbus> and downloading it may be too, but not for the host.

20:02:44 <Morbus> you're saving a host 300% bandwidth, easy, if you do If-Modified-Since, or ETags, or other similar checks.

20:03:01 <Morbus> likewise with supporting gzipped streams.

20:03:33 <Morbus> (ie. headers are 1k, most rss feeds are 4k+, 3k savings if you check headers first).

20:04:01 <Epesh> ja, makes sense

20:04:20 * Epesh has no pity for his host, so it hasn't mattered :)

20:04:29 <Epesh> Any recommended format for the ETag?

20:04:53 <Morbus> i don't remember how Apache does it. hold on.

20:05:43 <Epesh> Morbus: I'm sorry, I didn't mean to have you do all the work :)

20:05:54 <Morbus> don't worry - i give up quick ;)

20:05:59 <Epesh> I really did find the ETag reference on w3c once I knew what it was I was looking for :)

20:06:11 <Morbus> ah. ok. yeah.

20:06:29 * Morbus gives up.

20:07:14 <Epesh> Are you running an apache server (that has ETag support?)

20:07:28 <Morbus> i think so, yeah ;)

20:07:53 <Epesh> would you give me your hostname? (I don't actually run Apache, so ...)

20:08:20 <Morbus> yup: http://disobey.com/dnn/index.xml

20:08:24 <Morbus> ETag: "99bd-a076-3e9b4f96"

20:08:37 <Morbus> etags work only on static pages - so, http://disobey.com/ is a dynamic, so no etags.

20:08:43 <Epesh> hmmm.... I wonder if I should follow that

20:08:57 <Epesh> but... aren't rss feeds dynamic by definition?

20:09:07 <Morbus> if you're serving ETags, you should just MD5 the content and serve the hash.

20:09:13 <Morbus> i don't think so.

20:09:25 <Epesh> I was thinking about the Java hashcode :)

20:09:41 <Morbus> i post on some sites once a day - i don't want the extra "overhead" of 3000 angry aggregators hitting a feed that hasn't changed, but is being generated dynamically

20:09:43 <Epesh> I just didn't know if ETags needed to be formatted a specific way and the W3C just wasn't telling me as usual

20:09:55 <Morbus> apache does 'em via inode, etc.

20:10:05 <Morbus> see http://httpd.apache.org/docs-2.0/mod/core.html#fileetag

20:11:06 <Epesh> ah, cool

20:11:21 <Epesh> Morbus: thank you. It'll take some time to get etag support in, but that's fine :)

20:11:26 <Morbus> sure.

20:11:37 <Morbus> you've an aggregator?

20:11:43 <Epesh> yes and no

20:11:50 <Morbus> heh.

20:11:51 <Epesh> It's going alpha tomorrow

20:11:55 <Morbus> in java?

20:12:07 * eikeon does not see why ETages couldn't be used for dynamic pages too.

20:12:09 <Epesh> I have a *reader* - one that need not concern itself with ETags. But I also have an aggregator that uses it. :)

20:12:33 <nwalsh> Sigh. Norm should not be fooling with this stuff. But he is anyway.

20:12:33 <Epesh> Morbus: I'll be trying to make a splash with it tomorrow, when it goes alpha :)

20:12:37 <eikeon> ... just no so easy to calculate the ETag for them.

20:12:42 <Morbus> eikeon: you can force it on dynamics, but you need XBitHack on for Apache, and then chmod 755 all your dynamic pages. it's a pain.

20:13:12 <nwalsh> I have two files. Do they constitute valid RDF?

20:13:15 <eikeon> Ah... I was thinking in general... not in terms of apache.

20:13:22 * Epesh doesn't use apache anyway

20:13:24 <Epesh> piece of crap

20:13:33 <Morbus> excuse me? <g>

20:13:35 <nwalsh> Oops. Sorry. I've overlapped your conversation.

20:13:59 * Epesh grins. I haven't run apache on any of my servers in a long, long, long, long time. Years.

20:14:08 * Morbus grumbles some more.

20:14:26 <Epesh> Morbus: jealous? :)

20:14:31 * eikeon uses apache for static content... but not for dynamic.

20:14:33 <Morbus> lol. not in the least ;)

20:14:38 <Epesh> eikeon: not me

20:14:42 <Morbus> apache has done everything i've ever need it to do. it's been good to me.

20:14:49 <Morbus> especially considering it launched my writing career ;)

20:14:54 <Epesh> Morbus: Really?

20:14:58 <Morbus> yup.

20:15:10 <Epesh> Morbus: what did you write?

20:15:17 * Epesh looks over at his bookshelf

20:15:19 <Morbus> http://www.oreillynet.com/cs/catalog/view/au/779

20:16:03 <Epesh> ah, I see

20:16:13 <Epesh> I have only their "apache: the indefinite guide" book

20:16:43 * Epesh has yet to be truly impressed with that one

20:17:48 * Morbus shrugs.

20:17:50 <Morbus> never read it :)

20:20:28 <Epesh> It's not very good

20:21:27 <Morbus> i know the guy who wrote it ;)

20:21:31 <Morbus> one of them, at least.

20:21:38 <Morbus> he uses my aggregator.

20:21:59 <Epesh> what's your aggregator written in?

20:22:16 <Morbus> crossplatform Perl, without needing Perl.

20:22:21 <Morbus> mac/win/lin.

20:22:36 <Epesh> which one is it?

20:22:41 <Morbus> amphetadesk.

20:23:12 <Epesh> I think I've tried that one. Not sure.

20:42:03 <nwalsh> brb

22:16:49 <danbri>http://www.pipetree.com/qmacro/2002/May/24#etag

22:16:49 <dc_rdfig> L: http://www.pipetree.com/qmacro/2002/May/24#etag from danbri

22:16:56 <danbri> L:|ETag-enabled wget

22:16:57 <dc_rdfig> Titled item L.

22:29:49 <danbri> <danbri> I'm trying to use etag support against w3.org... Am I missing something?

22:29:49 <danbri> <danbri> HEAD http://www.w3.org/ | grep ETag

22:29:49 <danbri> <danbri> ETag: "3ea6d375;3e2eee38"

22:29:49 <danbri> <danbri> then... wget -U 'blagg/0+4i+ (wget.pl/0+1b)' --timeout=60 -s --quiet --header='If-None-Match: 3ea6d375;3e2eee38' -O - http://www.w3.org/

22:29:49 <danbri> <danbri> ...always retrieves content

22:29:51 <danbri> <danbri> Shouldn't it return nothing? (I tried adding quotes around the value too...)

22:29:56 <danbri> anyone know what I'm doing wrong?

22:41:36 <Morbus> danbri: at first glance, that looks like it should work. researching

22:41:50 <danbri> based on http://www.pipetree.com/qmacro/2002/May/24#etag

22:42:02 <danbri> I was expecting first time it runs to see a body returned, 2nd time, nothing...

22:46:06 <Morbus> man alive, i don't have HEAD installed.

22:46:11 <Morbus> someone dcc me HEAD ;)

22:57:06 <wkearney_away> wkearney_away is now known as wkearney99

23:02:01 <Morbus> danbri, i don't have a clue. i've been testing against my own server with LWP::UserAgent, and same results.

23:02:55 <Morbus> actually.

23:02:55 <Morbus> nevermind.

23:02:57 <Morbus> i think i got a bead.

23:04:33 <Morbus> yeah.

23:04:45 <Morbus> [[[

23:04:51 <Morbus> #!/usr/bin/perl -w

23:04:51 <Morbus> use strict; use LWP::UserAgent; use HTTP::Request;

23:04:53 <Morbus> my $ua = LWP::UserAgent->new;

23:04:53 <Morbus> my $req = HTTP::Request->new(GET => "http://www.w3.org/");

23:04:53 <Morbus> $req->header('If-None-Match' => '"3ea6d375;3e2eee38"');

23:04:54 <Morbus> my $res = $ua->request($req);

23:04:56 <Morbus> print "code was ", $res->code, "\n";

23:05:00 <Morbus> ]]]

23:05:02 <Morbus> that prints 304.

23:05:07 <Morbus> the quotes were necessary.

23:05:20 <Morbus> a raw dump of the headers I sent:

23:05:34 <Morbus> 'user-agent' => 'libwww-perl/5.69',

23:05:34 <Morbus> 'if-none-match' => '"3ea6d375;3e2eee38"'

23:06:24 <danbri> thanks. odd that wasn't working w/ wget though...

23:06:35 <danbri> does case of headers matter?

23:06:47 <Morbus> yeah, i duplicated the non-working with wget. lemme try again.

23:06:48 <danbri> logger_1, pointer?

23:06:50 <Morbus> seems lwp always lowercases then.

23:06:53 <danbri> logger_2, pointer?

23:06:53 <danbri> See http://ilrt.org/discovery/chatlogs/rdfig/2003-04-24#T23-06-53

23:07:01 <Morbus> or, at the very least, internally, it lowercases.

23:07:10 <Morbus> i didn't try wget with quotes. hold.

23:07:17 * danbri holds

23:07:50 <Morbus> well.

23:07:57 <Morbus> minimalistically, it seems to work.

23:07:59 <Morbus> [[[

23:08:03 <Morbus> morbus@disobey:~ > wget --header='If-None-Match: "3ea6d375;3e2eee38"' http://www.w3.org/

23:08:03 <Morbus> --19:09:11-- http://www.w3.org/

23:08:07 <Morbus> HTTP request sent, awaiting response... 304 Not Modified

23:08:07 <Morbus> 19:09:12 ERROR 304: Not Modified.

23:08:07 <Morbus> ]]]

23:11:18 <Morbus> if i use your pasted line, save for the quoting and removal of --quiet, it works too.

23:11:32 <Morbus> GNU Wget 1.8.2,

23:13:11 <danbri> ah, it was the \ before " ie \" not needed.

23:13:13 <danbri> thanks :)

23:13:46 <Morbus> aaah.

23:13:50 <Morbus> cool.

23:16:47 <danbri> i'll add etag and robots.txt support to my harvester at the weekend now, cheers!

23:17:34 <Morbus> robots.txt? what lang? i wouldn't mind having robots.txt support.

23:18:35 <danbri> ruby, but it shouldn't be too different from the perl...

23:26:34 * danbri -> zzz; g'nite!