Semantic Web Interest Group IRC Chat Logs for 2003-12-16

00:32:57 <eaon|away> eaon|away is now known as eaon

00:57:21 <icepick> icepick is now known as icepick-away

01:04:51 <danbri> sbp`! any news re domain name trauma?

01:05:10 <sbp`> yeah. I think infomesh.net is back up and running

01:05:18 <sbp`> thanks to everyone for the concern

01:05:41 <danbri> phew

01:05:46 <danbri> hope it didn't break the bank

01:06:01 <danbri> i tried calling you, the nos in whois, but couldn't get thru

01:06:23 <sbp`> nope, thankfully. I didn't deal with it all directly so I'm not much more informed than anyone else

01:18:26 <ChanServ> [#rdfig] This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome

01:30:22 <ChanServ> [#rdfig] This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome

02:51:52 <Davey> Davey is now known as D[a]vey

03:00:36 <x86processor> hello

04:39:43 <kota> ah, oh, RDF and OWL became PR...

04:40:15 <kota> I've just finished printing all the documents...

04:41:50 <danbri> how many pages?

04:42:23 <kota> all of them,,, hmm,,, i donno how many...

04:42:39 <danbri> just curious!

04:42:41 <kota> the biggest one was rdf primer, over 100 pgs.

04:43:26 <kota> heh, yeah, i can check the diffs.

04:43:32 <kota> but, what a nice timing...

05:06:46 <_karlcow> and it's a primer....

05:10:26 <danbri> evening, karl

05:13:29 <_karlcow> evening dan ;)

05:13:46 <_karlcow> Imagine now that I have to translate that in french :p

05:20:25 <kota> good job ;)

05:23:45 * danbri has http://www.daemonhost.com/phriday/tbl1.jpg drawn to his attention

05:26:33 <_karlcow> lol @dabvri

05:26:42 <_karlcow> even danbri

06:27:01 <anselm> heh

10:28:36 <fidothe> morning

10:48:09 <iwaim> iwaim is now known as iwaiAway

11:20:38 <Davey> Is there any Schema already defined for storing addresses?

12:18:01 <swh_> swh_ is now known as swh

13:26:13 <_karlcow> _karlcow is now known as karlcow

13:52:11 <dajobe> anyone around here got OSX 10.3 and would be willing to test some code for me?

13:52:59 <darobin> dajobe: depends on the intensity of the testing and your deadline :)

13:53:14 <darobin> I can do it, but in another ~2h

13:53:27 <dajobe> tarball, configure, make, make install, tell me what happens

13:54:39 <dajobe> ... would be nice to have an account somewhere to do proper testing, we've only got 10.2 on our ibook

13:55:48 <dajobe> sourceforges' compile farm is also on 10.2

13:56:58 <darobin> dajobe: gimme a pointer, if it's just that I'll launch it in the bg

13:58:27 <dajobe> ok, get http://www.redland.opensource.ac.uk/tmp/osx/redland-0.9.15.tar.gz

13:58:31 <dajobe> then ./configure --with-perl

13:58:34 <dajobe> make; make check; make install

13:58:51 <darobin> k

13:59:10 <dajobe> if it works, you'll need to rm -rf /Library/Perl/darwin/RDF /Library/Perl/darwin/auto/RDF

13:59:32 <darobin> or I can keep it and toy with it :)

13:59:37 <dajobe> sure :)

14:00:24 <dajobe> oops, it'll probably install into /usr/local also

14:00:30 <darobin> that's fine

14:00:37 <darobin> my Perl install is a kitchen sink already

14:01:23 <darobin> launching configure

14:01:44 <karlcow> *configure is long :p*

14:02:06 <dajobe> fwiw, it worked ok on 10.2 before I asked

14:02:17 <dajobe> but 10.3 has perl 5.8.x which broke things

14:02:27 <darobin> wow, you have more language bindings than I thought if configure is any indication

14:02:32 <karlcow> Raptor build summary:

14:02:32 <karlcow> XML parser : libxml(system 2.5.4)

14:02:32 <karlcow> WWW library: libxml(system 2.5.4)

14:02:34 <karlcow> Redland build summary:

14:02:34 <karlcow> Berkeley/Sleepycat DB: Missing

14:02:34 <karlcow> Languages available: perl(5.8.1) python(2.3) java(1.4.1_01) tcl(8.4) php(4.3.2) ruby(1.6.8)

14:02:35 <karlcow> Language APIs built: perl

14:02:37 <karlcow> Triple stores: memory hashes(memory)

14:02:39 <karlcow> RDF parsers: raptor(internal)

14:02:41 <karlcow> Content digests: md5(openssl) sha1(openssl) ripemd160(openssl)

14:02:41 <darobin> ah, I don't have stock 5.8 from the 10.3 install

14:03:15 <darobin> I have 5.8.2 built from source -- the stock one is a bit broken in itself (blame Apple, it's not even a released Perl)

14:03:30 <dajobe> I guess make check might not work without BerkeleyDB, I forget. Hmm.

14:03:44 <dajobe> I kinda assume people install fink to get all the bits OSX omits.

14:03:47 <darobin> I think I have it

14:04:07 <karlcow> dajobe, I have made a clean install one week ago for 10.3

14:04:22 <darobin> ah, it didn't find BerkelyDB even though I have it, is that an issue?

14:04:24 <dajobe> if bdb's not in the usual place, there's a configure arg --with-bdb=/wherever

14:04:25 <karlcow> ;) and many things I had installed, had... been removed

14:04:26 <karlcow> :)

14:04:30 <darobin> ok

14:04:36 <dajobe> karlcow: ok

14:04:48 * dajobe curses bdb again

14:05:53 <darobin> yeah, by default it installs itself in the weirdest place

14:06:04 * darobin reruns configure with the flags to find bdb

14:08:17 <dino> dino is now known as dinoff

14:11:19 <darobin> making cjeck

14:11:43 <darobin> some fail

14:13:24 <darobin> the perl tests fail, not finding /usr/local/lib/libraptor

14:13:43 <darobin> ah

14:13:51 <darobin> you can't run make check before make install

14:14:03 <darobin> otherwise, you're testing the previously installed version, if any :)

14:14:37 * darobin rechecks after install

14:14:40 <dajobe> hmm, I thought I fixed that linking

14:15:23 <dajobe> if it works after install, that'll be good

14:15:31 <darobin> the perl tests are looking for the installed lib which is a normal thing for perl tests to do. I think Perl tests should be run later

14:15:39 <darobin> all succesful after install

14:15:48 <dajobe> great! thanks a lot

14:16:05 * darobin bows, most happy to help the advancement of Perl and RDF at the same time :)

14:16:10 <LotR> darobin: usually make test uses blib/* no?

14:16:19 <darobin> LotR: not for C libs

14:16:26 <darobin> only for its own stuff

14:16:27 <LotR> that is stupid

14:16:34 <dajobe> IIRC perl thinks it knows better about the link lines

14:16:35 <darobin> it likely is

14:16:43 <dajobe> and rearranges the -L args so they don't work

14:17:13 <darobin> well, MakeMaker is a little overeager sometimes indeed

14:17:28 <LotR> MakeMaker is *dumb*

14:17:54 <LotR> it isn't hard to do everything from auto* for perl

14:18:00 <darobin> dumb, obfuscated, over-eager, and beats every other solution I've seen in any language save Module::Build :)

14:18:22 <LotR> MakeMaker certainly doesn't beat auto*

14:18:54 <darobin> last time I looked into auto it was a pain to create a new project

14:18:59 <darobin> perhaps that's changed

14:19:21 <LotR> you just copy from an old project. easy.

14:19:28 <darobin> heh

14:19:31 <LotR> just like you do with MakeMaker

14:19:57 <darobin> I can write Makefile.PL by hand, I just don't because mkpm writes it for me

14:21:50 <LotR> it isn't harder to write a configure.in by hand that does as little as a Makefile.PL

14:22:21 * darobin will have to look again

14:22:57 <LotR> of course, for C you'll usually need more than a Makefile.PL does :)

14:23:43 <LotR> I think auto* have improved quite a bit recently, but I haven't used the newer versions

14:25:17 <dajobe> if you want to be scared, read the Makefile.PL at the top level of redland-0.9.15 dir

14:25:42 <dajobe> it's part of the configuration to make redland work inside CPAN

14:26:48 <LotR> if I want to scare myself, I look at rdf :)

14:27:05 <dajobe> it's just triples, not anywhere as hard as xml

14:27:11 <darobin> ick

14:27:29 <darobin> indeed, that's not the simplest mf.PL :)

14:27:42 <LotR> dajobe: I guess my brain is wired weird. I vastly prefer XML :)

14:27:50 * darobin likes both

14:28:10 <dajobe> maybe you can explain about validation of external dtd subsets then?

14:28:25 * darobin giggles

14:29:05 <darobin> dunno if LotR plans to answer that one (provided the question was indeed for him) but of course DTDs Must Die

14:29:17 <darobin> painfullly

14:30:00 * eric is revisiting his XML 2003 notes to write an article for XML.com

14:30:30 <eric> are you guys aware of anything similar to Kal Ahmed Topic Maps Design Patterns in RDF?

14:31:04 <LotR> I'm not an xml validator or anything. I just think the xml formats I've seen are much easier to grok than the rdf I've seen. I'm sure you can obfuscate either more than I'd like to be aware of

14:31:17 * eric would think that RDF Schema should help a lot if we wanted to do something similar.

14:31:42 <dajobe> LotR: you are clearly sticking to a small XML subset. Similarly you can do that with rdf.

14:32:09 <LotR> dajobe: but minimal XML is much more minimal than minimal rdf it seems

14:32:22 <dajobe> haha

14:32:29 <LotR> maybe I should say rdf/xml

14:32:40 <LotR> which seems to be the dominant format

14:32:40 <dajobe> ah, comparing apples to elephants

14:35:35 <LotR> *shrug*

14:38:04 <dajobe> if you want to know my thoughts, read http://www.ilrt.bris.ac.uk/discovery/2003/11/new-syntaxes-rdf/

14:39:43 <kota> new syntax, that's interesting

14:40:07 <dajobe> I feel the need to create an IslandTopic just to annoy Danc ;)

14:40:28 * DanC is gonna shut down the wiki if folks won't use it as a wiki

14:40:39 <DanC> ;)

14:42:31 <dajobe> gah

14:42:43 * eric hates one of the XML design reqs (QNames in attributes values)

14:42:59 <dajobe> yeah, but people brought it up

14:43:24 <DanC> QName is a requirement? where?

14:43:47 <dajobe> it's on rdfcore issues list

14:44:00 <dajobe> and with it in XSD, XSLT

14:44:01 <DanC> re design patterns, yes, see the esw wiki...

14:44:05 <DanC> .google esw wiki

14:44:06 <datum> esw wiki: http://esw.w3.org/topic/FrontPage

14:44:09 * darobin wonders if we can really do without QNames and remain author-friendly

14:44:15 <dajobe> darobin: suits me

14:44:24 * dajobe fails again to find anything on the wiki

14:44:28 <eric> But it has different rules for XSLT and WXS!

14:44:50 <DanC> what are you looking for, dajobe?

14:44:59 <dajobe> somewhere to even start writing a page

14:44:59 <eric> DanC: I haven't found much in the esw wiki

14:45:11 <darobin> eric: that's ok so long as you say which you use

14:45:24 <darobin> both are useful

14:45:32 <dajobe> "NOTE: Please use the back-links from CategoryFaq " <- wtf?

14:45:58 <DanC> you're not familiar with back-links?

14:46:03 <eric> darobin: I think both are ugly.

14:46:07 <darobin> heh

14:46:25 <DanC> one of the more active design pattern topics, though it's still very much in progress, is http://esw.w3.org/topic/GoodURIs

14:46:52 <darobin> eric: it's not that I'm in love with them but I need something that can be unique enough (namespaced, so ID is out), that doesn't require me to type in an URI, and that doesn't require entities

14:46:53 <DanC> I like to think http://esw.w3.org/topic/NaryRelations is a good topic; I recently factored several patterns out of it.

14:46:55 * dajobe adds a FAQ: Please say how to use-back links from CategoryFaq

14:46:58 <DanC> the overall pattern language is pretty raw

14:47:47 <DanC> say how to use back links? you just make a link, and the back-links work. Have you ever used a wiki before? I have a hard time explaining this, cuz I can't remember what it's like to not know it.

14:48:10 <eric> darobin: I have nothing against using prefixes, but the one we use for attribute values should be declared separately from namespace declarations.

14:48:14 <darobin> eric: I'd be happy to see DSDL Part 6 find a better way ;)

14:48:55 <darobin> eric: that would avoid many problems, yes indeed

14:51:46 * DanC can't see where dajobe added "Please say how to use-back links from CategoryFaq"

14:52:19 <dajobe> I daren't

14:53:49 <DanC> I edited http://esw.w3.org/topic/FaqIdeas ; is it clear now?

14:53:55 <dajobe> so I made http://esw.w3.org/topic/MinimalRdf

14:54:36 <DanC> hm... why did you start the text with a heading? I don't understand why people do that. There's already a heading above.

14:55:04 <dajobe> SinceWikiNamesAreHardToRead

14:55:34 <DanC> ah; the c2.com wiki puts the spaces in the heading. we could do that.

14:57:20 <dajobe> FaqIdeas still doesn't answer how I use these WikiBadge things. What do I type?

14:57:50 <DanC> it's not clear to me that MinimalRdf is a pattern; i.e. a problem/solution pair. If it's one person's design-sketch, it's not really clear how other community members (e.g. me) are expected to contribute to it. If you really just want to express your own thoughts, I wonder if the wiki is the best place.

14:58:15 <DanC> sandro and i made up a BlueSky badge for shared design sketches. is that what this is?

14:58:50 <dajobe> <LotR> dajobe: but minimal XML is much more minimal than minimal rdf it seems

14:59:10 <dajobe> seems worthy of answering or investigating

15:02:40 * DanC adds a link from http://esw.w3.org/topic/SemanticWebArchitecture ...

15:03:39 <DanC> last modified 2003-12-16 09:03:28

15:04:41 * DanC -> telcon

15:06:38 <dajobe> oh, so you make a forward link on a page to CategoryFaq

15:06:52 <dajobe> then use the backlinks to find them

15:07:10 <dajobe> which is most easily done by clicking on the CategoryFaq title on the CategoryFaq page

15:12:15 <darobin> eric: quick question, do you agree that "\i\c*:\i\c*" defines a QName that *must* have a prefix?

15:28:27 <eric> darobin: yes, and so does ".+:.+" and (because of the lexical space of xs:QName)

15:28:51 <darobin> ah yes, that's smart, thanks a lot

15:29:11 <darobin> (but I'll stick with the more verbose form as I don't trust processors to be conformant enough)

15:30:14 <eric> :-)

15:30:59 <eric> note that \c includes colons and that your regexp still allows "a:a:a:a:" !

15:31:25 <eric> meaning that you are still relying on the lexical space of xs:QName!

15:34:32 <darobin> oh, I thought \c was the space of NCName!

15:36:03 * darobin switches to .+:.+, since this is for a Rec we should assume conformance

15:36:07 <darobin> thanks a lot!

15:43:47 <D[a]vey> D[a]vey is now known as Davey

16:15:14 <larsbot> does anyone know what the status of reification in RDF is at the moment?

16:15:22 <larsbot> is it still officially part of RDF? in the same way as before?

16:15:26 <larsbot> and do tools support it?

16:23:57 <dajobe> it's still in RDF

16:24:09 <dajobe> as with anything, some tools support it better than others

16:24:47 <danbri> reification vocab is there, but i never advocate its use

16:25:10 <libby> jena has quite good support I believe

16:25:22 <AndyS> The vocabulary for reification is still there - and clarification that reification is anout statings (instances) not statements.

16:25:49 <larsbot> so it's there, but not really recommended, is that it?

16:26:22 <larsbot> my concrete use case is TM->RDF conversion

16:26:34 <larsbot> I'm wondering what to do with reification and scope on, say, occurrences

16:26:37 <AndyS> Jena originally choose the statements interpretation but "we" (not me personally) have made sure both forms are conveniently used. The core system deals in statings.

16:26:44 <larsbot> whether to use RDF reification to capture it, or whether to just omit it

16:27:11 <AndyS> How would you capture it using reification?

16:27:25 <larsbot> AndyS: that sounds good. if I do a query, for example, will the query also take reified statements into account (as if they'd been written in non-reified form)

16:28:21 <larsbot> well, say I have an occurrence "homepage" that is "http://www.example.com" for topic X

16:28:41 <AndyS> Yes - if you mean that 4 RDF triples appear in the model (they may be held more compactly but the API can't see that)

16:28:46 <larsbot> I'd have to reify (X, homepage, "http://..."), then attach the scope with a tm:scope predicate

16:29:18 <larsbot> would a query react differently to {X, homepage, "..."} and to the same thing in reified form?

16:29:35 <larsbot> "reified form" == 4-triple form

16:29:53 <AndyS> In RDF, the reification of a statement does not presume that the ground statement is asserted.

16:30:29 <AndyS> But, as I understand your case, you do not want the original statement to be universally true.

16:30:57 <AndyS> A query can query the reification or the ground assertion. or both.

16:32:18 <AndyS> IMHO reification is not pretty but it can do a lot. The context mechanisms proposed are a somewhat neater way of achiveing these effects. It just that different properosal do different things.

16:32:37 <larsbot> it's not pretty, no

16:32:54 <larsbot> I didn't realize there was a distinction between reified-and-asserted statements and reified-and-unasserted statements

16:33:04 <larsbot> that makes a certain amount of sense, though it does make my life harder

16:33:17 <AndyS> :-)

16:33:26 <larsbot> scope is used to qualify statements in TMs, so probably I should go for the reified-and-unasserted variety

16:33:49 <larsbot> reification is usually *not* used for qualification, so I should probably go the other route for that

16:34:14 <larsbot> however, some statements turn into RDF nodes in my conversion, and those can't be unasserted (*argh*)

16:34:25 <larsbot> anyway, this helps a lot

16:34:31 <larsbot> so thanks for the info

16:34:39 <larsbot> what are the context mechanisms being proposed? any URIs?

16:35:58 <AndyS> Usually quads (a fourth arge to the "triple") but the meaning of the fourth slot, while a URI, can be different.

16:36:33 <larsbot> is the fourth slot effectively the identity of the statement?

16:36:43 <larsbot> or is it the identity of the context?

16:36:48 <dajobe> either

16:37:00 <dajobe> search for 16:18 Contexts in http://www.w3.org/2001/sw/Europe/events/20031113-storage/

16:37:18 <larsbot> interesting. this is precisely what we Ontopians have argued (internally) that RDF should have been :-)

16:37:29 <larsbot> though we would have gone for the first option

16:37:58 <larsbot> certainly would have made TM->RDF conversion enormously much easier

16:38:10 <AndyS> 4th slot: can be the id of the statement (or stating) or can be the source of the statement (c.f. RSS aggregation) or can be an internal truth maintenance thing

16:39:01 <larsbot> you can build all the rest on the first, though...

16:39:44 <larsbot> plus more (like the data type stuff & xml:lang)

16:40:03 * dajobe groans, another triple of 6 parts or so

16:40:16 <larsbot> nope. separate statements about the statment...

16:41:11 <AndyS> You can build anything on statements but it is like assembler. And thats what reification gives you. Need conventions on top of teh building block

16:41:30 <AndyS> Bother - s/statements/statings/

16:41:33 <larsbot> yep

16:57:41 <reagleBRKLN>http://lists.kde.org/?l=kde-bindings&m=107102539403704&w=2

16:57:42 <dc_rdfig> A: http://lists.kde.org/?l=kde-bindings&m=107102539403704&w=2 from reagleBRKLN

16:57:51 <reagleBRKLN> A:RDF/KDE bindings

16:57:52 <dc_rdfig> Added comment A1.

16:58:32 <reagleBRKLN> they might start a rdf.kde.org !

17:01:52 <evlist> sounds good!

17:04:05 <MarkB> A:eek, "Since every method

17:04:06 <dc_rdfig> Added comment A2.

17:04:27 <MarkB> oops

17:05:45 <MarkB> A2:eek, "Since every method in KDE will have a URI"

17:05:46 <dc_rdfig> Replaced comment A2.

17:11:49 <dajobe> A:

17:11:49 <dc_rdfig>http://lists.kde.org/?l=kde-bindings&m=107102539403704&w=2

17:11:50 <dc_rdfig> (1:reagleBRKLN) RDF/KDE bindings

17:11:51 <dc_rdfig> (2:MarkB) eek, "Since every method in KDE will have a URI"

17:11:55 <dajobe> A:|RDF/KDE bindings

17:11:56 <dc_rdfig> Titled item A.

17:12:01 <dajobe> A1:""

17:12:01 <dc_rdfig> Deleted comment A1.

17:21:10 * sandro sheds a small (but heartfelt) tear at one more deployment of RDF with URIs with fragments. Sigh.

17:27:37 <AndyS> Where is the fragment?

17:29:30 * AndyS getting to like using URNs for concepts not on (no representation) the net.

17:30:17 <dajobe> like info: ?

17:30:56 <AndyS> Like people? :-)

17:31:40 <sandro> But it's sooooo nice to be able to just use any browser to get good information about the thing. (we need more demos of that fact, of course.)

17:32:02 <sandro> And that doesn't work well with fragments or with URNs.

17:32:24 <AndyS> Oh yes. But somethign have no page. And I am not my home page.

17:32:33 <dajobe> most of the big ones I've seen recently don't use frags - CSAktors for example

17:33:01 <AndyS> How can you identifiy me as opposed to a page about me? (pace topic maps)

17:33:06 <Wack> actually an URI like 'http://rdf.kde.org/kde/3.2/kdeui#KPushButton::addButton(const+QString&,+bool)' is used to describe an abstract resource, but is an rdf document as well

17:33:31 <Wack> well, no, not exactly because of the fragment ...

17:33:53 <AndyS> So is that a thing(concept)-inside-a-page?

17:34:08 <larsbot> :-)

17:34:13 <sandro> AndyS, I do it via 303 redirects. http://esw.w3.org/topic/SlashRedirection

17:35:49 <AndyS> So what is the URL and what is the URL after redirection? Both pages? Or is it that the 302 has a "#" in it that makes it special?

17:36:15 <karlcow> karlcow is now known as _karlcow

17:36:16 <Wack> perhaps using <rdf:seeAlso> is a better way to explicitly reference an RDF (or not rdf, for that matter) document that further describes an otherwise undereferenceable URI

17:36:56 <Wack> but then you'd have to have the seeAlso information in advance

17:37:13 <sandro> No, the # is irrelevant. EG http://hawke.org/2003/canon-g3/002193 names a picture I took (in theory -- not working right now), but a GET of it gives you a redirect to a web page about that picture (including the picture itself, and some metadata).

17:37:55 * bijan recalls some discussion way back when twixt himself and Daniel Krech about a scheme to allow some standard operators on uris to get "related" identifed objects.

17:38:24 <sandro> (But if you use #, the server cannot gracefully handle bad URI-Refs. And if you do it in the original URI (the URI of the thing) you can't serve both RDF/XML and HTML.)

17:38:25 <AndyS> So how do I GET (a rep) of the picture? Seems not to separate description and representation.

17:38:50 <bijan> Similar to the general bnode, definite description technique

17:39:18 <sandro> Okay -- a picture is actually a bad example, since it can have respresentations in some sense.

17:39:57 <AndyS> No - its a good example as there is a representation and a description and they are different.

17:40:04 <sandro> http://hawke.org/2003/sandro is *me*. there is no way to serialize me, so you be redirected to something which can be serialized.

17:40:04 <dc_rdfig> B: http://hawke.org/2003/sandro from sandro

17:40:11 <Wack> this has been discussed before, there's some pointers in http://esw.w3.org/topic/GoodURIs

17:40:31 <AndyS> OK - is this because you say it is you?

17:41:04 <Wack> and http://esw.w3.org/topic/OverloadedUri

17:41:15 <AndyS> And there is no web page there? Else what if I say <http://hawke.org/2003/sandro> :length "2345 bytes"

17:41:15 <sandro> (Wack, I wrote all that stuff.)

17:41:40 <sandro> RIght, AndyS, there is no web page there. There are no representations of me.

17:41:46 <_karlcow>http://www.w3.org/TR/chips/#gl1

17:41:46 <dc_rdfig> C: http://www.w3.org/TR/chips/#gl1 from _karlcow

17:41:56 <Wack> (sandro: uuh ;)

17:42:02 <AndyS> I can see how it works IF there is no web page.

17:42:09 <_karlcow> C:| Chips, Guideline 1: Choose URIs wisely

17:42:09 <dc_rdfig> Titled item C.

17:42:46 <AndyS> It is when there is web page - does that mean it is the web page and not "you"?

17:42:54 <sandro> As for the photograph -- similarly no representations, no serializations, because to have that stuff you need to pick resolutions and stuff. So the redirect takes you to some RDF or HTML which gives you names for the picture at various sizes, etc.

17:43:16 * DanC notes GoodURIs and SubjectIndicator chat... whee!

17:43:18 <sandro> Sorry -- "wrote all that stuff" is not really fair -- other people have contributed to some of those pages, esp DanC.

17:43:36 <AndyS> Don't follow. All that stuff is just rep (MIME) negotiation.

17:44:09 <AndyS> There is still the difference of a picture and the thing the picture is of (e.g. Paris)

17:44:12 <sandro> How do you say whether you want the 1600x1200 or the 400x300 version of the picture via Mime?

17:44:19 <sandro> s/Mime/Con-Neg

17:44:36 * AndyS always wants the highest res piccy!

17:45:04 * DanC reviews http://esw.w3.org/topic/SubjectIndicator to see it should persuade AndyS that http URIs with fragments are better than URNs

17:45:12 <sandro> If there is a web page there, that's pretty good evidence to me that the URI does not name a person, or other physical object, or anything else which cannot be serialized.

17:45:37 <bijan> Hmm.

17:45:42 <bijan> Serialized

17:46:01 <bijan> I can be serialized, can't I?

17:46:06 <AndyS> DanC - I can see how a number of appraoches work. #frag or URN to me makes little difference as they both differentiate description and representation

17:46:28 <bijan> Serialization is a strange criterion

17:46:31 <AndyS> Sandro - what if a web page appears at that URI? Does meaning change? Or is it just bad practice?

17:46:41 <DanC> ah, but http#frag grounds the term in communication protocols and practice in a way that URNs don't.

17:46:48 * AndyS follows DanC's link and rereads.

17:47:05 <sandro> My actual URIs for pictures are like http://example.com/034534/size=1600x1200/crop=4,5,102,61/gamma=0.45 which is sometimes incredibly useful. THere are nigh-infinite interesting respresenations of more-or-less the same "image"

17:47:07 <AndyS> But but but ... :-) ... HTTP does not see the #frag

17:47:58 <DanC> the only reason I've seen people use URNs is that (A) they don't want to be accountable for their garbage, e.g. look at the use of unregistered URN namespace id's (e.g. urn:microsoft:...) or (b) they're too lame to run a web server and keep it still.

17:47:59 * AndyS notes it is easier to poke at the proposals than come up with one of his own

17:48:16 <sandro> If a web pages appears at that URI.... well, then you have conflicting information. Which seems like bad practice, but I really can't be sure. Maybe the state of the world has legitimately changed in a way which makes that legitimate.

17:48:21 * AndyS apologies for that :-)

17:49:00 <AndyS> I can see different, all workable, ideas here. There "just" has to be one way forward.

17:49:29 <DanC> er... sandro, pictures aren't very good examples of things that one might want to use a URN for; pictures are, pretty much, documents. to "represent" a picture with a sequence of bytes and a mime type seems much, much more natural than to represent a person that way.

17:49:38 <bijan> DanC: That's interesting...Kendall was just wondering the other month why people didn't use URNs more often, especially for things they didn't want to have dereferencable stuff for

17:50:01 <sandro> Yeah.... Redirect is the best I know, but it's not perfect. I hate that extra round-trip.

17:50:54 <AndyS> "people" are a good (testing) example. I am not my home page. If I am a #frag (no probs) then it is different for RDF and a browser (jump to #place) but workable.

17:50:56 <sandro> DanC, I agreed pictures were a bad/confusing example, but Andy wanted to use them, I think, for just that reason. And I do have a notion, myself, of pictures as abstract from particular serializable concrete forms of the picture.

17:51:04 <_karlcow>http://gmpg.org/xfn/

17:51:04 <dc_rdfig> D: http://gmpg.org/xfn/ from _karlcow

17:51:11 <DanC> my answer to "but what if I don't want to publish representations or descriptions of the thing" is: then you're not doing what you can to establish consensus around your terms.

17:51:13 <AndyS> Exactly - pictures are testing.

17:51:14 <_karlcow> D:|XHTML Friends Network

17:51:14 <dc_rdfig> Titled item D.

17:51:30 <_karlcow> D: a competitor of FOAF for HTML

17:51:30 <dc_rdfig> Added comment D1.

17:51:51 * AndyS thanks _karlcow

17:52:08 <_karlcow> D: [http://gmpg.org/xfn/xfn-foaf| FOAF - XFN comparison]

17:52:08 <dc_rdfig> Added comment D2.

17:52:17 * sandro has to run. Sorry for hit and run. :-)

17:52:23 <bijan> hmm.

17:52:33 <AndyS> I want to just adopt the right answer - not make hard/base-level decisions

17:52:42 <bijan> Well, I don't always want to do *everything* I can to establish consensu around my terms.

17:52:46 <DanC> to use a URN to establish well-known names for, say, people or countries would be like running for office without telling anybody your name or position on issues. Why would you expect to get any votes?

17:53:01 <sandro> logger, pointer?

17:53:01 <sandro> See http://www.ilrt.bris.ac.uk/discovery/chatlogs/rdfig/2003-12-16#T17-53-01

17:53:10 * bijan skeptical

17:53:28 <bijan> Especially if I use them in rich, potentially freestanding documents

17:54:17 <bijan> It's an interesting empirical matter, though.

17:54:23 <bijan> Why should I *not* expect to get any votes?

17:54:25 <bijan> :)

17:54:46 * DanC didn't say "everything", didn't mean it, wonders why bijan attacks positions nobody takes

17:54:59 * bijan notes that tons of pols get elected without telling anyone their positions

17:55:05 <AndyS> But - again - do I need to know when it is a description and when a rep (e.g. asserting the title of "it").

17:55:10 <kendallclark> oh, politics?! ergh.

17:55:10 <bijan> I didn't claim it was your position

17:55:32 <kendallclark> that was off-limits here last week! :>

17:55:44 <bijan> Though, it's hardly a unfair reading of:

17:55:46 <bijan> DanCmy answer to "but what if I don't want to publish representations or descriptions of the thing" is: then you're not doing what you can to establish consensus around your terms.

17:56:12 <bijan> How is thart an answer if I'm doing other things to establish consensus?

17:57:09 <DanC> s/doing what you can/meeting community norms and expectations/

17:57:35 <DanC> so... then you're not meeting community norms and expectations to establish consensus around your terms.

17:57:43 <bijan> Well, aren't we partly trying to see whether it's worth establishing other norms and expectations?

17:57:50 <DanC> I'm not

17:58:06 <AndyS> So - do "I" have to choose different URIs (or differ just by #frag) for me and my home page?

17:58:18 <DanC> I advise you to, yes, AndyS

17:58:29 <DanC> (as does the TAG, fyi)

17:58:38 <kendallclark> is the uri for "him" to be deref'able?

17:58:50 <MarkB> you don't have to, but concensus is easier to reach when you're unambiguous

17:58:58 <AndyS> Differ by #frag? I could have URN for me, and URL for home page.

17:59:02 <DanC> I advise using http://...address-of-homepage#fragForPerson

17:59:14 * bijan gives up

18:00:14 <AndyS> is that because GETting then pulls a rep? i.e. #frag UR for me is not completely opaque?

18:00:21 <DanC> yes.

18:00:22 <AndyS> s/UR/URI/

18:00:37 <DanC> it grounds http://...address-of-homepage#fragForPerson in common communication mechanisms and practices.

18:01:09 <AndyS> Ok- I can see how that works. Principle seems to be that one should always make a page avialable. That's cool.

18:01:47 * bijan offers the explanation of why "he attacks positions that nobody takes" that what was meant was clear from what was written; see the rewrite of DanC's position

18:02:40 * DanC puzzles over that explanation

18:07:12 <AndyS> (tries to "get" it) And so URI#frag does not name the same thing as URI(no frag).

18:07:59 <AndyS> Hence URN for URI#frag also works but can't GET on the derived URL (drop the frag). => less useful.

18:08:33 <DanC> yup

18:09:12 <DanC> sigh... the U.N. uses cookies for http://www.un.org/apps/news/story.asp?NewsID=9157&Cr=wsis&Cr1= why!?!?!

18:10:35 <libby> maybe the default with some package?...like with jakarta

18:11:41 <DanC> most likely... would be good to know what's behind it

18:12:17 <DanC> Server: Apache/1.3.22 (Unix) AuthTDS/1.1

18:12:23 <DanC> Set-Cookie: ASPSESSIONIDGGGGGGGG=HDCSZQLCDMEZJZTAQVSBFSQQPQQQOLKW; path=/

18:12:23 <DanC>

18:12:38 <DanC> no p3p headers

18:22:16 * DanC sends mail reporting the cookie problem

18:40:21 <MarkB> A:I take back the "eek", as they're just using it for identification, not invocation

18:40:22 <dc_rdfig> Added comment A2.

19:06:13 <Davey> Davey is now known as D[a]vey

19:57:16 <icepick-away> icepick-away is now known as icepick

20:59:05 <dinoff> dinoff is now known as dino

21:32:27 <D[a]vey> D[a]vey is now known as Davey

23:01:28 <Davey> Has anyone seen any publically available info on i18n of addresses?

23:03:23 <mortenf> that's a tough one, you might want to check the contact schema and the rdfweb-dev archives...

23:04:03 <mortenf> in short, don't go there unless you have plenty of time - use something like vcard instead (for now)

23:13:19 <eaon> eaon is now known as eaon|ill

23:58:58 <mdupont> mdupont is now known as md-zzz