Experimental IRC log swig-2009-05-29

Available formats: content-negotiated html turtle (see SIOC for the vocabulary)

Back to channel and daily index: content-negotiated html turtle

These logs are provided as an experiment in indexing discussions using IRCHub.py, Irc2RDF.hs, and SIOC.

00:02:57<timbl>ACTION scrolls back only some of te stuff he has missed
00:33:15<timbl>Great .. BBC programs are now connegged, as per Tom H's tweet
00:33:19<timbl>http://www.bbc.co.uk/programmes/b0074fll#programme
00:34:12<timbl>A: New Earth, an episode of Series 2 of Dr Who [data link]
00:34:23<timbl>A:| New Earth, an episode of Series 2 of Dr Who [data link]
00:41:00<PovAddict>A1:""
01:11:09<timbl>A: not sure it works completely for [http://purl.org/ontology/po/episode|episode] a predicate
01:11:21<timbl>A: BBC++
07:21:07<mhausenblas>good morning Web of Data
07:24:46<melvster>morning, i noticed that digg has gone creative commons "zero" http://mashable.com/2009/05/28/digg-creative-commons-zero/ might be a good candidate for triplification?
07:36:57<CloCkWeRX>don't they render rdfa?
07:42:02<melvster>some, not sure how good it is: <html xmlns="http://www.w3.org/1999/xhtml" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:dcterms="http://purl.org/dc/terms/" xmlns:cc="http://creativecommons.org/ns#" xmlns:review="http://www.purl.org/stuff/rev#" xmlns:sioc="http://rdfs.org/sioc/ns#" xmlns:rdfs="http://www.w3.org/2000/01/rdf-schema#" >
07:44:48<melvster><div rel='dc:subject' resource='http://digg.com/design'><span property='rdfs:label' content='Design'/></div>
07:45:22<melvster><span rel="foaf:thumbnail">
07:45:37<melvster>rel="dc:source d1sJC4" property="dc:title"
07:45:54<melvster><span property="dcterms:abstract">
07:46:22<melvster><a rel="dc:creator" href="/users/gamebittk">
07:46:35<melvster><span property="dc:date"
07:48:49<melvster><strong id="diggs-strong-1" property="review:totalVotes">2832</strong>
08:21:37<bblfish_>hi
08:21:52<bblfish>hi
08:23:33<reto>hi hi
08:23:41<kwijibo>morning
08:29:54<bblfish>for those not following the social web xg I put up 5 user stories for foaf+ssl here http://www.w3.org/2005/Incubator/socialweb/wiki/UserStories
08:32:54<melvster>bblfish: i thought they were very good
08:34:12<bblfish>thanks melvster :-)
08:34:18<melvster>i thought "Linking to a remote friend" was a good example
08:34:39<melvster>i wonder how many networks allow linking to a remote frien
08:34:45<bblfish>none
08:34:48<melvster>s/frien/friend
08:34:50<bblfish>well foaf does
08:34:57<melvster>foaf does, xmpp does
08:35:02<bblfish>true
08:35:05<melvster>identi.ca does
08:35:24<melvster>you can cross link to a friend on another server
08:35:42<melvster>friendfeed is a grey area or not?
08:35:45<bblfish>using foaf probably?
08:36:23<melvster>identi.ca use foaf, not sure how polished it is, but in principle i think it works
08:36:51<melvster>noserub perhaps
08:37:07<melvster>id need to look more closely
08:37:15<bblfish>I need to look more at XMPP. I wonder how the linking to friends on different networks does it. Does it really allow you to do that?
08:37:48<melvster>sure, xmpp is great, gtalk uses it, so you can link from one of those accounts to a jabber account or one on your own server
08:38:42<melvster>the pidgin client has come a long way, you can use it as a meta client for, xmpp / aim / icq / msn / yahoo / irc etc. but also for facebook / myspace / skype -- in fact im using it now
08:40:21<bblfish>is it like Adium?
08:40:54<bblfish>I use that, but I seem to have had to create accounts in many different places
08:41:18<melvster>yes adium is the mac version
08:41:31<melvster>both run on libpurple which is the underlying engine
08:41:46<melvster>yes so most networks dont let you cross link, but xmpp does
08:42:09<melvster>google use it too, facebook say they want to use it
08:46:21<bblfish>melvster: I hope I was not too critical of the work of Dirk and Peter
08:46:49<bblfish>for the xmpp extension
08:47:10<bblfish>they are proposing http://www.ietf.org/internet-drafts/draft-meyer-xmpp-sasl-cert-management-01.txt
08:49:39<bblfish>(they did not reply to the post http://lists.foaf-project.org/pipermail/foaf-protocols/2009-May/000603.html )
08:51:59<melvster>i thought you made some good points, didnt come accross to me as overly critical, i spoke to dirk, he seems a really nice guy
08:53:21<bblfish>ok, just follow up with him. :-) It would be interesting for someone to prototype a xmpp server that just used foaf+ssl
08:53:28<bblfish>for identification
08:54:33<melvster>yes we'll keep in touch, ill try and get one working with XEP0257 but i think it's not ready yet, spark beta i think is working on it
08:55:03<drewp>it seems harder than usual to get tabulator to dive into my data. I can get to a class, but not the instances of that class
08:55:05<mhausenblas_>Wikier?
08:55:13<mhausenblas_>seen:
08:55:14<mhausenblas_>http://www.techcrunch.com/2009/05/28/sourceforge-acquires-open-source-data-community-ohloh/
08:55:26<bblfish>the fun thing would be to see if one could even make getting an xmpp account automatic with foaf+ssl, not even requiring a username
08:56:06<mhausenblas_>B:| SourceForge Acquires Open Source Data Community Ohloh
08:56:50<bblfish>and you could then perhaps even get your xmpp network set up automatically with all your foaf friends in there
09:00:33<melvster>bblfish: yes that's possible or vice versa, export your contacts into an access controlled FOAF
09:01:47<bblfish>yep makes sense to me. Then of course you could extend jabber to let you see foafs
09:03:27<melvster>yes i think so, they have a "roster" which seems to be a simple data structure
09:08:32<Wikier>mhausenblas_: tell me
09:08:44<Wikier>:-O
09:09:02<mhausenblas_>Wikier: well ...
09:09:21<Wikier>"It’s not clear of Ohloh will be absorbed into SourceForge.net. "
09:09:58<Wikier>B:wondering about how this will affect to RDFohloh
09:10:10<Wikier>B:because "It’s not clear of Ohloh will be absorbed into SourceForge.net."
09:10:34<Wikier>mhausenblas_: thanks for the pointer
09:14:10<mhausenblas_>yw, Wikier
09:16:33<timbl>drewp?
09:17:05<timbl>drewp, re "it seems harder than usual to get tabulator to dive into my data. I can get to a class, but not the instances of that class" -- check the folder icon for a view of a class which lists its instances.
09:17:17<timbl>for some reasonit is no longer the default.
09:18:43<drewp>yeah i wasn't familiar with the folder
09:19:22<drewp>i also get some predicates labeled 'Error' with object 'HTTP error 404 Not Found' that seem new
09:19:46<drewp>i guess they mean that tab'r tried to fetch my resource, and made an error edge on it to show me what happened
09:20:35<drewp>i am seeing all my stuff, at least. I added a foaf:primaryTopic to my graph. Didn't there used to be a "here are all the types we found in this document" thing?
09:22:22<drewp>aha 'mentions'. it's still there, but it only shows up if you also have a primaryTopic
09:24:24<reto>I'm at the requirement gathering metting of http://www.iks-project.eu/, currently brain-storming, on the storage layers we have jsr170, cmsis and rdbms as suggestions, not just triple-store
09:25:35<drewp>another Error edge: "Unhandled content type: image/png" :(
09:48:34<melvster>reto: just thinking out loud i wonder if anyone has used hadoop http://hadoop.apache.org/core/ for storing semantic data sets, i know facebook and yahoo use it, but it looks like that may be overkill in this case
09:50:46<mischat>melvster: i think this http://www.systap.com/bigdata.htm bigdata thing is based on hadoop
09:51:02<bblfish>I was just about to saz the same http://www.systap.com/bigdata.htm
09:51:03<mischat>i could be wrong
09:51:21<bblfish>ACTION austrian keyboard
09:51:29<mischat>last.fm do all of their recommeding off of hadoop too
09:52:10<melvster>thanks, didnt know that, cool
09:53:05<mischat>the last.fm founder blogs about hadoop and stuff http://www.metabrew.com/
09:58:27<mhausenblas_>AFAIK, sindice uses Hadoop as well - cygri?
09:59:37<kwijibo>timbl: what if one were to do <document> log:semantics :Dataset
10:00:27<kwijibo>where the dataset is actually a superset of the semantics of the document
10:01:28<kwijibo>ie, would :Dataset is log:semantics of <1.rdf>, <2.rdf> . be wrong ?
10:02:07<kwijibo>(and <1.rdf> and <2.rdf> contain different data)
11:11:12<timbl>kwijbo, <document> log:semantics :Dataset would be wrong. You could write <document> log:semantics :Dataset. log:includes [ is log:semantics of <document> ]log:semantics <document> log:semantics :Dataset
11:11:16<timbl>oops i mean
11:11:56<timbl>kwijbo, <document> log:semantics :Dataset would be wrong. You could write :Dataset. log:includes [ is log:semantics of <document> ].
11:13:31<timbl>A document's N3 semantics is deemed to be has has a precise set of triples (on a good day!) which you get from it.
11:26:02<kwijibo>thanks timbl
13:10:45<melvster1>hi all lets say ive got a bunch of (20) triples in a document, can i split the document into 2 halves, one 'contains' the first 10 triples, and one contains the 2nd 10, so that I can then be able to access control each half of the document by referencing a URI? eg lets say i want the first half public, and the 2nd half private, or is it better to have 2 documents?
13:16:18<libby>imo 2 docs...but bnodes would be a problem if split
13:16:29<melvster1>libby: thanks
13:17:12<melvster1>what im trying to do is for a FOAF file, some thing id like public, and others friend viewable or private
13:17:30<melvster1>but for web access control i need to refer the access control to a URI
13:19:23<libby>yeah nice idea that
13:19:35<libby>ACTION tried to do it with encrption for calendars once
13:21:20<melvster1>you mean for entries in a personal calendar, encrypted so that only people with the right key can read it?
13:23:02<melvster1>or i wonder if i can put a wrapper around each triple, in order to reference it, then the reference can go into the ACL file...
13:24:02<melvster1>does RDF:Description do something like that?
13:24:21<melvster1>i think ill go through the rdf book again :)
13:29:03<melvster1>seems not, "The RDF Description wraps one or more resource predicate/object pairs. "
13:35:38<Anchakor>melvster1: you mean reification?
13:36:35<Anchakor>I think the named graphs are better for this
13:38:12<ldodds>melvster1: you can have 2 RDF documents, each of which is a PersonalProfileDocument that describe you. And you can split statements across those documents how ever you like
13:39:26<Anchakor>so I would response to the original question: don't *split* the document, but copy the private triples into other graph which you reference with the access control rules, the server should handle the access control
13:39:51<melvster1>ok that makes sense thanks
13:40:29<melvster1>then in the access control doc i can have rules for "foaf_public" and "foaf_private"
13:58:11<timbl>melvster1, no you can't address individual triples. You can use N3 to have two separate graphs in the same file, and say different things about them. But then you will need a special web server of course to do the ACL and generate diffrerent files. and people will get confused as they woulkd see diff troiples for te same URI.
13:58:18<timbl>Two docs i sbest
13:58:56<melvster1>timbl: thanks, ill do that then
15:29:49<MacTed>timbl - hm. I think that getting different triples for the same URI is not necessarily a problem. consider --
15:29:49<MacTed>I belong to a group, the Rebel Alliance.
15:29:49<MacTed>I want to include that assertion in my FOAF when viewed by others who I assert to be in the Rebel Alliance, e.g., General Solo, but not when viewed by anonymous visitors (such as Count Duku).
15:29:49<MacTed>When Solo uses FOAF+SSL+OAuth and GETs my FOAF URI, he should see the triples that say I'm a Rebel, as is he.
15:29:49<MacTed>When Duku or others GET my FOAF URI, they should *not* see those triples.
15:29:50<MacTed>I should *not* need two FOAF URIs, one for Rebels and one for others.
15:29:52<MacTed>I *could* have two backing Named Graphs, which will of course have their own URIs, and invisibly blend them (or not) depending on the requesting user -- but this leaves your objection of confusion unanswered... as I think it should be.
15:32:48<drobilla>heh. example of the day
15:33:23<MacTed>:-)
15:35:52<Shepard>MacTed: you have to be aware that such personalisation destroys caching. I think it's not RESTful either
15:46:03<MacTed>oh, drat, I missed timbl's earlier /part. timbl -- please see http://chatlogs.planetrdf.com/swig/2009-05-29.html#T16-29-47
15:46:03<MacTed>Shepard - I'm not sure that the caching objection carries much (if any) weight; consider http://shop.example.com/my-cart which also cannot be cached, as it is personalized based on login cookies and other such invisible elements. I'm curious to hear your "not RESTful" reasoning, as such thinking has often proven incorrect.
15:50:41<Shepard>I'm just saying that it is not RESTful, not that this is something you should strive for :)
15:51:38<Shepard>(the RESTful shopping basket would be http://shop.example.com/basket/123 and might have access restrictions so that only the owner can see it - or not)
16:00:07<MacTed>so ... you're suggesting that a RESTful FOAF would be something like http://example.com/MacTed/foaf with a seeAlso triple pointing to http://example.com/MacTed/foaf/GeneralSolo which carries access restrictions? I think that fails to scale because I have to include seeAlso triples for every friend in the first file, which reveals that I *know* Solo to those who (again) shouldn't see those triples...
16:02:31<Shepard>if you actually want such fine-grained differences in your replies. but are you really willing to edit visibility for every person you know for every fact in your FOAF? normally you group users together
16:02:54<Shepard>then you never have to reveal who is in which group
16:03:08<Shepard>and you restrict access group-wise
16:15:04<MacTed>so ... I reveal RebelAlliance (or something) which you then know exists, and have reason to bang against to try to learn more -- because you're not allowed in -- which in the Empire is prima facie evidence I am a Rebel (even if it's really just a Family group, and I'm protecting my kid's bathtub picture)
16:15:51<PovAddict>example of the day indeed
16:16:04<MacTed>the goal is that I can say "allow Rebels to see this, that, and the other info" of course, not "allow Solo to see"
16:16:22<PovAddict>if you're protecting your kid's bathtub picture, the permission checking should be in the picture itself
16:16:30<MacTed>but I want that to happen based on engine-powered reasoning, not my own manual rule mongering
16:16:38<PovAddict>hiding the URI to the picture is pointless
16:16:58<MacTed>PovAddict - I want to protect *both* ways.
16:17:20<PovAddict>in the case of "don't show I'm in the RebelAlliance", obviously you need to hide the triple saying so
16:17:22<MacTed>PovAddict - security by obscurity fails on its own. but it augments security by permissions.
16:21:02<Shepard>why would you reveal rebel alliance? if there is a risk that someone might see the URI of that group then just name it "group123"
16:25:14<Shepard>I think you can't always avoid personalisation but you can keep it to a minimum. in REST you would have a few previously known entry points to the application (the public site) which might reveal links to resources that are only valid for that requester. so the fact that every requester gets different links is the personalisation
16:26:01<Shepard>the actual resources are not personalised really, you just get links to different ones depending on your request
16:26:22<PovAddict>in HTML it's hard to do that, since you usually want things like showing "you're logged in as <username>" and a logout link on every single page
16:26:37<Shepard>alternatively you can provide links to generic resources (http://shop.example.com/my-cart) which then redirect to your personal resource
16:26:44<PovAddict>(I'd love to use HTTP Basic Auth but browser UI for that generally sucks)
16:26:59<Shepard>PovAddict: yes, authentication is a problem
16:27:32<PovAddict>what about read/unread icons next to every thread title on a forum page?
16:28:15<PovAddict>(no, client-side-tracked visited/unvisited links won't do :P)
16:29:09<Shepard>obviously for user-facing resources there are some requirements where usability is more important than caching
16:29:25<Shepard>but we were talking about FOAF files, no? :)
16:29:31<PovAddict>yep :)
16:31:26<Shepard>and regarding the engine-powered reasoning as opposed to manual editing of rights: yes, you can still do that, even in a RESTful way. you just provide links to dynamic resources which calculate the result based on the social graph instead of explict group assignments for example
16:31:44<Shepard>the idea is just to maximise cacheability as far as possible
16:46:32<drobilla>I'm not a fan of webdav /requiring/ this xml property format at all...
16:46:59<drobilla>propfind and friends would work fine with RDF given the appropriate content type headers etc.
16:57:23<mhausenblas>off for ESWC09 now - hope to see some of you there (and that we have a network that works, this time ;)
16:57:34<mhausenblas>good nite Web of Data
18:40:03<PovAddict>"As a parent, you're allowed to kill your children. If you do not wait for your children they'll become zombies and if you die, your children will be adopted by someone else. Maybe you should finger the user before making a connection. UNIX is full of great metaphors and such..."
18:44:14<gsnedders>Oh that is lovely…
18:44:44<PovAddict>lol
18:45:09<PovAddict>I recall a similar thing about C++
18:45:19<PovAddict>"in C++, your friends can access your privates"
21:07:32<ericP>what's the word we RDF'ers use to describe access to "old" data technologies? fits in "access to @@ data", and it's a little less pajorative than the word "vestigial".
21:08:04<melvster1>pushback used the term 'legacy'
21:08:44<ericP>exactly, tx!
21:09:13<PovAddict>use "prehistoric" or "jurasic" if you talk about mainframes
21:13:34<ericP>i think i'll go with "jurasic"
22:26:48<dajobe>http://github.com/dajobe/turtle/tree/master
22:27:04<dajobe>C:|GIT repository of historical turtle rdf spec development
22:27:12<dajobe>C:converted RCS->CVS->GIT
22:27:30<PovAddict>RCS?!
22:27:32<dajobe>C:might be of interest to people
22:27:42<dajobe>sure, perfectly fine for VCing single files
22:28:25<PovAddict>ancient stuff :)
22:28:42<dajobe>so are the most stable and reliable parts of many systems
22:28:58<dajobe>I'm still using RCS, CVS, SVN and GIT for different things
22:32:57<armenb_>join ##swig
22:33:00<armenb_>er, sorry.
22:35:46<armenb>is there a channel for the interface generator?
22:36:02<PovAddict>there might not be
22:36:11<armenb>nod. ok.
22:36:13<armenb>thanks
22:36:22<PovAddict>that was as vague as I could do

Back to channel and daily index: content-negotiated html turtle