Semantic Web Interest Group IRC Chat Logs for 2003-01-06

00:21:20 <dmiles> dmiles is now known as dmiles_afk

00:33:21 <NephLab> NephLab is now known as dmiles

03:41:12 <icepick> icepick is now known as icepick-away

03:41:18 <icepick-away> icepick-away is now known as icepick

06:34:26 <sbp-> sbp- is now known as sbp

09:51:14 <Squizz> Squizz is now known as mattwhe

11:36:45 * danb_lap waves pointlessly to Bijan

11:36:52 <bijan> In real life!

12:54:58 <JibberJim> JibberJim is now known as FredBloggs

12:55:54 <FredBloggs> FredBloggs is now known as JibberJim

13:08:16 <sandro> G: Wow, ambitious depth.....

13:08:17 <dc_rdfig> Added comment G2.

13:08:50 <sandro> G2: Wow, that looks like seriously thorough coverage....

13:08:50 <dc_rdfig> Replaced comment G2.

13:40:09 <icepick> icepick is now known as icepick-away

15:20:30 <_joshua> good morning

15:25:42 <sandro> 'morning

15:34:27 <_joshua> it goes, aparently

15:34:53 <_joshua> pixel: your post of geourl has blossomed into much traffic

16:46:17 <pixel> it looks that way, doesn't it!

16:46:33 <pixel> lots of people in my general vacinity now.

17:01:04 <_joshua> pixel: TheShiftedLibrarien got it from you, and i got a bunch of traffic from her. Then Jon Udell got it from her, and now I'm getting a bunch

17:01:13 <_joshua> Sunday it was on Blogdex, Popdex and Daypop all at the same time, too

17:02:34 <_joshua> Dissapeared today, but i think it'll be back

17:02:35 <_joshua> .

17:03:03 <_joshua> I bet i'd get a big boost if i releasted a geopoint RDFS

18:07:44 * danbri noticed it on daypop too

18:32:54 <deus_x> Anyone here want to field a few rdflib questions?

18:36:50 <eikeon> deus_x: Sure.

18:38:06 <deus_x> Actually, I think I just have one for now: How are the add and remove methods of Resource used? I thought it would be something like: bag.add(LI, item), but that doesn't seem to work. Looking at the source for Resource.py, I'm a bit confused about the self.predicate refs there

18:40:50 <eikeon> Hum... yeah, they should just be predicate not self.predicate. And are just shortcuts for adding triples using the first subject that the resource has.

18:41:25 <eikeon> Or in the case of remove... removes all predicate object pairs from any of the resources subjects.

18:42:32 <eikeon> The Resource object is a fairly new addition of rdflib's interface... and is not as mature as the rest yet.

18:43:12 <deus_x> Ahh okie. I might be a little mixed up then - when would a Resource have multiple subjects?

18:43:40 <deus_x> Oh, and in case you're interested in making fun of my code: http://www.decafbad.com/downloads/archiver_play.py

18:44:51 <eikeon> Most of the testing I have done on it has been to use it "fold" foaf:Person triples into one resource. And have used the __getitem__ __setitem__ a fair bit for properties of cardinality one.

18:50:21 <eikeon> Code looks good to me... if you have not already you may want to grab rdflib-1.2.0 -- the InformationStore has some more code for managing contexts. Things like timestamps of last update, sending If-Modified-Since, etc.

18:51:06 <deus_x> Okay, was hoping that I wasn't doing anything blatantly horrible :) Grabbing 1.2.0 now.

18:51:56 <deus_x> Looking to work this into the personal web proxy I'm building. I want to build up metadata about all the pages I browse, then later do some searching & reporting.

18:52:54 <eikeon> Looking closer at code... not sure about the use of LI

18:53:49 <deus_x> As in bad rdflib usage, or bad RDF in general? (still learning RDF itself)

18:57:02 <eikeon> Hum... how to explain...

18:57:25 <deus_x> Uh oh. :)

18:58:36 <eikeon> LIs get expanded when parsing RDF/XML into rdf:_1 rdf:_2 etc.

18:59:13 <eikeon> This expansion does not happen when adding triples to the store using add. If that is what you where expecting.

19:01:01 <deus_x> Hmm, still getting the hang of the collections. Is there a better way to build up a list of things?

19:01:45 <danbri> eikeon, is that 'folding' you mention the identity-based reasoning, smushing, inverseFunctionalProperty approach?

19:02:11 <eikeon> danbri: Think so.

19:02:29 <danbri> cool :)

19:03:08 <eikeon> This is all I am currently doing to identify a list of subjects:

19:03:09 <eikeon> [[[

19:03:15 <eikeon> for p in self.store.subjects(TYPE, UNAMBIGUOUSPROPERTY):

19:03:15 <eikeon> for o in self.store.objects(self.subjects[0], p):

19:03:15 <eikeon> for s in self.store.subjects(p, o):

19:03:15 <eikeon> if not s in self.subjects:

19:03:15 <eikeon> self.subjects.append(s)

19:03:15 <eikeon> ]]]

19:03:53 <eikeon> Where a resource at least has one subject... a subject it was created using.

19:04:42 <eikeon> deus_x: rdflib does not yet have all the interfaces it should to make dealing with lists easy.

19:05:36 <eikeon> There is a method for iterating over the values in the rdf:first rdf:rest style collections.

19:06:27 <eikeon> It is called items and is defined in rdflib/store/AbstractTripleStore.py

19:07:29 <deus_x> Hmm, I need to look into the first/rest thing. Read about it, but it's slipped my mind.

19:08:40 <eikeon> In some cases you may just want to use a predicate with cardinality greater than one. Unless the order is important.

19:09:34 <deus_x> Hmm, so far the lists I'm making have no important order. Was planning on sorting by things like dates.

19:09:59 <eikeon> For instance, I almost wish RSS 1.0 worked that way... since I typically just sort on the dates.

19:10:02 <eikeon> :)

19:10:25 <deus_x> But if I understand right, I really should use <rdf:_nnn> in the store, instead of <rdf:li> since that's really just a convenience for RDF/XML serialization?

19:10:55 <eikeon> That is my understanding too.

19:11:31 <deus_x> Hmm. Now I just have to figure out how to derive the last nnn and get nnn+1 when I add one :)

19:12:53 <danbri> logger, pointer

19:12:53 <danbri> See http://ilrt.org/discovery/chatlogs/rdfig/2003-01-06#T19-12-53

19:13:00 <danbri>http://ilrt.org/discovery/chatlogs/rdfig/2003-01-06#T19-12-53

19:13:01 <dc_rdfig> A: http://ilrt.org/discovery/chatlogs/rdfig/2003-01-06#T19-12-53 from danbri

19:13:12 <_joshua> hey hey danbri

19:13:30 <danbri> A:|Scroll back a bit for Eikon's summary of RDFLib's node-folding machinery

19:13:30 <dc_rdfig> Titled item A.

19:13:33 <danbri> hi there!

19:13:51 * danbri passing thru on way to dinner...

19:14:23 <danbri> eikeon, thanks for the summary, haven't time now to think it thru, but looks like same sort of thing i'm doing, though i only do it for sql-backed store currently

19:14:45 <deus_x> bon appetit!

19:15:00 <danbri> ta :)

19:15:02 * danbri heads off

19:15:58 <deus_x> eikeon: Oh, and thanks for the help!

19:16:12 <eikeon> Your both welcome :)

19:16:23 * deus_x wanders off to a meeting, wheee!

19:17:17 <danbri> (today's meeting count: 4)

19:17:26 * eikeon wouldn't mind comparing notes with danbri sometime re: node-folding

19:18:26 <eikeon> So far folding at query time is holding up performance wise... surprisingly. But will probably need to do something different for it to scale better.

19:18:53 <eikeon> Perhaps another index someday.

19:20:37 <danbri> A:I want to compare/contrast with the half-done [http://www.w3.org/2001/12/rubyrdf/pack/lib/squish.rb|rubyrdf] code (see 'def defrag' method). Currently I have a little harvester which defrags/compacts/smushes an SQL store each time it adds a parsed RDF doc.

19:20:37 <dc_rdfig> Added comment A1.

19:21:04 * danbri nods to eikeon, it'd be good to talk properly about this... (am really leaving now ;-)

19:21:24 <danbri> (also to write some test cases for such features of an rdf toolkit)

19:21:29 <eikeon> danbri: sounds good... talk later.

19:21:32 <danbri> cu

19:21:51 <danbri> oh, foafbot-over-redland also has this functionality, somehow (in Python over Redlands API I think; ask Edd)

19:29:18 <icepick-away> icepick-away is now known as icepick

19:38:53 <icepick> icepick is now known as icepick-away

22:49:21 <icepick-away> icepick-away is now known as icepick

22:50:59 <icepick> icepick is now known as icepick-away

23:02:08 <DanConn> anybody else getting 'connection refused' from http://www.w3.org/ ?

23:02:35 <deltab> no

23:04:01 <niq> danconn, there's some DNS trouble floating around the 'net

23:04:49 <DanConn> I think it was a (very) local DHCP problem.

23:05:34 <niq> webthing came under a DOS attack today, too ... and my dialup ISP (demon) is malfunctioning in some areas. Don't know how widespread trouble is

23:06:54 * niq tried to access www.spews.org but can't resolve its DNS from here:-(

23:07:17 <deltab> directly?

23:07:22 <niq> nor osirusoft - which smells of an attack by spammers

23:08:15 <niq> deltab, was able to resolve it manually by asking a root server and following the trail

23:08:50 <deltab> whoa, the gtld servers have changed names

23:11:18 <niq> webthing ran out of disc space on /var, due to massive growth of firewall log at 100s of entries per second:-(

23:11:40 <niq> who knows where else mischief might be happening?

23:13:11 <deltab> no useful responses from spews.org's 12 nameservers

23:13:56 <deltab> niq: which server?

23:14:17 <niq> which server what in particular?

23:15:03 <niq> if you mean me, I only have the one - serves various virtualhosts from the same box

23:16:14 <deltab> which nameserver gave a useful response - I'm not seeing any

23:17:29 <deltab> doesn't really matter, since most requests will fail anyway

23:19:02 <niq> indeed. I don't understand how this can happen without it getting reported, though ...

23:19:30 <niq> (I mean places like theregister)

23:26:13 <niq> aha, some light on it at http://groups.google.com/groups?hl=en&lr=&ie=UTF-8&threadm=courier.3E197E29.0000339A%40ny.email-scan.com&rnum=3&prev=/groups%3Fas_q%3D%26num%3D10%26as_scoring%3Dr%26hl%3Den%26ie%3DISO-8859-1%26btnG%3DGoogle%2BSearch%26as_epq%3Dspews.org%26as_oq%3D%26as_eq%3D%26as_ugroup%3D%26as_usubject%3D%26as_uauthors%3D%26as_umsgid%3D%26lr%3D%26as_qdr%3D%26as_drrb%3Db%26as_mind%3D6%26as_minm%3D1%26as_miny%3D2003%26as_maxd%3D6%26as_maxm

23:26:13 <niq> %3D1%26as_maxy%3D2003%26safe%3Dimages

23:26:30 <niq> oops ...