Semantic Web Interest Group IRC Chat Logs for 2003-12-03

00:44:47 <MorbusIff> MorbusIff is now known as MorbusIffOnly

00:46:15 <MorbusIffOnly> MorbusIffOnly is now known as Morbus

00:56:26 <mdupont> mdupont is now known as ZZzzZz-md-ZzzZz

01:43:04 <Davey> Davey is now known as D[a]vey

06:23:34 <eikeon_> eikeon_ is now known as eikeon

08:05:57 * DanC calls it a night after a bit of work on the wiki

08:06:31 <ericP> night

09:17:06 <ZZzzZz-md-ZzzZz> ZZzzZz-md-ZzzZz is now known as mdupont

09:21:34 <mdupont> mdupont is now known as md-zzz

10:02:46 <swh> swh is now known as swh_away

10:34:34 <md-zzz> md-zzz is now known as mdupont

11:10:25 <__andrew__> i been talking to a few people about a java rdf api

11:10:53 <__andrew__> as i've been trying to integrate kowari and jena

11:11:09 <dajobe> you better give a pointer to kowari as jena is much better known

11:11:27 <__andrew__> :-) umm...the download page?

11:11:41 <__andrew__> there's no real online doco for it yet...it's in cvs on sf...

11:11:42 <dajobe> which is what?

11:11:49 <__andrew__> ahh

11:12:07 <__andrew__>http://www.pisoftware.com/cgi-bin/kowari.pl

11:12:08 <dc_rdfig> A: http://www.pisoftware.com/cgi-bin/kowari.pl from __andrew__

11:12:17 <__andrew__> It's been noted that the registeration page puts people up

11:12:33 <dajobe> I typically stop at such things and go elsewhere.

11:12:40 <__andrew__> I don't know when it's going to be put up properly on SF which is http://sf.net/projects/kowari

11:12:48 <__andrew__> Me too

11:13:09 <__andrew__> But that sf project is no more useful than metadatabase

11:13:11 <dajobe> until then, it's not something high on my list of things to do

11:13:19 <dajobe> sorry

11:13:27 <__andrew__> Well the common RDF API is going to be separate from Kowari

11:13:47 <__andrew__> What I want is for Kowari, Jena, Sesame and any other Java RDF project to use it...much like the w3c XML APIs

11:14:16 <__andrew__> it's going to be under apache license too not mpl

11:14:28 <__andrew__> which should be a bit more friendly to developers

11:14:29 <dajobe> jena and sesame are the two major well known, open and deployed java apis (3-4 years) so I think they would have the userbase & experience

11:14:32 <swh_away> swh_away is now known as swh

11:14:44 <__andrew__> your right...

11:14:54 <dajobe> (there are other older java apis that I'm aware of but not significant for this chat)

11:15:32 <__andrew__> but even within sesame and jena there are different ways to create triples or different ways to get various properties out

11:16:15 <__andrew__> apparently there's been an idea that integration between jena and sesame would be a good thing but no one has time to do it

11:16:38 <dajobe> yes, that came up in the last week

11:16:53 <__andrew__> while i don't have the time to do a mapping between sesame and jena i have been given the task for kowari and jena and was hoping that by also keeping sesame in mind it would mean it could happen easily at a later date

11:17:19 <__andrew__> i personally think sesame is more correct

11:17:34 <dajobe>http://sourceforge.net/mailarchive/forum.php?thread_id=3545907&forum_id=8988

11:17:34 <dc_rdfig> B: http://sourceforge.net/mailarchive/forum.php?thread_id=3545907&forum_id=8988 from dajobe

11:17:58 <dajobe> B:|Jena and Sesame RDF java apis discussion, initial ideas

11:17:58 <dc_rdfig> Titled item B.

11:19:24 <dajobe> if you have the time, why don't you write up something (don't actually design it, do it) but consider what are pros & cons of approaches

11:19:42 <dajobe> i.e. expand on why "<__andrew__> i personally think sesame is more correct "

11:19:53 <__andrew__> also here : http://sourceforge.net/mailarchive/forum.php?thread_id=3545843&forum_id=9004

11:20:24 <dajobe> logger, chump b

11:20:24 <dajobe> B:See [http://ilrt.org/discovery/chatlogs/rdfig/2003-12-03#T11-20-24|discussion]

11:20:24 <dc_rdfig> Added comment B1.

11:20:59 <__andrew__> well sesame actually follows the current spec for a start

11:21:24 <dajobe> BLURB:Idea - RDF Java API scheduled chat

11:21:25 <dc_rdfig> C: Idea - RDF Java API scheduled chat from dajobe

11:21:25 <__andrew__> jena's graph api uses node, node, node to create a triple

11:21:55 <dajobe> C:this might be nice to organise on #rdfig soon. Signup if you'd like to attend.

11:21:55 <dc_rdfig> Added comment C1.

11:22:15 <dajobe> and what's wrong with that?

11:22:43 <__andrew__> umm...well it should be resource (bnode/uri ref), uri ref, node

11:23:05 <dajobe> oh that

11:23:26 <dajobe> It is probably checked somewhere

11:23:49 <__andrew__> yeah it is

11:24:23 <__andrew__> it just not as clean...if it was cleaner then it could be integrated more cleanly with kowari and probably sesame

11:24:45 <dajobe> that's not necessarily the case - "clean" is your design choice

11:24:56 <dajobe> it might be considered cleaner to have a uniform triple

11:25:30 <__andrew__> you're right...i'd like the rdf spec to be more uniform if that's the case

11:26:00 <__andrew__> i'm sure there's a good reason why subject and object aren't the same for example

11:26:00 <dajobe> the WG considered it but weren't going to change the rdf spec (weren't allowed in this case)

11:26:15 <dajobe> you can't write it in the syntax

11:26:28 <dajobe> it = literals as subjects

11:26:35 <dajobe> read the WG's postponed issue list

11:26:55 <__andrew__> yeah i remember there was a good reason :-)

11:27:32 <dajobe> and the other case is bnodes & literals as predicates

11:27:34 <__andrew__> so do you think a java binding is a good idea?

11:27:57 <dajobe> jena and sesame already provide that (amongst others)

11:28:04 <__andrew__> something for the w3c or something for developers to thrash out more

11:28:54 <__andrew__> like i said they are different and there are differences that can't be solved just by mapping one object model to another

11:29:26 <dajobe> can you please write down what those differences are?

11:29:39 <dajobe> in some document with paragraphs? pretty please?

11:29:45 <__andrew__> for example, in kowari most things are just a proxy around a disk representation...if you mapped them to jena objects you'd lose the ability to handle large objects

11:29:46 <__andrew__> ok

11:29:52 <__andrew__> i'll thrash it out properly

11:30:08 <__andrew__> document wise...

11:30:18 <dajobe> or a wiki page if you can do that

11:31:41 <__andrew__> are you aware of the rationale behind the w3c's xml api? was it just an industry/corporate push?

11:32:04 <dajobe> which xml api do you mean? aren't there ~5+ ?

11:32:14 <dajobe> but to answer that Q, no.

11:32:19 <__andrew__> :-)

11:32:55 <__andrew__> the problem is most of the rational i can think of are programming ones...

11:34:02 <dajobe> I still don't know which xml api you are thinking of.

11:34:24 <dajobe> XPath/XQuery for example isn't really very programatic

11:34:58 <__andrew__> dom

11:35:20 <__andrew__> org.w3c.dom...an then you have 3rd parties implementing around it

11:35:23 <dajobe> that originally (rationale?) came from web browsers

11:35:49 <__andrew__> so that rationale grew from javascript and then java

11:35:57 <dajobe> no idea

11:48:24 <__andrew__> what about this for why a standard is important: http://www.nitot.com/mozilla/portugal-join2003/slide19.html

11:49:35 <dajobe> the DOM (AFAIK) is one of the very few W3C software apis

11:49:58 <dajobe> so it clearly hasn't been a problem for, say, XML schema, xslt, or the other techs

11:50:14 <__andrew__> yeah but isn't it the same argument...you want to take your application written for jena and move it to sesame

11:50:23 <dajobe> that's just YA "why standards are good" talk

11:50:41 <dajobe> that's not something people actually ask

11:50:54 <dajobe> they ask "what can I use for RDF in Java?"

11:50:59 <dajobe> or "what is the RDF java API?"

11:51:04 <darobin_> a cool thing with having a WG to design an API is that you get more smart people working on the same thing than you would otherwise

11:51:04 <__andrew__> well the sesame/jena integration was about sharing data between the two

11:51:08 <darobin_> at least, you hope you do :)

11:51:36 <dajobe> I'm not one to judge if the DOM(s) APIs are a success, I don't work in that area and haven't use them.

11:51:49 <darobin_> the more I look at the DOM, the less I see other ways of getting it "right" ( for a tree based view of course)

11:52:32 <dajobe> but it doesn't match other views which is why SAX1, SAX2, XmlReader exist for example

11:52:34 <__andrew__> so you see the w3c defining protocols and data formats rather than programming apis

11:53:08 <dajobe> me? I'm just saying they haven't done a lot of software apis compared to formats and other things.

11:53:10 <__andrew__> the way i see rdf as a programmer is similar to databases...maybe that's just coloured by my world view

11:53:11 <darobin_> dajobe: I find that normal, designing one API for all uses would be impossible

11:53:25 <__andrew__> having jdbc for java is a good thing as a programmer to use databases

11:53:44 <dajobe> in that case, the work on a SQL-style query interface is on track to how the SQL world deals with cross-langauge issues.

11:54:01 <__andrew__> but it's just not querying...transactions for example

11:54:05 <dajobe> a graph network/web services api.

11:54:19 <__andrew__> sesame, jena and kowari all have 3 different apis for transactions

11:54:20 <dajobe> also being thought about, joseki, rdf net api, TAP, sesame's

11:58:06 <__andrew__> i'm not sure everything needs the overhead of a network/web services api sometimes you just want to get a statement out of a graph

11:58:27 <__andrew__> completely locally

11:58:34 <dajobe> and at that level I bet they have exactly the same api

11:59:22 <__andrew__> the api for parsing rdf is different to the forward facing/user api which is different again to how the triples are stored

11:59:43 <__andrew__> take a look at jena's model and graph apis...they are different although very close to being the same

11:59:53 <dajobe> there are reasons they made that change

11:59:57 <dajobe> graph is an internal SPI

12:02:36 <__andrew__> actually people use the graph...especially for integrating different stores

12:08:41 <__andrew__> cyas

18:05:35 Topic now Semantic Web hack-n-chat http://rdfig.xmlhack.com/ latest logs http://ilrt.org/discovery/chatlogs/rdfig/latest wiki http://esw.w3.org/topic/FrontPage

18:05:35 Users on #rdfig: logger dajobe AndyS larsbot eaon|away eikeon ericm timbl earle Wack roGer^work iwaim burtonator jql md-afk Arnia josek darobin_ LotR anselm2 shellac chrisc swh arnarl MarkB xover sbp eikco tansaku datum GabeW jordan sh1mmer kasei jmb D[a]vey setre mmealling grove danbri dc_rdfig karlcow CaptSolo dngor kota DanC deltab workbench

18:05:35 <NickServ> This nickname is owned by someone else

18:05:35 <NickServ> If this is your nickname, type /msg NickServ IDENTIFY <password>

18:05:35 <NickServ> Password accepted - you are now recognized

18:05:35 <MemoServ> You have no new memos

18:05:35 <ChanServ> [#rdfig] This channel is logged and blogged: http://logicerror.com/rdfIRCWelcome

18:07:10 <dajobe> B:I think andrew (newman) was suggesting that there are problems connecting the apis. I was asking him for detail.

18:07:11 <dc_rdfig> Added comment B5.