Semantic Web Interest Group IRC Chat Logs for 2002-03-16

00:20:02 <dmiles> back

02:11:28 <dmiles> dmiles is now known as infWERXTE

05:36:39 <DanC> ah... now maybe I can finish the thought that I was in the middle of when end-of-office-time came earlier today...

05:38:26 * DanC finds two newly-installed spam filters working. whew!

08:33:31 <theran> huh?

11:08:09 * danbri waves, checks out the new SOAP spec work

11:10:43 <danbri>http://lists.w3.org/Archives/Public/xml-dist-app/2002Mar/0115.html

11:10:44 <dc_rdfig> A: http://lists.w3.org/Archives/Public/xml-dist-app/2002Mar/0115.html from danbri

11:12:03 <danbri> A:|Rework on SOAP 1.2 Part 2: Section 2 and 3, Martin Gudgin and and Noah Mendelsohn work in progress within XMLP WG

11:12:03 <dc_rdfig> titled item A

11:12:40 <danbri> A:Please note "at this stage the rewritten sections DO NOT represent consensus from the XML Protocl Working Group. Rather they are work in progress on out way to consensus."

11:12:40 <dc_rdfig> added comment A1

11:13:55 <danbri> A:In particular, do take a look at the reworked [http://www.w3.org/2000/xp/Group/1/10/11/soap12-part2-mjg.html#datamodel|SOAP Data Model] section. "The SOAP data model represents information as a directed, edge-labeled graph of nodes."

11:13:55 <dc_rdfig> added comment A2

12:40:09 * JibberJim wonders if danbri has seen http://jibbering.com/rdf/foafwho.asp?person=mailto:danbri@w3.org

12:47:10 * danbri downloads isaviz for 1st time, tries to run it

12:47:11 <danbri> ./run.sh: line 3: 19375 Segmentation fault

12:47:22 * danbri waits for next version or until he upgrades Java

12:47:24 <danbri> Hi Jim!

12:47:34 <danbri> Yes, I caught a glimpse of it last night! Very cool :)

12:48:08 <danbri> (except that terrible photo of me at Sean's flat... I was really ill that day, in bed with fever the next...)

12:48:38 <danbri> I think there was some mention of RDF schema vocab needed for representing SVG fragments... did you see the same instance data I sketched?

12:55:15 <danbri> Jim, would it make sense to have http://jibbering.com/svg/AnnotateImage.html be bookmarklet-able? ie. http://jibbering.com/svg/AnnotateImage.svg might pick up the HTTP referer, or ?img=some-uri from URL...

12:56:59 <danbri> hmmm, can't fill in text fields in http://jibbering.com/svg/AnnotateImage.svg (moz .99, adobe 3)

13:02:36 <danbri>http://www.w3.org/2001/08/rdfweb/foafwho/svg-in-rdf.rdf

13:02:36 <dc_rdfig> B: http://www.w3.org/2001/08/rdfweb/foafwho/svg-in-rdf.rdf from danbri

13:02:50 <danbri> B:|Dan's SVG-inside-RDF example

13:02:50 <dc_rdfig> titled item B

13:04:09 <danbri> B:The is what [http://www.w3.org/2001/08/rdfweb/foafwho/imagemap/imagemap2meta.rb|imagemap2meta.rb] was going to use / export. Jim's [http://jibbering.com/2002/3/nadiaeric.rdf|SVG imagempap editor tools] export something similar.

13:04:09 <dc_rdfig> added comment B1

13:06:43 <danbri> B:So design question: we're both putting bits of an SVG path / polygon inside RDF. How much of the SVG should we reflect into the RDF graph? Does the path stuff need to be properties of a (jpeg etc) image, or an 'SVG' object?

13:06:43 <dc_rdfig> added comment B2

13:07:46 <danbri> B:Test case: if someone can create an SVG path against a small JPEG and the Batik rasterizer tool plus some RDF image metadata can be used to rasterize from a full detail JPEG (or TGA etc) image, we know we're getting this right-ish.

13:07:46 <dc_rdfig> added comment B3

13:08:42 <danbri> B:This also suggests that the various thumbnailizer scripts we all use for our photo collections could usefully leave detailed logs of the image transformations.

13:08:43 <dc_rdfig> added comment B4

13:19:43 <danbri> A:I just followed up [http://lists.w3.org/Archives/Public/xml-dist-app/2002Mar/0145.html|suggesting ntriples rather than ascii art] worked well for RDF syntax discussions, maybe same for SOAP-Enc too...

13:19:43 <dc_rdfig> added comment A3

13:57:40 <rreck> I cant seem to get the W3C RDF Validator to work for me, can someone tell me an URL and settings that work ?

13:59:03 <rreck> JHendler showed me it parse his homepage but i cant seem to reproduce it

14:12:39 <DanC> the validator is linked from http://www.w3.org/RDF/

14:15:34 <JibberJim> I hadn't seen the SVG-inside-RDF example... or I probably would

14:15:49 <JibberJim> not have rolled my own...

14:18:06 <JibberJim> I definately think the paths are properties of the jpeg at least in my model, they're just paths in the jpeg co-ordinate space which use SVG syntax. even in my example I process that path before outputting the svg.

14:23:38 <JibberJim> If you click in the text boxes of AnnotateImage.svg - can you not type into them? the main point of going to textboxes in svg was for Mozilla...

14:24:10 <danbri> re jpeg/etc, why should it be relative to jpeg? what if a PNG is content-negotiable at the same URL? (actually the rdfweb foaf stuff is vague on that too).

14:24:29 <danbri> we should use the Chump for this (since scanning the logs is increasingly time consuming for folk)

14:24:42 <JibberJim> does foaf:Image refer to a url that may be different sizes then?

14:24:46 <danbri> B:Jim replied " I definately think the paths are properties of the jpeg at least in my model, they're just paths in the jpeg co-ordinate space which use SVG syntax. even in my example I process that path before outputting the svg.

14:24:46 <dc_rdfig> added comment B5

14:26:37 <danbri> B:danbri: "re jpeg/etc, why should it be relative to jpeg? what if a PNG is content-negotiable at the same URL? (actually the rdfweb foaf stuff is vague on that too).

14:26:37 <dc_rdfig> added comment B6

14:26:55 <DanC> danbri, sometimes a pointer into the logs is easier

14:27:10 <danbri> good point. I forget how to do this!

14:27:14 <danbri> logger, pointer?

14:27:18 <danbri> See http://ilrt.org/discovery/chatlogs/rdfig/2002-03-16#T14-27-14

14:30:00 <JibberJim> I think your i:svg object is definately wrong - an SVG being vector shouldn't be constrained to 640 490, it's the Image that has the size.

14:30:48 * DanC is back (gone 110:22:14)

14:31:21 <JibberJim> other than that, we just use different names don't we?

14:32:51 <danbri> Crap, xchat just lost content :(

14:32:58 <danbri> I hit up arrow and it lost my half-finished comment.

14:33:15 <danbri> B:See [http://ilrt.org/discovery/chatlogs/rdfig/2002-03-16#T14-27-14|chat logs| for detail.

14:33:15 <dc_rdfig> added comment B7

14:35:15 <JibberJim> Mine would be very difficult (impossible?) to process with xslt without scripting.

14:36:42 <danbri> B:Here's a counterusecase, in favour of distinguishing the svg path overlays from any paricular image. Imagine a cityscape webcam: "this region is bristol docks, this is cabot tower, that's park street, that's some big old oak tree", with a 365*24 actual images per year. So paths best not modeled as being of some specific jpeg/photo; they can be more general.

14:36:42 <dc_rdfig> added comment B8

14:39:11 <JibberJim> Surely they are overlays of a resource though, this time the resource is not fixed, <webcam> jim:hasPart etc. just as a jpeg hasPart etc.

14:39:31 * DanC left cwm running last night doing the RDFS schema-closure of cyc.daml; it didn't finish. I'm watching it at --chatty=30 now

14:41:43 <JibberJim> Yes, you can have SVG representations, but then the Image/webcam etc. would be another property of the SVG.

14:53:25 <danbri> re foaf:Image, we've not used it ever for identifying images with different dimensions, but I could imagine using it for a content negotiable resource with PNG/GIF/JPEG 'views'

14:54:35 <JibberJim> I've needed to add height/width properties to it anyway which even content-negiotiated would need to be fixed (simply 'cos I can't get SVG to show an image if it doesn't know the size.

15:00:57 <DanC> cwm is plugging away at cyc...

15:00:58 <DanC> Concluding definitively C2->TangibleProduct C->HandTool _g8->0_work

15:00:58 <DanC> Size of store changed from 26753 to 26754

15:00:58 <DanC> Concluding definitively C2->TangibleProduct C->FoodVessel _g8->0_work

15:00:58 <DanC> Size of store changed from 26754 to 26755

15:01:07 <DanC> this is perhaps a worst-case scenario for forward-chaining.

15:01:26 <DanC> I wonder how well Euler would do with these data. Jos, are you listening/reading-the-logs?

15:01:55 <DanC> oops... killed it; the command-line was:

15:01:57 <DanC> python ~/w3ccvs/WWW/2000/10/swap/cwm.py --chatty=30 --rdf http://opencyc.sourceforge.net/daml/cyc --n3 ~/w3ccvs/WWW/2001/05ve/rdfs.n3 --think

15:02:51 <danbri>http://www.cikon.de/Text_EN/SWKS.htm

15:02:51 <dc_rdfig> C: http://www.cikon.de/Text_EN/SWKS.htm from danbri

15:03:07 <danbri> C:|The Semantic Web - a Web of Knowledge and Services, Hans-Georg Stork, Bollendorf

15:03:07 <dc_rdfig> titled item C

15:06:36 <danbri> <JibberJim> I think your i:svg object is definately wrong - an SVG being vector shouldn't be constrained to 640 490, it's the Image that has the size.

15:07:00 * danbri suspects you're right. The SVG doesn't have size as such (but maybe a coordiantion system? I'm mostly bluffing as far as my SVG knowledge goes)

15:08:17 <JibberJim> It's only when using raster images that you should use pixels as the co-ordinate system though, like in css pixels are bad...

15:09:17 <danbri> I don't think I *said* they were pixels, the dimensions jsut happend to map 1:1 onto the pixel structure of the overlaid image.

15:09:33 <danbri> But I didn't think too hard about it, nor do anything yet with the resulting RDF!

15:09:34 <JibberJim> That's true...

15:11:33 * danbri tries Jim's SVG again

15:13:07 <JibberJim> As to prefilling it with a referer or something, that's a good idea, I'll look to do that.

15:13:47 <danbri> (nitpic), some minor unfairness in http://jibbering.com/rdf/foafwho.html re "I found this lacking as I idn't know who was who", since in http://www.w3.org/2001/08/rdfweb/ stuff we admit it lacking and try using Amaya to address this...

15:13:56 * danbri prefers the pure SVG implementation though :)

15:14:36 <danbri> With something like this (your SVG doo-dad), I could imagine getting multi-1000s of image descriptions over the next year! :)

15:14:53 <danbri> (training that many members of public to use Amaya isn't going to happen, sadly)

15:15:18 <JibberJim> I can get svg annotations posted to annotea to and coming back with the right namespace (by working around the Annotea bug.)

15:15:56 <danbri> Neat. Which bug is this? the whitespace in content-type bug?

15:16:07 * danbri wonders about bugzilla for swad stuff...

15:16:19 <JibberJim> yes.

15:16:31 <JibberJim> We also need to get rdfweb to store and generate the same as my foafweb, it's not going to scale as it re-parses the entire RDF each time.

15:17:03 <danbri> what content type should it be returning? I'll ask Eric to update the live site...

15:17:58 <JibberJim> He's not fixed it and they're scheduled for a release of the new code on the 22nd, so probably best to leave it 'til then as it's only a week.

15:18:35 <danbri> 'k

15:19:37 <danbri> Do you have a list of RDF files RDFWeb should harvest and index? It (or my version, I think Libby's too) follows rdfs:seeAlso pointers between documents, so you could add a URL to a table of contents doc on your site, and it'd get re-indexed periodicially.

15:20:40 <JibberJim> Er. no, the svgfoafwebs are all in one big http://jibbering.com/2002/3/nadiaericsbpdanbri.rdf I'm not that good at organising what I do, but I'm trying to get better.

15:20:50 <danbri> trying http://jibbering.com/svg/AnnotateImage.svg again: no joy. no sign of the text fields being able to accept focus.

15:21:26 <danbri> I tried 'show SVG' and it showed me text of a basic SVG doc (though not clear how to flip back to other mode).

15:22:23 <JibberJim> Does the textboxes at http://www.kevlindev.com/gui/textbox/textbox.svg work?

15:22:48 <JibberJim> (that's where I borrowed them from...)

15:23:55 <danbri> No, they're inactive too.

15:24:35 <JibberJim> works on win32 Mozilla 0.9.8

15:24:53 <danbri> bummer :(

15:26:08 <JibberJim> I was going to use them aswell as part of my attempt to add decent non win-32 support to www.showcaster.com I'll head back to the drawing board.

15:27:56 * danbri visits showcaster

15:28:11 <danbri> heh, Operating System not supported. see what you mean!

15:28:50 <JibberJim> It's also not true, there is a beta flash version, and you can listen to the mp3 of presentations...

15:29:40 <danbri> So for the SVG / imagemap editor... how would we do the following: associate an RDF class with the image region. For example, the beer in my picture would a depeiction of an instance of the class http://xmlns.com/wordnet/1.6/Beer

15:30:02 <danbri> ...I'd like to know what I could do server side to make it easier to add wordnet cataloguing into the SVG frontend.

15:30:32 <danbri> Could do it all post-imagemap, with just HTML and maybe server-side Batik?

15:31:01 * danbri thinking out loud, that's all...

15:31:09 * danbri takes a break for a while

15:31:44 <JibberJim> The annotateImage.svg would do it, if you could enter the boxes, instead of it currently just creating SVG with regions, it could create RDF with the wordnet class coming in.

15:32:25 <JibberJim> I _think_ I can port my RDF parser to SVG which means I could then have it importing schema like in the editor which may make it easier.

15:33:20 <danbri> port to _svg_? SVG-with-javascript, or does SVG have enough innate scriptability? (I hadn't thought so...)

15:33:37 <JibberJim> SVG has the same core script engine as Mozilla.

15:34:14 <JibberJim> The only thing that is missing in SVG that's needed for my parser is a way of importing the RDF to start with, which I think it's got.

15:34:54 <JibberJim> SVG in Adobes implementation of course.

15:35:46 <danbri> right

15:36:22 <JibberJim> (which also means I would be able to make my editor in just SVG rather than it being embedded as it is now.)

15:36:54 * danbri torn between hacking on this, getting some missed sleep, and working on his ruby project (RDF db of cross-package subclassing using 78 Ruby .tar.gz pacakges I've just downloaded).

15:37:25 <danbri> For that sort of thing, I can live with server-side work, ie. return an SVG with embedded triples from some .asp or CGI

15:38:24 <JibberJim> And presumably the reverse to, for getting the RDF out, have the SVG+triples and generate the RDF?

15:38:54 <danbri> Yes, doing a POST to something server side (which I think you do already to some extent).

15:39:17 <JibberJim> You can only do a GET unfortunately, but yes.

15:39:21 <danbri> Maybe even a PUT...

15:39:25 <danbri> Oh, right. pity.

15:39:58 <danbri> Generate HTML locally with a button that does a POST? GET maybe isn't big enough for the amount of data...

15:40:07 <JibberJim> If Mozilla sorted out the scripting of plugins as they keep threatening, then you could do a put.

15:40:20 <JibberJim> Mozilla won't let you.

15:40:29 <danbri> (the script that receives the GET could generate a form prep'd for the POST, which when submitted would be the operation with sideeffects...)

15:40:39 <danbri> that's a shame

15:40:56 <JibberJim> The GET also has to return to SVG rather than to a new page...

15:41:20 <JibberJim> mozilla is really broken in this area.

15:43:52 <JibberJim> Right, I need to do some shopping... see ya.

15:44:11 <danbri> OK, happy shopping. cu...

16:02:22 <Galahad> Galahad is now known as xena

17:11:08 <danbri> A:[another example|http://www.w3.org/2001/12/rubyrdf/examples/gudge/intro.html] based on xml-dist-app thread with Martin Gudgin

17:11:08 <dc_rdfig> added comment A4

19:47:16 <thc> thc is now known as tomc

20:29:16 <DanC> I can't ping www.mit.edu; can anybody else?

20:32:32 <dajobe> no

21:16:40 <DanC> sandro, can you ping www.mit.edu?

21:17:06 <DanC> ah... it's back. hmm.

21:17:34 <sandro> Connectivity to MIT has been weird the past hour. Symptoms I can't make sense of. But nothing weird the last ten minutes.

21:22:19 <DanC> do you have any idea what EricP is up to with this stuff? http://www.w3.org/2002/03/11-RDF-XSL/

21:23:11 <sandro> Yes, I think so.....

21:23:43 <sandro> I couldn't explain "= variable assignment" though.

21:24:37 <DanC> can you ping qtwu2.intuit.com?

21:24:41 * DanC is working on taxes

21:25:25 <DanC> so aside from quirks/details, what is he doing?

21:25:36 * sandro should be working on bills/taxes, but is working instead. (surveying triplestores)

21:27:04 <sandro> He's playing with techniques for parsing XML into RDF.

21:27:25 <sandro> ... where XML/RDF is among the harder kinds of XML to parse into RDF.

21:27:43 <dajobe> no sh*t

21:27:44 <sandro> intuit.com is giving me routing loops.

21:27:54 <DanC> er... XML/RDF already is RDF, no?

21:28:48 <sandro> rdf != rdf :-) Perhaps read n-triples for "RDF" above. Pure model stuff. (thus his defining an RDF API.)

21:29:07 <DanC> "pure model"??? n-triples is syntax.

21:29:57 <DanC> the abstract of his page is "This document describes an XSLT stylesheet that transforms application/xml+rdf to a series of RDF database API calls." er... why would one want to do that?

21:30:29 <DanC> if you want an RDF parser in XSLT, I wrote one of those in Apr 2000. it's incomplete, but I don't see why start over.

21:30:40 <dajobe> jeremy carroll wrote a complete one

21:30:43 <dajobe> paper in www2002

21:30:48 <DanC> there.

21:30:56 <DanC> again: 4tf is ericp up to?

21:31:12 <danbri> xslt?

21:31:18 * danbri sends mail to www-rdf-interest to similar effect

21:31:35 <DanC> ericp isn't even writing the XSLT. he's getting somebody else to do that.

21:31:46 <sandro> He's been traveling for I think 4 of the last 5 weeks, Dan.....

21:32:03 <danbri> which imho is willful wasting of someone elses time. At least one of them should've heard of Google... sheesh...

21:32:20 <dajobe> jjc's work: SNail - http://www.hpl.hp.co.uk/people/jjc/snail/

21:32:37 <sandro> His interest in this was, I think, to better understand what one can and cannot nicely parse-to-triples using XSLT. This was at least an educational exercise in that direction.

21:32:45 <DanC> ok, being a little random is understandable. But Sandro, you said you did understand what he's up to.

21:33:21 * danbri recalls http://www.xmlhack.com/read.php?item=757

21:33:28 <danbri> [[

21:33:29 <danbri> Dan Brickley, urged the

21:33:29 <danbri> two authors to bring one of these tools to completion:

21:33:29 <danbri> I'd really like to see one of both of these efforts finished to completion, as part of

21:33:29 <danbri> our attempt to pin down the various issues/problems with the spec. [...]

21:33:29 <danbri> In particular, Bag/Seq/Alt handling seems crucial if the XSLT parsers are ever going

21:33:31 <danbri> to be used in anger.

21:33:33 <danbri> ]]

21:33:40 <danbri> this was sept 2000 for heavensakes

21:33:50 <danbri> Are we doomed to cutesy prototypes for all time!

21:34:12 <danbri> 1 XSLT RDF parser was cool, two were promising. three were crowdy, four is worth making a fuss over.

21:34:16 <dajobe> snail is a complete (caveats) parser using just xslt to ntriples (caveats)

21:34:18 * sandro doesn't program in XSLT because it looks like cutesy prototypes and quick hacks are about what it's good for.

21:34:46 * danbri grins at dave's newfound N-Triple ultracaution ;-)

21:34:47 <DanC> in what way is this stuff an "RDF database API"?

21:34:51 <DanC> predicate = uri("http://example.com/schema#f")

21:34:51 <DanC> object = uri("file://localhost/#abc")

21:34:51 <DanC> addTriple(rType, object, uri("http://example.com/schema#g"))

21:34:51 <DanC> addTriple(predicate, subject, object)

21:34:51 <DanC> subject = object

21:34:52 <DanC> object = NULL

21:34:58 <danbri> It uses xslt extension functions to call some API

21:35:16 <danbri> other alternative is to textualize the triples for consumption elsewhere (eg. ntriples or xml normalisation)

21:35:17 <dajobe> nah, I meant it doesn't quite write the correct ntriples due to xslt->text deficiencies

21:35:57 <DanC> chuckle! "

21:35:57 <DanC>

21:35:57 <DanC> Snail

21:35:57 <DanC> Excruciatingly Slow RDF Parsing

21:35:57 <DanC> "

21:36:14 <dajobe> yeah, he gave a nice ironic presentation of it too - you had to be there

21:37:18 <DanC> [[ Features

21:37:19 <DanC> * Glacial, even with the fastest CPU

21:37:19 <DanC> ]]

21:37:25 * danbri grins

21:37:27 <sandro> In what way is that stuff not an RDF database API, DanC?

21:37:36 <sandro> ... or not using part of one.

21:38:22 <DanC> er... I'd expect to see insert/update/delete in a database API.

21:39:12 * danbri tries to separate insert from update/delete (not least least because I like databases that never forget)

21:39:16 <DanC> or prepare/execute or whatever

21:39:50 <danbri> DanC, can we get W3C T-shirt bounty to bribe people into finishing off some of this stuff?

21:39:57 <DanC> what stuff?

21:40:07 <danbri> At least one XSLT RDF parser, for example...

21:40:26 * danbri knows an upcoming EU project that might have some resources... (as might darker corners of MIT/W3C store cupboards...)

21:40:26 <DanC> jeremy finished his.

21:40:43 <danbri> One that could be used in non-glacial time :)

21:40:54 <DanC> what for? there are plenty of RDF parsers, no?

21:41:00 <danbri> There's an XSLT-to-C convertor I've been meaning to try...

21:41:11 <DanC> there are RDF parsers in C.

21:41:44 <DanC> the purpose of my RDF parser in XSLT was to play with alternative RDF syntaxes. I've since traded XSLT for python and gone with N3.

21:41:45 <sandro> Is there any RDF parser that's less than 6 months behind the WG? ARP, maybe?

21:41:51 <danbri> So for my Ruby library, I'm currently using N-Triples output from Redland. When I/we get Ruby via SWIG glued in, that'll be an option too. I've also got Jason Diamond's parser hooked up via Ruby XSLT.

21:41:57 <danbri> Dave's maybe?

21:42:10 <dajobe> thanks for noticing

21:42:44 <sandro> Dave's has fatal bugs, which I've reported to him and had him say "I've fixed that in an unreleased version, which has worse bugs anyway."

21:42:54 <danbri> what kind of bugs?

21:42:54 <dajobe> ARP's very close to the current state of play

21:43:55 * DanC still doesn't see what's novel about what EricP is doing. there are zillions of RDF parsers with a addTriple() interface.

21:44:16 <danbri> Was it claimed as novel? I think its the XSLT aspect...

21:44:24 <sandro> The one in the released version had to do with naming of bNodes, as I recall. Sometimes they came out genid' or sometihng. It's been a while.

21:44:26 <danbri> ...they didn't do a Google search (or track the mailing list etc)

21:44:39 <danbri> Sandro, do you have an RDF/XML parser?

21:44:46 <dajobe> sandro: that's not too fair, parser has bugs since it isn't #1 priority in my job to work on it

21:45:05 <DanC> well, if he's not claiming it's novel in some way, why is he mailing www-rdf-interest?

21:45:08 <sandro> Sigh.

21:45:13 <danbri> sigh?

21:45:20 <sandro> No, I don't have an RDF/XML parser.

21:45:33 <danbri> Sigh cos you want one?

21:45:40 <danbri> (want to write one?)

21:45:53 <sandro> I use raptor, and I very much appreciate and love it. I just wish Dave would fix this silly little bug. Sorry if I'm coming off wrong about that.

21:46:16 <danbri> use the src, lukesandro... ;-)

21:46:19 * DanC wonders why sandro doesn't fix the bug if it's silly, little, an getting between him and his goals

21:46:22 <danbri> (he says, hypocritically)

21:46:35 <sandro> The RDF/XML parser I write, when/if I write one, is going to just be a blindfold grammar.

21:47:18 <danbri> Is blindfold expressive enough to deal with charsets, namespaces and the full beauty of rdf/xml syntax

21:47:19 <sandro> I have a workaround for the raptor bug -- some line of perl I run on the output that sets things right.

21:47:39 <danbri> The RDF/XML parser I write, when/if I write one, is going to just be a SOAP Encoding handler.

21:47:44 <DanC> what's interesting, to me, about XSLT and RDF is taking RDF with syntactic sugar and reducing it to more vanilla RDF. http://www.w3.org/2001/04rs22/

21:47:46 <dajobe> the bug you reported was on the M&S schema, which works OK now int he current sources

21:48:08 <sandro> That's the $64,000 question, danbri. It should be, but ... what was the line ... it's not my main priority at my job.

21:48:23 <danbri> dave, do we have anon cvs checkout for cvs.ilrt.org? or nightly snapshots for redland etc?

21:48:29 <dajobe> all of that

21:49:37 <DanC> dave, do you have a harness for running your parser thru the WG tests?

21:49:42 <dajobe> yes

21:49:52 <danbri> sandro, I sympathise: getting tools that consume and *demonstrate the value* for using the RDF model is imho more important than yet another parser. I'd like one decent parser per implementation language, with C bindings (presumably to Dave's or repat) and XSLT as fallback for languages not blessed with native support.

21:49:57 <dajobe> DanC: and it is, um, encouraging me to fix things

21:50:34 <DanC> what are the odds your harness could be used with another parser? I think it might be worthwhile to finish off sax2rdf.py (I realized just last night it doesn't grok <propElt propAttr="val"/>)

21:50:37 <danbri> When we can show *Why* folk might want to export RDF views of their data, we'll imho have terrabytes of RDF before we know what to do with it. Right now most can't see the value in exposing RDF.

21:50:49 <danbri> AaronSW had a test harness for a few parsers too, at one point.

21:51:07 <dajobe> my harness is scripts+perl bits, does the job but needs tidying

21:51:24 <dajobe> aarons was for the old rdf parser tests thing i did which generated test descriptions

21:51:29 <DanC> I hope we get weekly/bi-weekly tables of implementations-vs-the-test-suite. We'll need such a table to skip CR.

21:51:47 <dajobe> I've been talking to jang about his ntriples-compare c program and folding/including it into raptor release/sources

21:51:58 <DanC> scripts+perl bits sounds like about what I need; what do you use to compare ntriples files?

21:52:01 <dajobe> tables: sure

21:52:05 <sandro> ugh. I shouldn't be having this discussion on family time.

21:52:10 <dajobe> jang's ntc; hold on

21:52:33 <dajobe>http://ioctl.org/rdf/ntriples/ntc.tar.gz

21:52:34 <dc_rdfig> D: http://ioctl.org/rdf/ntriples/ntc.tar.gz from dajobe

21:52:39 * danbri shouldn't be having this discussion on slacking off time

21:52:49 <dajobe> D:|ntriples comparison utility in C++ - Jan Grant

21:52:49 <dc_rdfig> titled item D

21:53:00 <danbri> so fwiw my gripe on www-rdf-interest, http://lists.w3.org/Archives/Public/www-rdf-interest/2002Mar/0111.html

21:53:01 <DanC> on that we are agreed: we shouldn't be having this conversation. ;-)

21:53:58 <sandro> I have an n-triples comparison tool too, that finds the differences, somewhere.

21:54:58 * DanC wanders off...

21:55:46 <danbri> danbri has changed the topic to: SemanticWeb 24x7 hack'n'chat -- entering completer-finisher phase, please check Google before coding! -- http://rdfig.xmlhack.com/

21:55:59 <dajobe> and my rdf guide too?