Semantic Web Interest Group IRC Chat Logs for 2004-06-09

06:46:28 * sanctius is back (gone 10:19:01)

06:46:32 * sanctius Bonjour tout le monde / Good Morning all

06:46:47 <luuk> lu

07:34:15 <alf> If anyone has a minute, I'd appreciate some critical eyes being passed over the RDF parts of this: http://www.pmbrowser.info/rvw/0.2/ to make sure it's valid (particularly the optional use of rdf:Bag and rdf:Seq)

07:42:45 <luuk> Minor thing: rvw:reviews doesn't match rvw:review

07:43:59 <luuk> And check rvw:rating

07:47:35 <alf> Do you think that rvw:review should be in a rvw:reviews Bag even when there's only one?

07:48:50 <alf> for the rating I missed the end / - thanks

07:49:05 <luuk> also identifier

07:49:48 <alf> what about the identifier?

07:49:56 <luuk> If it's possible to have multiple reviews, then it makes sense to use a bag, even if there is only one for this particular track

07:50:05 <luuk> the closing tag is identifer

07:50:59 <alf> well spotted :)

07:51:15 <luuk> I ran it thru my xml/rdf parser/validator ;)

07:51:20 <alf> ok, i'll put them all in Bags then

07:52:09 <luuk> It complained about the comments too

07:52:41 <alf> yeah, that's ok, it's not supposed to be used directly from the page with comments

07:53:02 <alf> which parser/validator do you use?

07:53:06 <luuk> <-- should be <!--

07:53:07 <luuk> Ah, ok

07:53:16 <luuk> I created one from raptor

07:53:25 <luuk> with libxml

07:54:03 <luuk> Might be a bit dated though. It's < version 1

08:07:42 <kao_> kao_ is now known as kao

08:10:48 <alf> thanks luuk

11:38:22 * sanctius is away: ?_ absent pour le moment _?

11:55:10 <bengee> another (potentially silly) n-triples question: when do I add the @[lang_code] to a literal? only if it's _not_ typed?

12:01:09 <bengee> that's how I understand the data model section in the syntax spec, although I thought it would make sense to have a "something"^^<...#string>@en

12:05:00 <bengee> and related: if an untyped literal does only have numbers and stuff in it (e.g. "2004-06-09") should I still add the language info, or is this only done when there are alpha chars in the literal value?

12:05:19 <bjoern_h> bjoern_h is now known as bjoern_

12:05:32 <bengee> any help appreciated. (sorry for the noise..)

12:08:35 <shellac> bengee: um - not sure I understand

12:09:08 <shellac> you can have lang and/or datatype

12:10:23 <shellac> "foo"@lang "foo"^^<datatype> "foo"^^<datatype>@lang

12:10:40 <shellac> (hmm - did I get the syntax right?)

12:11:26 <bengee> that's what I assumed. but the spec (http://www.w3.org/TR/rdf-syntax-grammar) says (section 6.1.8 and 6.1.9) plain literals can have a language, typed literals only a datatype.

12:11:43 <bengee> if I understand the spec's syntax right..

12:13:53 <shellac> oh, I see what you mean

12:14:44 <bengee> I _think_ the w3c validator doesn't care... will have a look

12:16:29 <bengee> ah no, it _does_ care

12:16:44 <shellac> what - it objects?

12:17:44 <bengee> it does not show language info when there is a datatype.

12:18:09 <shellac> well you learn something new every day

12:18:36 <shellac> section 6.5 is a bit clearer on this

12:19:40 <shellac> but it appears to say that you can't give a literal typed as a string an language, which is odd

12:20:17 <bengee> that' s why I started wondering..

12:20:30 <bengee> but the spec is the spec.

12:21:39 <bengee> so I think, IF there is some language info AND the literal is not typed: add the @.

12:21:57 <bengee> (only if)

12:23:03 <shellac> I'm trying to think of non-trivial cases where you want both

12:23:16 <bengee> and obviously no matter what the literal value is (e.g. can be numbers).

12:24:42 <bengee> hm, the datatype kind of "overwrites" the language info, so toolmakers have to make sure to remove the datatype attr if the language info is to be preserved

12:25:09 <shellac> I'd point out "The language tag relates only to human language text" - so I would say it is legal, but redundant, for eg dates

12:29:13 <bengee> yeah. in a framework one could additionally look at the vocabulary and check the range before generating n-triples..

12:30:40 <bengee> thanks for the help, shellac. things are more clear to me now :)

13:02:54 <mortenf> mortenf is now known as mortenfMAD

13:16:46 <[GNU]> hello all

13:21:47 <DanC> hmm... the grddl abstract could use elaboration.

13:41:47 <mortenfMAD>http://www.wasab.dk/morten/2004/06/photos/madrid/2/

13:41:49 <dc_rdfig> A: http://www.wasab.dk/morten/2004/06/photos/madrid/2/ from mortenfMAD

13:42:04 <mortenfMAD> A:|Image Description Workshop - Whiteboard

13:42:04 <dc_rdfig> Titled item A.

13:42:09 <mortenfMAD> A:Notes by danbri

13:42:10 <dc_rdfig> Added comment A1.

13:42:24 <mortenfMAD> A:From day 2, tuesday

13:42:24 <dc_rdfig> Added comment A2.

13:43:04 <mortenfMAD> A1:Notes by danbri and chaals

13:43:04 <dc_rdfig> Replaced comment A1.

13:48:44 <DanC> ok, http://www.w3.org/2004/01/rdxh/spec 1.49 has beefed-up abstract. feedback, anyone?

13:51:33 <dajobe> hmm

13:51:47 <dajobe> [vocab ... is fixed, or slowly evolving, ...]

13:52:02 <dajobe> I find that sentence rather long and saying too many things

13:52:27 <DanC> hmm... yeah... the XHTML bit is sort of a diversion

13:52:37 <dajobe> starts off, xhtml is a dialect [of xhtml] and at the end, there are lots of dialects of xhtml itself

13:53:17 <DanC> I want to start with "HTML is ubiquitous" and then talk about XHTML.

13:54:53 <dajobe> sentence 2 looks like it was going to say while [XHTML is this thing.... BUT..] however it never gets to the BUT

13:55:18 * DanC wordsmiths...

13:57:57 <DanC> try 1.50

13:58:25 <DanC> "for used to" oops

13:59:13 * DanC saves 1.51 with s/for used/used/

14:01:26 <dajobe> better

14:01:40 <DanC> hmm... what motivated me to dink with the abstract was the idea that there's lots of semantic-web-ish data that's not in RDF, and GRDDL can bridge the gap. I'm not sure I actually said that, though

14:02:20 <ndw> DanC, there are encoding problems in Dom's name

14:02:46 <DanC> ppphphtphtphtphtphphphtphtpht

14:02:53 * ndw shrugs

14:04:13 <DanC> I have to demangle names about 3 times a day.

14:04:35 <DanC> I stick to us-ascii and &#numeric-character-references; but there's some tool that mangles them

14:06:33 * DanC demangles it. 1.52

14:06:49 <dajobe> there was something got me, apache's default content-encoding (iso8859-1) vs the documents. I forget the detail.

17:29:32 <bblfish> any comments on GRDDL versus IBM's Semantically-Adapted Service Data Objects

17:29:39 <bblfish> ?

17:30:28 <sbp> .g Semantically-Adapted Service Data Objects

17:30:29 <phenny> Semantically-Adapted Service Data Objects: http://dannyayers.com/archives/2004/04/09/ibm-emerging-technologies-toolkit-semantic-sdo/

17:31:02 <sbp> I don't see the relevance

17:31:33 <bblfish> sbp: I thought both were ways of converting XML into RDF

17:32:53 <sbp> dunno. attaching RDF/OWL to an "SDO DataGraph" isn't all that clear

17:33:42 <bblfish> It's not that clear what the licence of IBM's stuff is, and it seems like a huge download

17:34:30 <bblfish> phenny: thanks for the link, I had found a few others but yours is a bit longer than the others I found

17:34:39 <sbp> she's a bot

17:34:54 <phenny> Thanks anyway, though!

17:35:27 <bblfish> ? :-|

17:35:29 <sbp> ohh, you're Henry Story

17:35:38 <sbp> heh. one day I'll remember your nickname -> name mapping

17:35:41 <sbp> phenny: help?

17:35:41 <phenny> Hi! I'm phenny, datum's replacement. My commands include:

17:35:43 <phenny> cp, ety, g, mw, purl, swhack, t, tell, time?, val, w, xena

17:35:44 <phenny> For help: 'phenny: help command?'.

17:35:45 <phenny> My owner is sbp. Thanks.

17:35:49 <bblfish> yes

17:35:52 <sbp> phenny: help g

17:35:52 <phenny> sbp: .g <phrase> - Google for <phrase> and return the first result.

17:36:53 <bblfish> sigh :-/

17:37:00 <bblfish> :-)

17:38:22 <sbp> I talk to her too sometimes. wouldn't worry about it

17:38:30 <sbp> phenny: tell bblfish it's no problem

17:38:30 <phenny> Okay, I'll tell bblfish that when they're around.

17:38:43 <bblfish> heh

17:38:43 <phenny> bblfish: sbp told me to tell you: it's no problem (at 2004-06-09 17:38:30.683266)

17:39:41 * ndw chuckles

17:39:44 <ndw> phenny, help cp

17:39:44 <phenny> ndw: .cp <regexp> - Search for a particular Unicode codepoint.

17:39:52 <ndw> .cp bullseye

17:39:53 <phenny> 0298: LATIN LETTER BILABIAL CLICK

17:39:54 <phenny> 25CE: BULLSEYE

17:39:59 <ndw> sweet!

17:40:08 <ndw> But how's 0298 relevant?

17:40:19 * bblfish gota go for dinner and a drink of strong wiskey

17:40:38 * ndw mmmm's "whiskey"

17:41:48 <ndw> phenny, help mw

17:41:48 <phenny> ndw: .mw <word> - Search for the definition of word on m-w.

17:42:20 <ndw> .mw achieved

17:42:21 <phenny> Entry: achieve /&-'chEv/, verb

17:42:22 <phenny> Etymology: Middle English acheven, from Middle French achever to finish, from a- (from Latin ad-) + chief end, head -- more at CHIEF

17:42:23 <phenny> Entry: transitive senses1 : to carry out successfully : ACCOMPLISH <achieve a gradual increase in production>2 : to get or attain as the result of exertion : REACH <achieved a high degree [...]

17:42:38 <ndw> phenny, help swack

17:42:38 <phenny> Sorry, I don't recognize 'swack'

17:42:48 <ndw> phenny, help swhack

17:42:48 <phenny> ndw: .swhack <phrase> - Search for <phrase> in the Swhack logs.

17:42:55 <ndw> heh

17:43:00 <sbp> $ grep -i bullseye UnicodeData.txt | head

17:43:00 <sbp> 0298;LATIN LETTER BILABIAL CLICK;Ll;0;L;;;;;N;LATIN LETTER BULLSEYE;;;;

17:43:11 <ndw> heh

17:43:15 <ndw> thanks, sbp

17:43:19 <sbp> no problem

17:43:45 <sbp> val and purl are probably the best functions

17:43:57 <sbp> .val http://norman.walsh.name/

17:43:59 <phenny> http://norman.walsh.name/ is Valid

17:44:06 <sbp> .val http://nwalsh.com/

17:44:07 <phenny> http://nwalsh.com/ is Valid

17:44:09 <dc_rdfig> B: http://norman.walsh.name/ from phenny

17:44:10 <dc_rdfig> C: http://nwalsh.com/ from phenny

17:44:15 <sbp> whoops

17:44:16 <ndw> lol!

17:44:37 <sbp> I fixed that bug on the server but haven't restarted her yet

17:44:48 <ndw> phenny, help purl

17:44:48 <phenny> ndw: .purl <name> <URI> - Create a short link to the given URI.

17:44:48 <sbp> B:|Norm's Blog

17:44:51 <dc_rdfig> Titled item B.

17:44:59 <sbp> C:|Norm's Homepage (it's Valid!)

17:45:01 <dc_rdfig> Titled item C.

17:45:16 <sbp> .purl ndwblog http://norman.walsh.name/

17:45:17 <phenny> Created http://purl.org/net/swhack-ndwblog

17:45:29 <LotR> ndw: can I go back to #docbook yet?

17:45:35 <ndw> oh, yes, sorry.

17:45:37 <sbp> thanks to xover for help with .val and jcowan for help with .purl

17:45:46 <ndw> LotR: you were idle for like 142 hours, i figured you were on vacation

17:45:59 <ndw> The #docbook channel is now registered on freenode. And it has a bot too :-)

17:47:13 <LotR> ndw: no, the computer in my room doesn't boot, so I only peek in (very) occasionally from the gateway box that is in the hallway here for now

17:47:31 <ndw> hah

17:47:50 <ndw> To quote the final message, "we apologize for the inconvenience"

17:48:16 <ndw> hmm. 0298 is really a much better bullseye than 25ce (in my font in firefox anyway)

17:50:18 <sbp> but don't send it to anyone that you wouldn't kiss in real life

17:50:41 <sbp> (U+0298 being a kiss sound, roughly)

17:51:18 <ndw> rotfl! "bilabial click" ROTFL

17:51:26 <ndw> That's too funny

17:53:30 <sbp> Unicode is full of Freudianisms. you just need to know where to look :-)

17:56:31 <ndw> That one's getting blogged as the funniest thing to cross my desk in some time

17:57:36 * sbp does the "made n.w.n for the second time in as many years!" dance

18:03:40 * DanC wanders by... "n.w.n"?

18:03:51 <DanC> ah... norman.walsh.name

18:04:19 <ndw> got it in one

18:12:29 * DanC mulls over http://norman.walsh.name/2004/06/04/dvr ...

18:12:33 <DanC> "but really, just how convenient do we have to make television anyway? Watching less of the stuff?d probably be a better choice." well said.

18:12:53 * DanC wants a kitchenputer like kbob's

18:13:01 <DanC> .google kitchenputer

18:13:21 <DanC> hey, what's with the service around here? it's gone to hell.

18:13:29 <DanC> ;-)

18:13:49 <DanC> . http://www.euglug.org/blogs/kbob.php?itemid=68

18:16:23 <DanC> more recent: http://www.euglug.org/blogs/kbob.php?itemid=104&catid=5

18:20:06 <ndw> .g kitchenputer

18:20:07 <phenny> kitchenputer: http://www.euglug.org/blogs/kbob.php?itemid=68

18:20:35 <ndw> Phenny has a slightly different command vocabulary, I guess

18:26:44 <DanC> skberp! 404 @ http://www.advogato.org/person/connolly/

18:27:03 <sbp> phenny coexists with xena in some channels, hence the evasion of a command intersection

18:27:16 <DanC> ah... "Advogato is temporarily down" -- http://www.advogato.org/

18:29:59 <ndw> .g skberp

18:29:59 <phenny> skberp: sorry, no results were found.

18:30:11 <ndw> ok, it's not just me then :-)

18:31:13 <sbp> skberp, /sk&-burp'/, intj. - 1) Polite Dancian for "oh crap, one of my homepages is down".

18:34:11 <ndw> lol

18:35:31 <DanC> heh

18:47:34 * DanC wonders if sbp has seen http://www.w3.org/2000/10/swap/admin/bugStatus

19:06:12 <sbp> yeah, I think I commented on it a while ago

19:06:19 <sbp> it's a good system. seems there are quite a few resolutions omitted though: is that because TimBL/you haven't got around to them yet, or because you can't hook up the resolution messages to the bug raisings?

19:13:32 <DanC> prolly because we haven't gotten around to them yet. example?

19:14:03 * DanC mentions sbp by name in case that reduces latency

19:18:57 <ndw> .t

19:18:57 <phenny> Wed, 09 Jun 2004 19:18:58 GMT

19:54:01 <DanC> my brother(!) writes: "It would be convenient to have a utility that would read the xml file into a DOM - show the XML - let me pick a node, let me enter some XPATH and it would show me what it selects."

19:54:12 <DanC> any clues? ndw, seen anything like that?

19:57:33 <ndw> sounds familiar, but no specific answers spring to mind

19:57:56 <ndw> maybe one of the XSLT debugging kits, Paul Prescod was working on one wasn't he? While still at ActiveState?

19:58:00 <ndw> There must be others.

19:58:21 <mattmcc> Would something like http://www.zvon.org:9001/saxon/cgi-bin/XLab/XML/extras.html?stylesheetFile=XSLT/xpath.xslt&lang=eng be what he wants?

19:58:24 <ndw> sorry I can't be more directly helpful

19:58:57 * DanC looks at http://www.activestate.com/Products/Visual_XSLT/ ...

19:59:24 <DanC> "XPath Workshop

19:59:24 <DanC> Build and test XPath expressions"

19:59:55 <DanC> ooh... mattmcc... wicked!

20:00:15 <ndw> Evening Libby

20:00:44 <DanC> logger, pointer?

20:00:44 <DanC> See http://www.ilrt.bris.ac.uk/discovery/chatlogs/rdfig/2004-06-09#T20-00-44

20:01:32 <mattmcc> Hmm. http://www.opikanoba.org/lantern/ also looks useful.

20:02:38 <libby> heya norm

20:03:26 * libby looking for mortenf's photos from last night, guess they're not up yet

20:04:56 <eaon> eaon is now known as eaon|away

20:06:33 <libby> guess I could put mine up but they aren't that good

20:06:33 <DanC> man, mattmcc, where do you find these things?

20:06:50 <ndw> Off topic, what's the IRC bot development platform of choice? I tried POE in Perl and it seems to just stop responding after a while.

20:06:53 <mattmcc> DanC: The zvon one was from a vague memory of it existing back when I read their xpath tutorial.. The second from freshmeat.

20:07:30 <DanC> I tried google... thought about freshmeat, but you beat me to it. IRC rocks. :)

20:08:03 <libby> pirc in java is nice and easy to use

20:09:02 * DanC hopes debian and java will get along better soon

20:10:20 <ndw> I didn't have any trouble just grabbing the JDK and installing it DanC

20:10:28 <ndw> .g pirc

20:10:28 <phenny> pirc: http://www.pirc.co.uk/

20:11:13 <ndw> nope. that's not it. chuckle

20:12:19 <ndw> For log readers, it's www.jibble.org

20:15:07 <libby> what bot are you writing ndw?

20:15:08 <ndw> Yeah, I should use Java. Thanks, Libby

20:15:37 <libby> jibbler wrote pirc - he's here btw

20:18:30 <Jibbler> .g pircbot

20:18:30 <phenny> pircbot: http://www.jibble.org/pircbot.php

20:18:30 <libby> ello :)

20:18:30 <Jibbler> pirc != pircbot :)

20:18:30 <libby> yeah sorry

20:18:30 <Jibbler> i should have chosen a more sensible name :)

20:18:31 <Jibbler> pircbot = paul's irc bot </sad>

20:18:31 <libby> heheh

20:18:33 <ndw> A little helper bot for #docbook; I'll start up the half-baked one again if you want to see

20:18:33 <ndw> Eventually it'll read the DocBook RNG and work "automatically" from that. At least that's the plan. In my spare time, you know :-)

20:18:40 <libby> cool :)

20:18:44 * Jibbler forgets what spare time is these days

20:18:59 <ndw> It's the other, other 24 hours in the day

20:19:02 <libby> I love bots though I find the syntax hard to design

20:19:19 <DanC> spare time is when you're only double-booked.

20:19:27 <ndw> bingo!

20:20:01 <DanC> or: spare time is when you're doing stuff you're not getting paid for.

20:20:08 <ndw> I was thinking of investigating the natural language parsers from the infocom games for that :-)

20:20:21 <ndw> I think there's a Java Z-code interpreter

20:20:35 <ndw> But I'll probably just tokenize and fake it :-)

20:24:13 <Jibbler> i just got the PDFs of IRC Hacks today, so i've gotta read through the whole lot in the next few days. i think i'm going to be dreaming about bots tonight :)

20:24:41 <ndw> PDFs?

20:24:52 <Jibbler> pdfs... adobe things

20:25:06 <ndw> Ah, PDF files of IRC Hacks, a forthcoming (O'Reilly?) book?

20:25:10 <Jibbler> yarr

20:25:13 <ndw> kewl

20:25:17 <Jibbler> gotta check it's all okay before they print it

20:25:24 <ndw> that's one option :-)

20:25:32 <Jibbler> looks like it's going to be about 420 pages or so

20:25:53 <ndw> I'll plonk my $39.95 down for that :-)

20:28:04 <Jibbler> it'll probably be about $17-18 on amazon

20:47:27 * sanctius vous souhaite une très bonne nuit à tous et plein de beaux rêves dans les bras de Morphée ???

20:48:05 * sanctius is back (gone 09:09:41)

20:48:18 * sanctius is away: ?_ absent pour le moment _?

20:51:16 <eaon|away> eaon|away is now known as eaon

21:04:35 <libby> A:[http://swordfish.rdfweb.org/photos/2004/06/08/|libby's photos from the same day]

21:04:36 <dc_rdfig> Added comment A3.

21:37:31 * mortenf can't get to his server, so no tuesday pics yet :(

22:51:01 * bengee finally got his rdf parser running. He's going to have nightmares full of monster predicates screaming 'parseType="Collection"' at him 2night...

22:51:27 <drumm> bengee: what language?

22:52:07 <bengee> php

22:52:11 * bengee ducks

22:52:30 <drumm> rdf goes in, what comes out?

22:53:02 <bengee> array of triples with triples being arrays, too

22:53:24 <drumm> how fast is what happens in between?

22:55:03 <bengee> hmm, depends on structure of rdf/xml. varys betwenn 2.5 to 4 times faster than RAP, but I guess the comparison is not fair (RAP offering more functionality etc)..

22:56:05 <drumm> so you have to do the rdf to arrays, any other data cleaning, verification, enhancement, or anything extra in there?

22:58:07 <bengee> not much. no validation or anything, it can serialize to n-triples with proper datatypes and language info, but that's basically it.

22:58:31 <drumm> do you use php's built in xml parser or some glob of regexp?

22:58:44 <bengee> main advantage is having a single-file standalone parser..

22:58:57 <bengee> am using php's SAX

22:59:01 <drumm> good

22:59:17 <drumm> okay, how is it licensed and is it GPL-compatible?

23:00:33 <bengee> not sure yet. something along "free for non-commercial use, derivative work: OK"

23:01:39 <drumm> so I would want to get it into Drupal (drupal.org), maybe even Drupal core, but as a module processing FOAF for now

23:02:11 <drumm> as the demand for RDF parsing rises then a parser such as yours would need to be in the API and, that is definately GPL licensed

23:02:35 <drumm> and Drupal is used by a lot of consultants on commercial sites

23:03:26 <bengee> I see..

23:05:21 <bengee> but those consultants don't resell Drupal installations, do they?

23:05:45 <drumm> nope, they sell their services for setting it up

23:06:37 <drumm> I am working on a distribution of Drupal (civicspacelabs.org), which we will have a hosting company or three host typepad style

23:07:47 <bengee> oh, deanspace.

23:07:50 <drumm> it is definately used for commercial purposes, but the code is free being open source and gpl

23:08:06 <drumm> yep, it is deanspace

23:08:22 <bengee> that was a cool campaign.

23:09:33 <drumm> yeah, then it blew up

23:09:46 <bengee> ;)

23:11:02 <bengee> at least he couldn't blame the IT for that..

23:12:18 <drumm> well, we actually did have some sucky IT where it was important

23:13:13 <drumm> Iowa took the bid which was half the price of all the others for their voterfile system, the big db that lets you keep track of people who might vote for you

23:13:43 <drumm> and we got what we paid for. that thing was quite flaky

23:19:05 <bengee> so the gpl says that any code you add to the system must be gpl-licensed as well?

23:20:49 <drumm> yep

23:21:54 <drumm> joel de gan (I think) made a nice xml parser (xml->nested arrays via php sax)

23:22:05 <drumm> he made it public domain

23:22:29 <drumm> which is by far the easiest to get your code out there if you don't want monetary profit

23:27:04 <bengee> the tools I'm writing are going to be part of a commercial product, but I'd like to make them available to the community/bloggers/coders etc as well.

23:28:20 <bengee> have to spend some thoughts on the licensing stuff. should anyway run some rdf test cases first ;)

23:28:51 <bengee> g'night RDFers

23:30:19 <drumm> hrm, I have no way to contact him again

23:42:42 <sh1mmer> evenin

23:47:18 <ndw> Is there an ontology for XML constructs (element, attribute, PI, etc.)?

23:57:08 <sh1mmer> ndw could you maybe convert the dtd into one?

23:57:39 <ndw> Not a particular element in a schema; the concept of XML element

23:58:06 <sh1mmer> I know what you mean

23:58:32 <ndw> I just lied: http://www.w3.org/TR/REC-xml/element

23:58:36 <sh1mmer> Is there not a formal specification of xml in a dtd though (maybe I have missed something)

23:58:51 <sh1mmer> :)

23:59:18 <drumm> wow that sounds like way too many layers of xml describing xml