Friend of a Friend (FOAF) IRC Chat Logs for 2003-10-21

00:00:02 <JibberJim> It's something I've been meaning to test, dumping my data into a different schema, but then I have lots of things I've been meaning to test...

00:00:35 <mattb> reminds me, i need to ping edd for the latest foafbot src, check out his smushing algo

00:00:49 <mattb> there's a step missing from my latest one

00:01:04 <mattb> just recovering from a week's illness, brain's only just ticking over

00:01:27 * libby discovers a clash in sha1s :(

00:01:29 <libby> mysql> select * from resources where keyhash='-861849320';

00:01:29 <libby> +------------+--------------------------------------------+

00:01:29 <libby> | keyhash | value |

00:01:29 <libby> +------------+--------------------------------------------+

00:01:29 <libby> | -861849320 | mailto:amy@w3.org |

00:01:30 <libby> | -861849320 | _:f70ffae235c22616253231ab69526ab667533899 |

00:01:32 <libby> +------------+--------------------------------------------+

00:01:34 <libby> 2 rows in set (0.27 sec)

00:02:13 <mattb> i get different hashes via commandline sha1sum for those

00:02:19 <libby> hm

00:02:28 <mattb> ah, my mistake

00:02:30 <JibberJim> http://jibbering.com/foaf/smush.code.txt is my smush code

00:02:42 <mattb> strip the carriage return, the first 8 hex chars are the same

00:02:47 <libby> something has definitely gone wrong

00:02:58 <mattb> why do you throw away the rest of the sha1?

00:02:59 <mattb> int size?

00:03:39 <libby> not sure!

00:03:50 <libby> should probably stop :)

00:03:59 * mattb wonders how hash functions compare in performance

00:04:01 <libby> I'm using int(11) so shouldn;t matter

00:04:26 <mattb> i have a vague feeling sha1/md5 are unnecessarily cpu-heavy for non-crypto apps

00:04:30 <libby> I had 2 duplicates on 370000 triples

00:04:43 <mattb> they're not just about avoiding dupes, they're about not ever ever being able to recover the input from the output

00:05:00 <libby> yeah

00:05:04 <mattb> and about having good cascading behaviour

00:05:16 <mattb> where changing n bits in the output changes more than n bits in the output

00:05:20 <mattb> or similar

00:05:37 <mattb> so that if you do find a clash, it's not one that the user wouldn't be able to detect

00:05:43 <mattb> you can't just add a few spaces to the input text

00:05:52 <JibberJim> I only find ~1000 predicates, I reckon for predicates, no hash, but an in store/memory look up table would be quicker.

00:05:57 <LTjake_> LTjake_ is now known as LTjake

00:06:02 <mattb> dajobe mentioned using the perl internal hashing algorithm

00:06:27 <LTjake> Hi.

00:12:56 <danbri> is that documented for non-perl users?

00:14:40 <mattb> redland src says perl docs say it's http://burtleburtle.net/bob/hash/doobs.html

00:15:00 <JibberJim> Hi LTjake

00:15:11 <mattb> be interested to see if that algo gives any clashes on anyone's scutter store

00:15:33 <mattb> i doubt that cpu time taken by sha1 is significant in the face of mysql processing and java chewing RDF, though

00:15:43 <LTjake> G'day JibberJim.

00:18:04 <LTjake> JibberJim: got a few mins to instruct me on downloading the most recent foafnaut?

00:18:07 <JibberJim> Does anyone have any thoughts on my shipping my DB bnode ID's in the RDF for foafnaut to consume?

00:18:09 <libby> nn

00:18:38 <mattb> i've been wondering about that

00:19:01 <mattb> currently for passing between scutter and joseki i map all bnodes to http://hackdiary.com/ns#thebnodeid

00:19:17 <mattb> so i can refer to them again later if i need to smush them together

00:19:45 <mattb> if there's no explicit nodeid in the input rdf, i use the redland internal bnode identifier

00:21:56 <mattb> how are you planning to use it?

00:23:40 <JibberJim> Yes Jake, I'll just prepare a tarball for you.

00:25:27 <LTjake> groovy, thanks.

00:26:38 <JibberJim>http://jibbering.com/foaf/foafnaut.20031021.tar.gz

00:26:51 <JibberJim> Only a really tiny cache of files, but enough to test things I think.

00:27:33 <JibberJim> Currently matt, foafnaut relies on sha1sum as the index into people, I need to move from that as not everyone has an email address, and the only universal ID people have is my bnode.

00:27:43 <mattb> yeah, that's the problem

00:27:52 <mattb> you need an explicitly way of pointing at a bnode

00:27:59 <mattb> even if only temporarily

00:28:06 <mattb> such as between server and client

00:28:14 <LTjake> JibberJim: Thanks again.

00:28:52 <JibberJim> Yes, I was thinking of just shipping the subj which I have in the store as a foafnautID literal

00:29:24 <JibberJim> it won't be instantly understandable LTjake!

00:32:05 <LTjake> I'm new to SVG, so, no doubt!

00:32:25 <JibberJim> It's mostly javascript, all the SVG is pretty much abstracted out into the templates now.

00:33:06 <LTjake> nice.

00:33:38 <LTjake> that gedcom app i made is perl based.

00:33:40 <JibberJim> but you'll need to create new templates for the relationship schema, and modify the opComplete function to know about things other than knows.

00:34:42 <JibberJim> Well, the most important thing to do is have your geneogy data look something like http://jibbering.com/foaf/example.xml for each person

00:35:03 <danbri_> can you make the bnode IDs so they're unique, not going to clash with other RDF stores inventing theirs?

00:35:40 <mattb> that's why i used a namespace mapped from bnodeids instead of just the ids themselves

00:36:21 * danbri_ needs to be tired. stupid timezones.

00:36:56 <JibberJim> hmm, I could, but I'd rather not, it would increase verbosity for little gain - as I don't intend to be serialising it as RDF referencing those bnodes as objects, but as a literal, or do people think that is a bad idea?

00:37:19 <mattb> in this in your xml-based pseudo-rdf?

00:37:27 <danbri_> depends on likelihood of that rdf leaking out and being merged elsewhere

00:37:28 <JibberJim> Yeah.

00:37:37 * mattb echoes danbri_'s concern

00:37:54 <mattb> i worry slightly about ever republishing rdf

00:37:59 <mattb> someone else's rdf

00:38:10 <mattb> answering queries, fine

00:38:19 <JibberJim> but if I generate predicates such as <something> <jimsNS:foafnautID> "_:ahfdkjhfksdh" that shouldn't effect anyone should it?

00:38:20 <danbri_> or someone hacking the js to merge in other RDF from a different source (or scutter instance) with clashing IDs

00:38:25 <mattb> but doing such subtle rewriting of bnodes to concrete uris would be odd

00:38:36 <mattb> JibberJim: that sounds ok to me

00:38:39 <mattb> just another IFP

00:38:42 <mattb> for your own use

00:38:45 <danbri_> <JibberJim> but if I generate predicates such as <something> <jimsNS:foafnautID> "_:ahfdkjhfksdh" that shouldn't effect anyone should it?

00:38:47 <danbri_> <JibberJim2> but if I generate predicates such as <something> <jimsNS:foafnautID> "_:ahfdkjhfksdh" that shouldn't effect anyone should it?

00:38:50 <danbri_> <JibberJim3> but if I generate predicates such as <something> <jimsNS:foafnautID> "_:ahfdkjhfksdh" that shouldn't effect anyone should it?

00:39:00 <danbri_> probably not, Jims. But you never know ;)

00:39:17 <danbri_> might be other foafnaut instances, etc.

00:39:42 <JibberJim> Other foafnaut instances, it's hard enough keeping on the go!

00:40:06 <danbri_> it is a pretty pedantic concern, should be ok. can always change it again later.

00:40:16 <danbri_> hmm P900 http://www.allaboutsymbian.com/news.php?id=17015

00:40:29 <danbri_> audio in my phone been playing up. Reckon if I take it back in can get a P900?

00:41:04 <JibberJim> Yeah I reckon so.

00:44:12 <LTjake> JibberJim: should i bother using rel:siblingOf in these foaf files, because i can infer that through parents...?

00:46:03 <JibberJim> I'd but it in, because the way foafnaut works, it always requires a link from A-B otherwise B won't appear, even if there are linsks B-A - which is why foafnaut uses the jim:isKnownBy in there.

00:47:16 <LTjake> right, but i guess i'd only need that if i want links between siblings...

00:47:26 <LTjake> most family trees link through parents...

00:47:54 <JibberJim> That's true - but remember foafnaut is not "most family trees :-)"

00:48:01 <JibberJim> but yeah, you're right.

00:48:41 <LTjake> heh. true. i'm trying to shape it for my data.

00:50:50 <JibberJim> I have no experience of family trees mind you (coming from Devon it would probably be a bad idea to look too closely!)

00:51:56 <LTjake> i know very little. the data is my father && uncle's work.

00:53:53 <LTjake> were you saying i could link via rdf:ID? (i know you said it was good that i had unique ids, at least)

00:54:31 <JibberJim> I'm not sure what would be best, mattb, do you have a suggestion on this one?

00:55:26 <JibberJim> LTjake's got a load of geneology data, but there's no IFP's for the people - any idea what a good solution would be?

00:55:42 <mattb> how many people?

00:55:46 <LTjake> sample output: http://red13.alternation.net/gedcom/I0080/foaf

00:55:55 <LTjake> mattb: 1000+

00:56:59 <JibberJim> but I think this would be a good demo of all genealogy data, plugging foafnaut as a geneaology browser maybe :-)

00:57:11 <mattb> sounds very neat

00:57:24 <danbri_> that'd be sweet!

00:57:35 <LTjake> hell yeah.

00:57:49 <JibberJim> I was thinking on how to make it do parents/children/sibling in the right direction on the explosion too...

00:58:57 <mattb> unique IDs sound fine to me for linking up this dataset

00:59:06 <mattb> since it's all under your control

00:59:37 <LTjake> right.

00:59:56 <mattb> i guess a key thing about using IFPs to distinguish individuals in public data is so that two different data sources don't have to agree on a URI for an entity

01:00:08 <LTjake> can't combine gedcom sets, then, though.

01:00:08 <mattb> but here you can be a source of IDs

01:00:12 <danbri_> yep

01:00:21 <mattb> ah, so they're not unique IDs

01:00:26 <LTjake> well, yes.

01:00:28 <mattb> only unique within the dataset

01:00:31 <LTjake> for this data set.

01:00:41 <danbri_> more re IFPs in http://rdfweb.org/mt/foaflog/archives/000039.html

01:01:00 <danbri_> same thing crops up in FoafCorp btw., we know within the theyrule.net dataset that each name is unique...

01:01:15 <danbri_> which is fortunate, as we don't have much other info on the company directors in that db

01:01:27 <mattb> does the gedcom set have a unique id?

01:01:30 <mattb> for the set?

01:01:37 <LTjake> mattb: yes.

01:01:46 <mattb> you could prefix it onto the individual IDs

01:02:05 <mattb> to get a globally unique id

01:02:23 <JibberJim> Combining then would rely on owl:sameIndividualAs ?

01:02:40 <JibberJim> 8119 in an ecademy file :-(

01:02:47 <mattb> tricky

01:03:08 <LTjake> personally, i don't need to combine 'em =)

01:03:18 <LTjake> but it could be handled at the gedcom level.

01:03:55 <mattb> i wonder how geneology data gatherers do merging on a large scale

01:04:02 <mattb> what IFP-like heuristics they use

01:04:27 <LTjake> no clue!

01:05:29 <JibberJim> hmm, well if we assume familysearch.com is doing it then they don't seem to prompt for much other than name birth/death/marriage etc. and parents names

01:05:38 <JibberJim> .org I mean - http://www.familysearch.org/Eng/Search/frameset_search.asp

01:05:53 <mattb> i guess you can match if the 'nearby' graph seems to contain people with the same names in the right relationships

01:06:03 <mattb> i imagine you'd have to traverse a little

01:06:12 <mattb> not just use a single triple as an ifp

01:06:39 <LTjake> gedcom data can be quite extensive, if you want.

01:07:14 <LTjake> lots of stuff to match up.

01:08:05 <LTjake> of course, we're not even sure if our own data is correct.

01:08:29 <LTjake> 100% that is.

01:08:36 <LTjake> most of it is right =)

01:08:47 <JibberJim> I found a James Ley in Cornwall in 1881...

01:09:45 <KevinMarks> KevinMarks is now known as _KevinMarksEats

01:17:02 <LTjake> time for some Z's.

01:17:04 <LTjake> cya

09:04:32 <zool> morning

09:04:36 <edd> heya

09:04:41 <zool> hi edd

09:31:55 <zool> morning mattb

09:31:58 <zool> hey JibberJim

09:32:03 <JibberJim> hey zool

09:32:09 <zool> and dajobe and etc etc, sorry

09:32:16 <mattb> morning zool

09:32:25 <JibberJim> aww, I thought I was special :-(

09:32:29 <mattb> mornign jim

09:32:34 <dajobe> hello

09:33:02 <JibberJim> Foafnaut's DB died, disk errors...

09:35:02 <JimH> :-(

09:35:22 <JibberJim> well actually the table used for foafnaut queries is fine, but others are dead

09:58:31 <mattb> oops

10:27:45 <JibberJim> erm, that's a bit odd, Number of rows changed from 228214 to 389340

10:28:12 <JimH> I'd added a little to my Foaf file yesterday, but not *that* much...

10:28:40 <JibberJim> Where did it find 170,000 triples from? That was on a Repair table, normally they disappear, not find strange new ones.

10:48:56 * libby discovers spelt 'under' wrong all through my software

10:49:09 <edd> That's extreme.

10:49:17 * danbri_ gives up

10:49:23 <danbri_> how can you spell 'under' wrong?

10:49:27 <danbri_> undre?

10:49:29 <libby> beats me

10:49:33 <libby> 'unider'

10:49:35 <libby> sigh

10:49:43 <danbri_> that's individual

10:49:49 * edd chortles

10:49:50 * libby thought so

10:50:17 <danbri_> Results 1 - 10 of about 236.

10:50:22 <danbri_> you're not alone. But nearly so.

10:57:37 <edd> foo easynet

11:02:40 * danbri_ grins at http://www.indecorous.com/experiment/index.html

11:02:45 <danbri_> I always meant to do that.

11:02:59 <danbri_> grew out of it before i got around to it, thankfully

11:03:22 <danbri_> I like the idea of recording one's decline. But now we have weblogs.

11:04:20 <JibberJim> don't you get them only when you've already declined?

11:11:28 <mattb> ooh, indecorous

11:11:30 <mattb> that's ian malpass

11:11:33 <mattb> he works in my office

11:11:34 <mattb> nice guy

11:12:46 <JimH> with a headache?

11:13:38 <danbri_> heh, pretty random. I found him via googling for 'pseudoephedrine' (clearing out my pharmaceutical shelf!)

11:13:50 <mattb> heh, he's wrote that notebook many years ago

11:14:01 <mattb> there's a whole bunch of fun stuff on indecorous

11:14:57 <JibberJim> Do you mean clearing out, or trying one of everything in there danbri_?

11:17:09 <danbri_> ooh, a pink one

11:19:46 * libby and shellac get delicious birthday hamper

11:19:48 <libby> yum

11:20:21 <JibberJim> Happy Birthday to Damey, Happy Birthday to Damey...

11:21:39 <JibberJim> Happy Birthday to Damey, Happy Birthday to Damey...

11:21:55 <shellac> :-)

11:22:17 <shellac> thanks jim

11:28:44 * shellac also got whsmith vouchers

11:29:05 <shellac> they have 'Lone Wolf and Cub' - hurrah!

11:29:28 <danbri_> addict

11:29:46 <shellac> smiths stocking violet manga - tut

11:42:49 <darobin> yuk pernod with no water, no wonder he didn't like it

11:43:38 <edd> violet? what is the recommended colour?

11:51:38 <shellac> oops - is this thig workig?

11:51:50 <zool> mm?

12:06:30 <laszlo> shellac, it is ideed workig but you do't have ay 's.

12:07:30 <JibberJim> Plumbers are evil, they install a load of radiators, then turn of the boiler so there's not even hot water and leave!

12:57:15 <Netminder> Netminder is now known as NetHome

13:21:19 <nmg> morning (afternoon) all

13:21:29 <danbri_> hi there

13:23:01 <libby> heya nick

13:25:26 <JibberJim> Did anyone create Avatar-like image vocab?

13:25:51 <JibberJim> I was wondering about one for foafnaut - a tight head shot you could use in place of the blub.

13:26:51 <darobin> foaf:avatarPic?

13:27:32 <JibberJim> Yeah, but I think it needs to be bound to the forum etc. that uses it.

13:27:47 <danbri_> was talking to Anselm Hook and Jason Harlan about this in SF last week...

13:27:57 <JibberJim> the Avatar I want to use on the S&M web-board, is a bit different to the one on SVG-developers

13:28:08 <danbri_> Anselm does games stuff; Jason's dayjob is body models for clothing website...

13:28:08 <zool> boggle

13:28:31 <JimH> foaf:headShot?

13:28:38 * danbri_ assumes you meant M&Ms

13:28:46 <JimH> does it have to be an avatar?

13:29:00 <JibberJim> Oh sorry, horrible spoonerism - M&S, I talk about their foodhall...

13:29:21 <danbri_> if you look at the typepad feeds, lots of people choose non-representational self depictions

13:29:55 <JibberJim> not so much for foafnaut where yeah I really want a headshot, but I think it's in the same problem space as the avatars.

13:30:10 * JimH must be an old fart as he's ceased to see the point or humour in avatars...

13:30:24 * JibberJim agrees with JimH entirely!

13:30:28 <danbri_> portability between computer games?

13:33:05 <danbri_> mattb? mortenf? how do you connect your geko/serial to a PC?

13:33:23 * danbri_ looking at libby's spare serial to usb adaptor, she mentioned it apparently needed drivers for macosx...

13:33:28 * danbri_ about to try it on linux

13:33:36 <danbri_> otherwise, am off to Maplin.

13:37:46 <danbri_> ah http://www.hackdiary.com/archives/000039.html :)

13:38:26 * JibberJim sits down to a shiny new radiator

13:39:38 <JibberJim> it's stone cold, but at least it's progress

13:43:41 * JibberJim organises meeting in Bath on thursday afternoon...

13:44:14 <danbri_> what for?

13:44:19 <libby> wil you be passing through BRS?

13:46:54 <JibberJim> I was wondering about popping into wish you Happy B'day libby...

13:47:01 <libby> that'd be neat :)

13:47:14 <libby> we have a meeting all day and then later going bowling I think

13:47:19 <libby> not sure about that

13:47:32 <sbp> ooh, happy birthday libby!

13:47:47 <danbri_> not yet! she's still young.

13:47:47 <JibberJim> Not todal sbp, that's Damey

13:47:52 <sbp> heh

13:47:56 <libby> I think we should be back by 5.30

13:47:58 <sbp> oh man. happy birthday Damey!

13:48:01 <libby> heheh

13:48:06 <libby> thansk anyway sbp :)

13:48:20 <sbp> yep. well, now on the day I can just point to logs

13:48:28 <libby> I think we should be back by 5.30 - from the meetign I mean

13:48:29 <libby> heheh

13:48:49 <dajobe> reminds me, libby, danbri_ - do you want a lift thursday?

13:49:14 <danbri_> there was talk of a taxi one day...

13:49:22 <libby> that's fri...

13:49:29 <dajobe> I'm not going friday

13:49:36 <danbri_> we have max and charles here, so prolly need cab.

13:49:41 <dajobe> ah, ok

13:49:50 <danbri_> s/have/will have/

13:51:38 <libby> we could do somethign complicated with a black cab

13:52:08 <dajobe> well, let me know

13:52:29 <danbri_> ok ta, will have a think

13:53:01 <libby> yeah, thanskf ro the offer dave

13:56:07 <danbri_> http://swordfish.rdfweb.org/discovery/2002/02/paths/ is much easier to use now, with the name shortcuts

13:56:09 <danbri_> libby++

13:56:37 <libby> :)

13:56:45 <libby> dont know why I didnt thik of that before

13:56:58 <danbri_> can you explain what exactly it is doing, when i ask for path from 'dan' to 'dan' ?

13:57:12 <danbri_> it seems to fix 1st dan to me, then rotate thru the others, which is cool...

13:57:24 <JibberJim> hmm, it seems to have some problems...

13:57:29 <danbri_> or every combination?

13:57:31 <JibberJim> Jim Ley to John F. Kennedy via Jim Ley, Ian Hickson, Charles McCathieNevile, Tim Berners-Lee, Bill Clinton

13:57:35 <JibberJim> I go via myself?

13:57:52 <libby> hm

13:58:03 <libby> you might if there are two email addresse for you I think

13:58:39 <JibberJim> Unlikely... certainly not as an mbox, maybe as a sha1sum

13:59:08 <libby> hm, I think with youjim, it's behaving slightly oddly because there's a pic of you on your own

13:59:23 <JibberJim> I know, I have no friends :-(

13:59:34 <libby> yes you do!

13:59:41 <libby> but there's alos a pic of you on own

13:59:55 <libby> one of me too, so I wonder if it does that with me too

14:00:00 <JibberJim> Ah, right, but I've got Beer - that's always my friend.

14:00:10 * darobin noticed that

14:01:17 <libby> hm, yeah, it's behaving oddly

14:03:39 <edd> funny_beer--

14:03:57 <JibberJim> funny beer?

14:04:14 <JibberJim> I was in Germany, one had to fit in...

14:04:57 <danbri_>http://swordfish.rdfweb.org/discovery/2001/08/codepict/aboutimage.jsp?thumb=http://www.picdiary.com/~mattb/windowclip.php?size=50&file=/foafmeet/img_2300.jpg

14:05:12 <danbri_> lib, I think you need to escape the & etc. in the quoted URL

14:05:26 <libby> ah

14:05:32 * libby was wonderign what was up with those

14:06:10 <libby> there's also some odd ones e.g. of martin durst, where the pics have been removed

14:06:25 * edd refrains from comment

14:09:55 * edd wants to make jim a tshirt with "wordnet:Beer is my foaf:friend" on it

14:11:49 * JimH thinks about fighting Mr Ley for the rights to the shirt

14:12:02 * libby too

14:12:06 <libby> hm

14:13:50 * nmg tries to work out the OWL to say "wn:Beer is my only friend"

14:15:02 <libby> an owl usecase!

14:15:10 <edd> congratulations :)

14:15:36 <darobin> lol

14:15:39 * libby thinks we have a new *official foaf tshirt* ;)

14:15:52 <danbri_> here you go, <owl:FunctionalProperty rdf:about="&foaf;friend"/><foaf:Person rdf:about="&me;"><foaf:friend><wn:Beer/></foaf:friend></foaf:Person>

14:15:53 * darobin wants one too!

14:16:19 <danbri_> you could be fancier and make foaf:friend functional over a singleton class whose sole member is you...

14:16:55 <danbri_> does that work, nick?

14:18:03 <edd> isn't "only" how you want to qualify the Beer's friendness, not friendness in general?

14:18:33 * nmg agrees with edd

14:19:07 <nmg> of course, you could define a foaf:bestFriend that's functional

14:19:08 <danbri_> I read it more as 'there is only one friend of ... and that friend is beer.'

14:19:17 <nmg> (assuming you can't have more than one best friend)

14:19:24 <danbri_> uhoh. funny nobody has proposed that one yet!

14:19:35 <JibberJim> foaf:onlyFriend

14:19:37 <edd> i bet you aything livejournal users would

14:19:42 <edd> and they'd change it every day, too :0

14:19:47 <nmg> heh.

14:20:04 <edd> i loved the grouphug.us confession about being addicted to a 15 year old's livejournal

14:20:13 <nmg> they'd interpret "once looked at my livejournal" as "they are my frieeend"

14:20:13 <edd> there is a certain morbid fascination to some of them

14:20:23 * danbri_ missed that one, they kinda blur after a couple hundred pages

14:20:24 <nmg> car crash viewing

14:20:32 <nmg> can't drag your eyes away

14:20:46 <edd> nmg: dead on, like a certain other site which i've foresworn

14:21:17 * JibberJim was depressed to see there was 300 new ones yesterday, if there was maybe 10... but I must not look...

14:21:17 <nmg> which one's that?

14:22:27 <danbri_> www.filmratings.com -- try a search on empty string

14:23:20 <darobin> ouch

14:23:26 <darobin> that almost crashed me

14:23:26 <edd> lol, most violent first. cool.

14:25:56 <nmg> creature comforts? strong sexual content? in plasticine?

14:26:08 <sbp> hmm... [ foaf:nick "nmg"; rdf:type [ rdfs:subClassOf [ rdf:type owl:Restriction; owl:onProperty foaf:friend; owl:cardinality "1" ] ]; foaf:friend wn:Beer ] .

14:26:12 * edd diverts his mind onto other things

14:26:40 <nmg> sbp: that works for me

14:27:22 <JibberJim> ooh ooh, the radiator next to me is making strange gurgling noises.

14:27:41 <darobin> "it's a-liiiiive!"

14:27:51 <sbp> now we just have to lobby for foaf:friend

14:28:08 <sbp> odd that a Friend of a Friend schema wouldn't have a foaf:friend property, actually

14:28:39 * libby kinda likes that

14:29:05 * danbri_ too

14:29:09 <danbri_> makes you fink

14:29:16 <sbp> heh, heh

14:32:28 * sbp - aware that everybody's collecting provenance data but not really using it - publishes foaf:friend = foaf:knows in all his RDF

14:32:49 <danbri_> '=' ?

14:32:54 <danbri_> owl:sameAs?

14:33:05 <sbp> owl:sam... well, I was going for samePropertyAs, but yeah

14:33:15 <nmg> owl:equivalentPropertyTo, please...

14:33:32 <sbp> sigh. I still use all the DAML names

14:34:11 <nmg> sorry, just owl:equivalentProperty

14:34:13 <sbp> hmm. it's "owl:equivalentProperty", no?

14:34:14 <sbp> yeah

14:34:15 <sbp> heh, heh

14:34:33 * nmg gets confused too, since he was there for all the arguments about what to call these things

14:34:55 <sbp> good point

14:35:39 <JibberJim> hmm, it maybe alive, but it it's still stone cold...

14:38:00 <sbp> heh, DanC on #rdfig: <DanC_RSW> aharth, have we met? (I'm foaf:topic of http://www.w3.org/People/Connolly/)

14:38:31 <sbp> does that make foaf:homepage owl:inverseOf [ rdfs:subPropertyOf foaf:topic ] true?

14:39:01 <sbp> FOAF needs an axiomatic semantics

14:39:34 <danbri_> foaf:topic and foaf:page are inverses, and homepage is superproperty of page. So I don't think the inference can be derrived, but it makes sense...

14:39:40 <libby> wanna write one? ;)

14:40:09 <sbp> homepage is a superproperty of page?

14:40:09 <danbri_> draft stuff in http://rdfweb.org/topic/FoafAxioms ... we can put N3 rules linked from the ns uri if there's enough worth saying

14:40:15 <sbp> surely it should be a subpage?

14:40:19 <sbp> ...subProperty

14:40:51 <sbp> that is, ?x foaf:homepage ?y entails ?x foaf:page ?y, but not t'other way around

14:41:33 <danbri_> yeah, you're right. I always get confused cos the way we named those properties is backwards from normal conventions.

14:42:53 <sbp> yeah. just slurped it from the specification. so I guess that'd make foaf:homepage owl:inverseOf [ rdfs:subPropertyOf foaf:topic ] true... but it's more like foaf:homepage rdfs:subPropertyOf [ owl:inverseOf foaf:topic ]... I wonder if those two are equivalent?

14:45:43 * sbp asks on rdfig

14:45:45 <danbri_> foaf:topic is pretty useful...

14:47:13 <danbri_> hmm would we do a film review like this? <Document><topic><wn:Film><foaf:homepage><Document rdf:about="http://www.lost-in-translation.com/"/></foaf:homepage></wn:Film></topic></Document>

14:48:09 <sbp> looks reasonable

14:48:09 <danbri_> and find it again by asking for things ?x where ?x foaf:topic ?y . ?y foaf:homepage http://www.lost-in-translation.com/

14:48:25 <danbri_> which means the type on the intermediate thingie isn't that important

14:48:55 <sbp> unless you wanted to get reviews of a certain type of thing

14:49:12 <sbp> ?x foaf:topic ?y . ?y rdf:type wn:RadioShow .

14:49:32 <sbp> less likely, unless you were creating an index or something

14:49:56 <danbri_> yeah, a browse tree could use that...

14:50:33 <edd> danbri_: my P800 GnuPOC account just made Slashdot, as part of the P900 announce

14:50:35 * edd glows :)

14:50:36 <danbri_> I think people see FOAF and think 'social networking', and ignore or don't realise this aspect of FOAF (aka RDF + candy coating) is pretty powerful...

14:50:44 <danbri_> whee!

14:50:56 * danbri_ curious how many hits you get

14:51:23 <edd> well, they linked google's cache of the page, but of course i still get hit for the image

14:51:42 <danbri_> do they normally do that now?

14:52:01 <edd> dunno, but my logs are confirming plenty of traffic to the image.

14:52:37 <danbri_> [

14:52:37 <danbri_> Please don't mail me asking for more information, as this is as far as I've got. You should also read Dan Brickley's documentation of this adventure, as it is more precise in recording the detail than I have been. There's a lot more of this SDK left to get working, but at least some of the examples now build and install.

14:52:38 <danbri_> ]

14:52:51 <danbri_> wonder how many will get that far... involves scrolling.

14:52:58 <danbri_> do slashdot readers scroll?

14:53:07 <oPless> ? :)

14:53:08 <edd> You are in the best position to measure that.

14:53:19 <oPless> [03/10/21 15:52] <oPless> P900 review on /.

14:53:19 <oPless> [03/10/21 15:52] * Disconnected

14:53:26 <oPless> mirc-- :(

14:54:07 <danbri_> damian says stats are borken currently

14:54:53 <JibberJim> HEAT!!!!

14:55:05 * JibberJim dances around the room in joy

14:55:35 <danbri_> heh

14:55:42 * danbri_ has his window wide open still

14:56:15 <oPless> it's freezing here

14:56:40 <oPless> and I think I left my heating on "jungle turbo boost"

14:56:56 * oPless expects heating bill++ :(

14:59:06 * sbp gets led to http://epiphany.mozdev.org/

14:59:41 <edd> go sbp

14:59:49 <edd> ephy rocks

15:01:15 <danbri_> edd, what currently ails foafbot?

15:01:28 <edd> a deficiency in circular tuits

15:01:46 <edd> i believe i have now updated debian sid properly, so there's no obstacle for me in getting it running again

15:01:50 <edd> i did that last friday

15:01:50 <danbri_> sounds serious!

15:02:09 <edd> however, i've been caught up in complications due to caring for bedbound wife

15:02:14 <JibberJim> Danbri, what's the status of foaf-project redesign?

15:02:20 <edd> oh, and i'm just crap and ineffective, too

15:02:55 <dajobe> yeah, the latest redland now being in debian sid/unstable

15:03:10 <dajobe> foafboat was running on the cvs version equivalent to that, on the older pc

15:03:15 <danbri_> nicole froze her css stuff a while back, but we were 99% out of touch while she was in Boston for the summer.

15:03:16 <dajobe> foafbot :)

15:03:23 <edd> fofboat

15:03:40 <danbri_> she's back in paris now, am going to check her stuff into cvs when i get a tarball...

15:03:40 <edd> i shall attempt to restore it tonight

15:03:42 <edd> keep naging

15:12:06 <sbp> edd: will ephy let me put the navigation/address/URI field on the menubar? I've a very low screen res, so I like to have menubar + URI at top, then nav buttons and bookmarklets on second

15:12:24 <edd> sbp: no

15:12:49 <sbp> hmm. but perhaps I can cope with three bars...

15:13:22 <edd> i have two bars, i shove the bookmarklets in a drop-down on the right hand side of the URI

15:13:44 <edd> if you activate the 'edit toolbar' option, you can drag and drop the bookmarklet thingies around, too

15:14:09 <sbp> hmm. I tend to lean on my bookmarklets an awful lot; it'd probably be worth having an extra bar

15:16:05 * JibberJim has stopped using bookmarklets entirely... I used to use them a lot and all sorts of things.

15:18:04 <sbp> why'd you stop?

15:19:09 <JibberJim> dunno...

15:19:42 <edd> his hands got too cold to use the mous

15:19:42 <edd> e

15:19:55 <JibberJim>http://www.iamcal.com/misc/londonmap/

15:20:07 <edd> i like that you can use all the smart bookmark searchs in ephy from the URI bar

15:20:37 <JibberJim> what's a "smart bookmark search" ?

15:21:20 <edd> like a google search box, or a dictionary.com search box

15:21:28 <edd> you program in the URL with a %s where you want your text to be

15:21:39 <edd> type in the URI bar, and choose from the dropdown that appears which service you want

15:22:26 <JibberJim> Right, so the sort of thing Konqi and IE and Opera have had for years? I thought that was standard across all browsers now.

15:23:27 <edd> hmm, no that's not the same as in other browsers, you might be misunderstanding me

15:23:39 <edd> anyway, epiphany's selling point is simplicity, not overfeatured ness

15:24:06 <nmg> rehi

15:24:58 <JibberJim> Hmm, in IE4 you go "g chicken" or "d chicken" into the URI bar, and it does a google or dictionary search.

15:25:40 <edd> right. but here you type "chicken" and you get a drop-down menu. similar sort of thing

15:26:02 <JibberJim> Oh right, that's quite good then - don't need to remember the G etc.

15:26:04 <edd> handy for me, as i never remember the one letter cut

15:37:19 <sbp> hmm. that's why I use bookmarklets: click, type/paste, enter

15:37:42 <sbp> well, one of the reasons. w.a.o and HTTP HEAD etc. are different

16:06:25 <danbri_>http://www.w3.org/TR/NOTE-MCF-XML-970624/complex.gif

16:06:42 <danbri_> healthy overlap w/ foaf vocab, maybe worth re-doing the image and putting explicit ack in the spec...

16:12:06 <sbp> uh oh. epiphany.mozdev.org /.'d?

16:18:54 <edd> nope, just down by the looks of it

16:19:05 <edd> there's a #epiphany on irc.gnome.org if you need any expert info

16:19:45 <sbp> excellent; thanks

16:21:47 <JibberJim> hmm it seems like ecademy is now seeAlso'ing lots of non-RDF data - photos, blog, account details for each.

16:22:20 <libby> it's legal....

16:22:32 <nmg> doom, doom!

16:22:43 <JibberJim> I know, but it's a bit of pain making us download 4 for every 1 RDF doc.

16:22:47 <libby> yeah

16:23:02 <danbri_> yeah, legal ish.

16:23:25 <danbri_>http://esw.w3.org/topic/UsingSeeAlso

16:26:07 <nmg> Terry Payne and are have been doing some work on URI resolvers for SW systems (ie. where can I find more information about this resource) that go beyond simple http URLs dereferencing

16:26:25 <nmg> we've ended up with a couple of subproperties of rdfs:seeAlso for this very reason

16:26:27 <dajobe> rddl?

16:27:20 <danbri_> you prefer subprops to typing the thing referenced?

16:27:21 <nmg> one explictly points at rdf/xml files that may be retrieved, the other points at a service (a knowledge base) that can be queried to give a contextual sub graph containing the required info

16:27:59 <nmg> yes, because there are extra aspects of the thing being indicated which affect how you interact with it

16:28:27 <nmg> the general idea was to build a system which you could ask "what have people said about this resource" and get back some sort of meaningful answer

16:28:45 <nmg> approach very much influenced by WHOIS++ (and DDDS), for what it's worth

16:28:46 <zool> mmmm contextual sub graph

16:29:06 <zool> i've been doing such thing also recently

16:29:25 <nmg> zool: we've written something a little like this as a layer above 3store, but it isn't (yet) distributed

16:29:32 <danbri_> whois++ !!!!!

16:29:35 <danbri_> retrochic!

16:29:41 <zool> in simple fashion, the elements surroudning a context for a statement or extar bits of the model for things connected to what is at the uri

16:29:50 * dajobe hands danbri_ an IAFA-Template

16:30:04 * danbri_ had some scribbles here http://ilrt.org/discovery/2000/09/metamesh/

16:30:12 * nmg grins at danbri

16:30:13 <danbri_> somewhere hada whois++/rdf+http gateway script

16:30:21 <zool> e.g. at http://mutemap.openmute.org/dmz/a_node/30410413 i also get a project i'm connected to, vrey simply just the name

16:30:21 <nmg> just ecause it's old, doesn't mean it's not good

16:30:29 * nmg talked all about whois++ in his phd thesis

16:30:31 * danbri_ remembers meeting nmg and talking about this stuff in bath

16:30:33 <zool> 'metamesh' eh

16:30:46 <zool> i don't really know how whois works

16:30:48 <nmg> wow, that was years ago. 1996?

16:31:02 <danbri_> 1998. After RDF came out. Brian Kelly held a seminar.

16:31:03 <zool> sounds like a distr. federation message passing thing again

16:31:32 <nmg> Right - I hadn't realised it was that late (I was at soton by that point)

16:31:34 <danbri_>http://www.ariadne.ac.uk/issue15/events/stakis.html

16:31:57 <nmg> right, I'm off to lunch

16:32:01 <dajobe> "Z39.50 is regarded as a legacy system, which is a bit distressing if you are currently implementing Z39.50". -- danbri, 1998

16:32:15 <nmg> nmg is now known as nmg_lunching

17:21:42 <mortenf> did you get your geko connected danbri (i have a serial port on my machine, so i go directly)

17:22:40 <danbri_> mortenf, not yet, but been busy

17:22:40 <danbri_> or distracted, or both

17:23:03 <mortenf> :)

17:25:35 <dmwaters_> {global notice} Hi all, it appears that one of our main us hubs is having very serious routing issues, I'm going to do emergency hubbing to fix the situation, so please bare with me.

17:48:10 <lilo> [Global Notice] Hi all. Severe packet loss on a main rotation hub server caused repeated reconnects. It's been resolved. Parenthetically, still working on the global notices replacement. Thank you for your patience, and thank you for using freenode!

18:28:19 <nmg_lunching> nmg_lunching is now known as nmg

19:52:46 * mattb tries jibbler's cookie recipe

19:52:53 <Jibbler> mmm cookies

19:53:06 <Jibbler> i have used that recipe myself so many times now :)

19:53:14 <JibberJim> Cookies?

19:53:18 <Jibbler> the sad thing is that i have to keep looking it up on the web

19:53:28 <mattb> JibberJim: http://www.jibble.org/cookies.php

19:53:39 <JibberJim> yes google knew!

19:53:42 <mattb> heh

19:54:20 <JibberJim> Seems to be in some dodgy language, I wonder if I can port it to something else...

19:54:59 <Jibbler> there are 3 mistakes in my cookie recipe, btw

19:55:11 <mattb> in the recipe or the code?

19:55:15 <JibberJim> Hmm, and I've got a breadmaker too, so I could've used that, but the code doesn't seem to be there - I guess that is in the commercial version

19:55:23 <Jibbler> in the code

19:57:20 <mortenf> does anyone know the difference (if any) between = and == in squish?

20:15:51 <dajobe> ya foaf reinvented? http://www.kunekt.com/nie/index.php

20:16:11 <dajobe> that url courtesty of their browser sniffing

20:16:12 <mortenf> not quite...

20:16:19 <edd> think it was discussed here briefly t'other day?

20:16:27 <edd> wow, terms of serice!

20:16:42 <edd> foaf so needs a huge page of legalese to spread the love

20:16:52 <bitsko> is there a significant difference between dc:title and foaf:name? or does it just seem odd to "title" a person or organization, rather than "name" them

20:17:17 <mortenf> i think they're both (or should be) subpropertyof rdfs:label

20:17:32 <bitsko> ah, that's the connection I needed. thx

20:18:10 <mortenf> using dc sometimes seems to infer that the subject is a document

20:19:08 <gromgull> JibberJim, what was the url for your scutterplan again?

20:19:45 * mortenf now has an almost complete squish2myredlandmysql rewriter (with a restriced parser), based on mattb's and libby's impls.

20:21:06 <kasei> mortenf: in what language?

20:21:11 <mortenf> Perl

20:21:17 <kasei> cool

20:21:24 <kasei> is it available? :)

20:21:32 <mortenf> will be shortly

20:21:40 <kasei> excellent

20:22:17 <dajobe> handy

20:22:24 <mortenf> depends on when/if my redland contrib is accepted by its master ;)

20:22:31 <dajobe> oh, I was fixing other stuff

20:22:33 <mattb> nice one mortenf

20:22:40 <mortenf> np, don't hurry

20:22:51 <dajobe> I discovvered I'd left some un-cvs-commited code around

20:22:51 * gromgull found it , http://jibbering.com/foaf/scutterplan.1

20:22:52 <mortenf> thank you mattb, spend quite some time with your note

20:23:13 <mortenf> s/spend/spent/

20:23:13 <mattb> glad to hear it

20:23:27 <mattb> it only took me about 18 months after i came up with the algorithm to actually write it up

20:23:31 <mortenf> heh

20:23:34 <mattb> in those pre-hackdiary days

20:23:47 <mortenf> of course, i had to fix it for split-keys and table limits...

20:23:51 <edd> yep, you got it down to 17 months now

20:23:53 <edd> mattb++

20:24:20 <mattb> it needs fixing for all sorts of things i'm sure

20:24:23 <mattb> i'm glad people have found it useful

20:25:30 * mattb needs to do some writing this evening

20:25:48 <mattb> hence the cookies

20:25:59 <mattb> yak shaving

20:27:01 <mortenf> you got a wn:yak-3?

20:27:48 <JibberJim> gromgull, my scutterplan is far from complete

20:28:22 <JibberJim> I broke a scuttering mid scutter today when I lost network

20:29:03 <gromgull> That's fine, I was just after some url's for the foaf of people I don't know...

20:29:16 <gromgull> I wasn't after a full scutter (not yet anyway)

20:30:34 <mortenf> oh, mattb, one thing i found with your algorithm: it doesn't handle bindings that aren't selected

20:31:04 <mortenf> i.e. ?x some:prop ?y, ?y other:prop ?z - you have to ask for ?y as well

20:31:44 <mattb> ah yes

20:31:48 <mattb> that's true, i should note that

20:32:06 <mortenf> not hard to fix though

20:37:24 <mattb> mmmmmm

20:37:29 <mattb> Jibbler: great cookies, much obliged

21:28:38 * mortenf tweaks squish-like syntax to output context (provenance) as well for ?var+

21:30:32 <mortenf> somewhat not useful yet, as all the triples are from libby...

21:31:57 <mortenf> hmm, should scutter-data-interchange be done by reification to preserve provenance?

21:32:17 * dajobe inserts eek

21:32:21 <mortenf> yeah...

21:32:22 <JibberJim> EEK! I dunno, we still need to solve that

21:32:30 * mattb eeks too

21:32:44 <mattb> definitely needs thought

21:32:48 <mortenf> yep

21:33:33 <mortenf> or, perhaps a new n4 format?

21:33:49 <mortenf> (like ntriples with an extra term)

21:34:21 <JibberJim> hmm, do we want to exchange provenance data as the

21:34:39 <JibberJim> URI that gave us the triple - we might aswell just have the other scutter do that.

21:34:54 <mortenf> not sure i understand?

21:35:34 <JibberJim> I'm imagining that the stores which will exchange info with others will have tighter provenance - ie "this stuff is from Jim Ley" or "this stuff is from ecademy" etc. rather than just a URI

21:35:48 <mortenf> hmm

21:35:57 <JibberJim> forcing scutter distributers to do useful provenance checking and not just anonymous folk

21:36:25 <mortenf> well, in redland the context is just a node, so a pgp key id could do?

21:36:43 <edd> i always use a uri then use an RDF description about that uri

21:37:07 <mortenf> a local uri, and in the same model?

21:37:08 <edd> an invented uri that's unique for that fetch of that uri

21:37:13 <mortenf> :9

21:37:54 <edd> yeah, but it beats reification

21:38:04 <mortenf> oh, not bad at all

22:11:42 <JibberJim> 30,000 triples in Libby's scutterplan!

22:12:09 <mortenf> heh, i bet most of the triples in your store is from scutterplans :)

22:12:58 <JibberJim> Nah, I killed all mine from appearing

22:13:05 <JibberJim> So I think it's just libby's

22:17:03 <Mutiny> is there any way the common man can access these big scuttered foaf databases?

22:17:24 <JibberJim> You can download a dump

22:17:36 <JibberJim> libby has some sort of query engine - sbp knows about that.

22:17:38 <dajobe> what access would you want?

22:17:51 <JibberJim> What sort of queries do you want to run?

22:18:55 <Mutiny> yeah just downloading a dump would be cool

22:19:26 <JibberJim>http://jibbering.com/foaf/dumps/

22:19:51 <dajobe> I'll run a database or three if you like, and provide dumps

22:20:00 <dajobe> & some sort of web api

22:20:38 <dajobe> JibberJim: they seem at random times?

22:20:51 <dajobe> can you make a README preview before the 4M download?

22:22:54 <JibberJim> Yeah, they're completely random, just dumped there as and when, I should certainly create README... I'll add it to me list.

22:23:27 <dajobe> well tell me then, if i wget it, what do i get?

22:23:42 <JibberJim> You get a SQL dump of the foafnaut DB.

22:30:27 * edd almost done the necessary sysadminning to resurrect the mythical foafbot

22:30:52 <mortenf> nice