IRC log of dig on 2012-11-07

Timestamps are in UTC.

00:27:04 [melvster1]
melvster1 has quit (Ping timeout: 252 seconds)
03:40:42 [kennyluck]
kennyluck has quit (Read error: Connection reset by peer)
03:41:36 [kennyluck]
kennyluck (~kennyluck@119.161.158.96) has joined #dig
05:31:06 [timbl]
timbl has quit (Read error: Connection reset by peer)
05:32:00 [timbl]
timbl (~timbl@c-24-62-225-11.hsd1.ma.comcast.net) has joined #dig
07:50:21 [rszeno]
rszeno (~rszeno@79.114.103.154) has joined #dig
08:11:15 [tyteen4a03]
tyteen4a03 has quit (Read error: Connection reset by peer)
08:11:33 [tyteen4a03]
tyteen4a03 (tyteen4a03@n218250225173.netvigator.com) has joined #dig
08:41:54 [jmvanel]
jmvanel (~jmv@82.96.114.78.rev.sfr.net) has joined #dig
08:55:48 [kennyluck]
kennyluck has quit (Read error: Connection reset by peer)
08:56:06 [kennyluck]
kennyluck (~kennyluck@119.161.158.96) has joined #dig
11:01:18 [melvster]
melvster (~melvin@p4FF97A59.dip.t-dialin.net) has joined #dig
11:04:46 [melvster1]
melvster1 (~melvin@p4FF9629C.dip.t-dialin.net) has joined #dig
11:05:24 [melvster]
melvster has quit (Ping timeout: 248 seconds)
11:25:45 [webr3]
timbl, fyi: http://lists.w3.org/Archives/Public/public-rdf-wg/2012Nov/0079.html and http://lists.w3.org/Archives/Public/public-rdf-wg/2012Nov/0043.html
11:26:27 [webr3]
(push for triple ::= term term term . term ::= IRI | blanknode | literal .) in rdf
11:40:50 [danbri]
danbri (~danbri@80.71.30.113) has joined #dig
12:24:27 [RalphS]
RalphS (RSwick@30-7-118.wireless.csail.mit.edu) has joined #dig
12:38:48 [bblfish]
webr3: I wonder what a Literal in predicate position would mean, but otherwise it is interesting that they are looking at this
12:39:09 [bblfish]
betehess: would find that interesting - looks like Banana is ahead of time
12:43:48 [bblfish]
mind you webr3 it looks like they can't do anything as they are restricted by their charter
12:46:38 [bblfish]
perhaps they should change their charter
13:01:09 [jmvanel]
jmvanel has quit (Remote host closed the connection)
13:18:16 [rszeno]
bblfish, http://www.w3.org/TR/rdf-mt/#literalnote
13:53:42 [webr3]
bblfish, it's only for inference rules, non normative, and in a note.. i.e. recognize they (generalized triples) are needed and need to be mentioned somwhere, but don't actually allow them anywhere over the wire / in a media type / abstract syntax etc
13:54:14 [bblfish]
ah ok.
13:54:17 [bblfish]
makes sense.
13:54:28 [bblfish]
yes, for inference you definintely need them
13:54:42 [webr3]
bblfish, PatH made an interesting observation the other day, (roughly) that in all the cases he looked at through inference, bnode+iri+literal ended up in subject and object positions, and only bnodes+iris ended up in p
13:54:44 [bblfish]
that's why betehess' Banana RDF allows it
13:55:05 [bblfish]
yes I'd agree :-)
13:55:07 [webr3]
likewise why the rdf-interfaces and my own tools allow it
13:55:13 [bblfish]
literals in p does not make sense
13:55:42 [bblfish]
well I think it might in the future when we can work out clear RDF semantics for natural languages
13:55:45 [webr3]
unsure if that's a reason to preclude them, but in subject and bnode in predicate would make a world of difference
13:55:55 [bblfish]
"knows"@en == foaf:knows
13:56:24 [bblfish]
( but that would be very shaky)
13:56:47 [betehess]
bblfish: we have a URI in the predicate position https://github.com/w3c/banana-rdf/blob/master/rdf/src/main/scala/RDFOps.scala#L26
13:56:52 [webr3]
you can do many shay things in rdf, including { foaf:knows owl:sameAs "knows"@en } - it's already valid
13:57:18 [bblfish]
yes, but it's not clear it is true.
13:57:25 [betehess]
not having a URI here makes the fluent API much much more difficult to use
13:57:32 [betehess]
hey webr3
13:58:17 [bblfish]
you'd need clear agreement from linguists for something like foaf:knows owl:sameAs "knows"@en to be more than a hypothesis
13:58:21 [betehess]
but we could relax that, the same way we relaxed the condition on subjects...
13:58:37 [betehess]
I'd prefer the RDF guys to focus on cleaning up literals
13:58:46 [betehess]
whatever
13:59:22 [deiu]
deiu (~andrei@2001:470:8b2d:7d4:706f:c5d7:33c5:fca9) has joined #dig
13:59:22 [deiu]
deiu has quit (Changing host)
13:59:22 [deiu]
deiu (~andrei@unaffiliated/deiu) has joined #dig
13:59:31 [webr3]
n3-2+rules may be very interesting to do :) looking forward to having a stab at that one day
13:59:49 [bblfish]
betehess, timbl: I just wrote a (very primitive) proxy for w3c with WebID authentication using WAC ontology
14:00:37 [bblfish]
So I have a localhost play server running and I can give WebID authenticated users access to subparts of W3C web sites ( that are public )
14:01:15 [bblfish]
Now all one would need is to be able to map real users to WebID and do some interesting demos
14:02:03 [bblfish]
I mean if I could give my server the password of real w3c users then I should be able to authenticate them autmoatically using existing system, when they use WebID.
14:02:20 [bblfish]
one could then do more advanced things too...
14:02:34 [timbl]
timbl has quit (Read error: Operation timed out)
14:05:52 [bblfish]
It's kind of an interesting way to do demos for existing sites, of how they could look like with WebID enabled...
14:07:41 [deiu]
bblfish, is the meeting set for Friday?
14:08:05 [bblfish]
I am just trying to get zakim telecom slot now
14:09:10 [bblfish]
this was the time I asked for http://www.timeanddate.com/worldclock/fixedtime.html?msg=WebID%2FRWW+teleconf&iso=20121109T16&p1=195&ah=1
14:09:24 [bblfish]
our previous meeting is now slashing with LDP groups meeting slot
14:09:33 [bblfish]
s/slashing/clashing/
14:09:40 [bblfish]
DIGlogger: pointer?
14:09:40 [bblfish]
See http://dig.csail.mit.edu/irc/dig/2012-11-07#T14-09-40
14:10:21 [bblfish]
mh, diglogger does not correct s///
14:10:52 [bblfish]
is that ok deiu for you?
14:10:57 [deiu]
Yeah
14:11:17 [bblfish]
how was the conf in Boston?
14:11:34 [deiu]
the conf?
14:11:48 [bblfish]
was it Semtech you were going to?
14:11:50 [deiu]
ISWC is next week, if that's what you mean
14:12:07 [bblfish]
ah ok
14:12:57 [bblfish]
betehess: don't know if you saw but there was a lot of pushback on WebID list to some aspects of the new definition of WebID at TPAC
14:13:22 [bblfish]
so I sent you an updated version that I could get some agreement on
14:13:36 [betehess]
I'm afraid of what I'll see
14:13:46 [betehess]
not sure I want to loose time there
14:13:47 [bblfish]
( sent that to the WebID list too of course )
14:13:51 [betehess]
sounds very unproductive to me
14:14:24 [betehess]
I don't have the brain-cycle to process the noise introduced by some
14:14:24 [bblfish]
gaining consensus is a complicated task.
14:15:14 [betehess]
it's consensus to please one very lousy guy
14:15:30 [betehess]
anyway, don't have time today for that
14:16:08 [bblfish]
There were a few more really I think. Anyway I think the general gist is going in your direction.
14:17:14 [bblfish]
if you check the e-mail you'll see it's moving the right way.
14:18:29 [betehess]
link?
14:18:55 [bblfish]
http://lists.w3.org/Archives/Public/public-webid/2012Nov/0020.html
14:19:10 [deiu]
bblfish, I'm going to meet kingsley next week
14:19:18 [deiu]
maybe I can talk some sense into him
14:19:35 [bblfish]
he agrees with the points in mail above
14:20:01 [bblfish]
perhaps if you come to teleconf this Friday we can widen the consensus on this http://www.timeanddate.com/worldclock/fixedtime.html?msg=WebID%2FRWW+teleconf&iso=20121109T16&p1=195&ah=1
14:20:51 [deiu]
bblfish, so that's the final hour?
14:20:56 [betehess]
will kingsley be at the conference?
14:21:04 [betehess]
if not, then it's useless
14:21:10 [deiu]
there's a different meetup
14:21:29 [deiu]
he's organizing it
14:21:33 [betehess]
deiu, I think I'll go to this meetup
14:21:42 [bblfish]
which meetup?
14:21:43 [deiu]
same
14:21:48 [deiu]
http://www.meetup.com/The-Cambridge-Semantic-Web-Meetup-Group/events/78412762/
14:22:14 [betehess]
http://www.meetup.com/The-Cambridge-Semantic-Web-Meetup-Group/events/78412762/
14:22:17 [betehess]
heh
14:22:20 [bblfish]
ah good. Well he has one of OpenLink's people who is there.
14:22:21 [deiu]
betehess, good, otherwise I'd be going alone into the wolf's den
14:22:33 [bblfish]
I'll see if I can nudge kingsley into coming.
14:26:19 [danbri]
the new definition didn't seem to have a lot to do with SSL
14:26:36 [danbri]
rather, it's a page whose owner can demonstrate control
14:26:46 [danbri]
(via various techniques)
14:26:53 [jmvanel]
jmvanel (~jmvanel@82.96.114.78.rev.sfr.net) has joined #dig
14:27:57 [betehess]
danbri, if it's the case, then it's good
14:27:59 [bblfish]
that's the point danbri. We are trying to seperate the idea of a WebID from SSL
14:28:14 [betehess]
yes, Identity vs Authentication
14:28:24 [danbri]
If I understand right (re new direction), that's where I was heading with http://web.archiveorange.com/archive/v/egAMuIBH6WRojMBKNlOj 'tell-me show-me auth', i.e. a page makes a claim that can be demonstrated
14:28:43 [danbri]
http://microformats.org/wiki/RelMeAuth too
14:29:11 [betehess]
danbri, you were ahead of your time, and it's still very difficult to explain why this is important
14:29:36 [danbri]
WebID is a very nice general name, to 'use up' on a very specific (useful but universally appropriate) protocol technique
14:29:46 [danbri]
are people worried we're 'watering it down' to mean something super-general?
14:30:05 [bblfish]
that's the danger.
14:30:35 [bblfish]
We need to rename the WebID protocol to WebID Authentication over TLS
14:30:49 [danbri]
WIDAOT!
14:31:11 [danbri]
"Don't be without WIDAOT!"? Not quite so catchy :) but audience is engineers
14:31:53 [betehess]
WIDAOT :-)
14:33:22 [bblfish]
the nice thing is that we can claim to have 1 billion WebIDs
14:33:32 [bblfish]
but the danger is that people can say it's useless
14:33:50 [webr3]
do not say anything, just show it working
14:34:09 [danbri]
what'd make it seem more concrete, would be software tools that exploit the common abstraction
14:34:51 [danbri]
Would 'paste this random string in your page and tell us when you did, so we can check it' work?
14:35:13 [danbri]
(and the page might be a public access wiki page... that's another level of checks to do)
14:35:58 [bblfish]
yes, though then you have people moving away from point and click to pasting and strings, which is more prone to error
14:36:08 [bblfish]
and requires more education
14:36:43 [bblfish]
perhaps we can just reduce WebID over TLS Auth to WebID Auth
14:37:00 [bblfish]
ofter all the others are well known OpenId Auth, or OAuth
14:40:17 [bblfish]
The other thing we could do is create a Web Identity Interoperability spec, which shows how the different identity systems can be interlinked
14:40:24 [bblfish]
using logic
14:41:48 [danbri]
but who would the interlinking benefit?
14:41:58 [bblfish]
all of us
14:42:02 [danbri]
will it make it easier for users, developers, businessy types etc somehow?
14:42:10 [danbri]
through common software? ux?
14:42:12 [bblfish]
because i could use an OpenId to login on sites that don't have WebID
14:42:27 [danbri]
you can do that anyway
14:42:28 [bblfish]
or with devices I don't want to install a certificate on
14:42:32 [danbri]
using OpenId
14:42:42 [danbri]
that's already possible
14:42:49 [bblfish]
ok, vice versa then
14:42:52 [danbri]
it might look like a turf grab
14:42:59 [bblfish]
no this is just logic
14:43:11 [bblfish]
showing how all these systems can tie into the same ACLs
14:43:12 [danbri]
well, just a prediction
14:43:38 [danbri]
but if you go out there and effectively say OpenID 2 or OAuth 2 are basically WebID, ...
14:43:52 [danbri]
...without showing some strong value for the rebranding, it'll likely get a lot of pushback
14:43:56 [danbri]
logic or no logic :)
14:44:04 [bblfish]
agree. I argued that at TPAC
14:44:12 [bblfish]
so that would be a different document
14:44:40 [danbri]
if there is software that shows why calling these things by a common name reflects common implemented tooling, that's vastly more impressive
14:44:48 [bblfish]
" Web Identity Interoperability spec" would show how given an ACL with an OpenId it can be verified if all we know is that someone has a WebID
14:47:20 [bblfish]
or if the ACL restricts users to people with an foaf:mbox, then we can show how using perhaps WebFinger you can go from a WebID to an mbox principal
14:47:33 [bblfish]
etc. etc.
14:47:59 [bblfish]
that then shows that authentication is one thing and authorization another one.
14:49:39 [bblfish]
I find it difficult to see which of these paths is the right one to follow.
14:49:41 [betehess]
danbri, the point of the new definition was to be able to rely directly on LDP
14:50:18 [danbri]
what's in LDP that makes this an attractive goal?
14:50:28 [betehess]
read-write ?
14:50:35 [betehess]
containers?
14:50:47 [betehess]
RDF everywhere?
14:51:03 [betehess]
standards for all the above?
14:55:53 [bblfish]
The WebID as the name of the URL that identifies someone seems like a good idea. Having an (HTTP(s)) URL one can do LDP work also seems right, because that is one that can be linked to - creating linked data.
14:56:41 [bblfish]
At the same time saying that you can have a WebID for me is a bit weird.
14:56:59 [danbri]
who amongst potential WebID adopters are those things appealing to?
14:57:02 [bblfish]
Usually an ID is something that I possess
14:57:18 [danbri]
14:56 bblfish: At the same time saying that you can have a WebID for me is a bit weird.
14:57:26 [danbri]
yeah, you put your finger on one of the tricky things
14:57:29 [betehess]
danbri, it's appealing to me because it gives me a platform to relies on
14:58:14 [danbri]
so you might expect in a few years that there might be Apache or IBM or whatever LDP implementations 'out of the box' doing much of what you need for an WebID-centric site?
14:58:37 [betehess]
not sure I understand
14:58:47 [danbri]
what kind of numbers are we talking about? thousands/millions/billions? (of users...)
14:59:05 [danbri]
well, a W3C spec is not a 'platform to rely on' directly; you need software too
14:59:08 [betehess]
well, I hope it's "all the Web"
14:59:17 [bblfish]
+1
14:59:21 [danbri]
but the hope is that LDP-compatible software will be very useful for WebID
14:59:27 [betehess]
yes, but I don't see the point of this remark
14:59:30 [danbri]
I mean in one installation
14:59:31 [betehess]
yes!
14:59:50 [betehess]
that's what I meant by leveraging LDP
15:00:02 [kennyluck]
kennyluck has quit (Quit: kennyluck)
15:00:07 [danbri]
ok, just trying to get to the bottom of the appeal
15:00:09 [betehess]
the fact that there is a WG says a lot
15:00:37 [betehess]
and we have already some implementations, willing to comply with the LDP spec
15:03:06 [timbl]
timbl (~timbl@30-5-244.wireless.csail.mit.edu) has joined #dig
15:44:36 [scor]
scor (~scor@w0052335.mgh.harvard.edu) has joined #dig
15:44:37 [scor]
scor has quit (Changing host)
15:44:37 [scor]
scor (~scor@drupal.org/user/52142/view) has joined #dig
16:01:31 [bblfish]
people who were present at TPAC http://www.w3.org/2005/Incubator/webid/tpac/group.n3
16:03:06 [deiu]
cool stuff bblfish :)
16:03:13 [bblfish]
I must be missing some still...
16:41:48 [Snuupy]
Snuupy has quit (Excess Flood)
16:43:11 [Snuupy]
Snuupy (~Snuupy@2607:5300:20:201::1:3077) has joined #dig
16:45:00 [Snuupy]
Snuupy has quit (Changing host)
16:45:00 [Snuupy]
Snuupy (~Snuupy@unaffiliated/snuupy) has joined #dig
16:45:29 [melvster1]
danbri: I dont mind what it's called so long as it's not 'WebID -- The Living Standard' -- because that would spell WebID TLS :)
16:46:38 [melvster1]
danbrI: yes I think the idea was to split into two specs, one concerned with identity (how to identify someone on the web) one with authentication, which can leverage SSL for example
16:48:19 [melvster1]
danbri: Im not sure you can say OpenID 2 or OAuth2 are webid ... how do they define a user or agent? is that even clear?
16:48:39 [betehess]
melvster1: I didn't pay attention to the new proposals. do you believe that they are still ok with this goal?
16:48:44 [danbri]
TLS :)
16:49:15 [danbri]
have a look at relmeauth
16:50:33 [melvster1]
betehess: first of all i think all the proposals we have are both OK, and an improvement, in so far as we've decoupled webid from having a public key and that is deferred now to webid auth ... I personally think it's all good ... I know exactly the technology i need to get things done and the properties, namely URIs, linked data, opacity, universality, tolerance, independent invention ... the axioms of the web are the same ... naming i can live w
16:55:30 [melvster1]
danbri: I think all the identity systems inc. OpenID, OAuth, WebID, Persona, SAML etc. should consider that identity has 3 parts ... identification / authentication / authorization ... a good system will have a clean modular design and a separation of concerns between these three concepts, and even more importantly the key terms should be clearly defined
16:56:34 [melvster1]
there are examples of identity that have no authentication by default, three I can think of are, the postal service, the telephone and email
16:57:56 [melvster1]
not saying any of those 3 are perfect, or even good, but a well designed system tends to be clean and modular, loosely coupled
17:00:09 [bblfish]
+1
17:05:59 [danbri]
what would it take, for folk here to consider http://en.wikipedia.org/wiki/Stephen_Fry a WebID for the person it describes?
17:06:47 [danbri]
note e.g. that it informally describes his twitter account, http://en.wikipedia.org/wiki/Stephen_Fry#Twitter ... which is something you could check control of
17:06:59 [melvster1]
he's on dbpedia too ;)
17:07:18 [mattl]
and that's a photo of him i added :P
17:09:09 [melvster1]
thankfully twitter have dispensed with hash bangs https://twitter.com/stephenfry
17:11:15 [scor]
danbri: are you considering the http-range-14 here? would be nice to align our design on the one of schema.org while we're at it
17:12:33 [melvster1]
danbri: I think the advice you gave facebook, ie to put a # on the end in the turtle, was very good and pragmatic
17:12:46 [scor]
OpenID doesn't care about differentiating the peron and the page describing the person for example
17:13:48 [melvster1]
scor: the web has a principle of tolerance ... if you can more or less determine what someone means you dont have to be too strict ... but to avoid ambiguity people should be encouraged to follow standards where possible
17:17:28 [melvster1]
it's often advantageous to be able to split a page up into multiple parts each with a subject, e.g. if you want to get a whole schema with a single GET, or if you have a number of financial transactions in, say, the form of a statement
17:17:59 [melvster1]
the one subject per page is an example of clean design for the most part, but there are certainly cases where that is not efficient, imho
17:22:05 [bblfish]
danbri: I think if the wikipedia page is a WebID for Stephen Fry, then we have really lost all sense of this being something one can use to authenticate with.
17:22:36 [bblfish]
then one can wonder why one should create a special class of Agent URIs at all
17:22:54 [bblfish]
anyway here is the teleconf for WebID/RWW/SocialWeb http://lists.w3.org/Archives/Public/public-webid/2012Nov/0023.html
17:23:08 [bblfish]
for the after TPAC meeting
17:24:31 [bblfish]
This friday 15:00-16:00 UTC
17:29:21 [bblfish]
anyway those are tricky issues
18:17:40 [danbri]
bblfish; what if the claim within the wikipedia page that http://en.wikipedia.org/wiki/Stephen_Fry#it has :account https://twitter.com/stephenfry was robust (by community review processes, etc.) against malicious changes?
18:18:09 [danbri]
then the person it describes - namely Stephen Fry - could use it for logging in, couldn't he?
18:18:24 [danbri]
much as we use openid by delegation today; or as relmeauth is proposed.
18:18:39 [bblfish]
that would fit the new definition of WebID that I gave, since it is 1) a URL 2) it refers to an Agent 3) the description is inverse functional
18:19:14 [bblfish]
but usually people thing of an ID being something the person who owns the ID shows to someone... So we're kind of missing that part in the definition
18:19:58 [bblfish]
So it's a marketing issue then
18:20:17 [bblfish]
Is it interesting to name URI's that describe Agents.
18:20:21 [bblfish]
?
18:20:34 [bblfish]
Or is it more interesting to name an authentication method built into the browsers?
18:20:49 [bblfish]
that in one click identifies you.
18:20:57 [bblfish]
if you want to be identified
18:22:10 [rszeno]
what about control over resource used to identify somebody?
18:22:26 [bblfish]
I am not sure that is the important bit
18:22:42 [bblfish]
say if your workplace controls your work identity
18:22:50 [bblfish]
or the Government controls your gov identity
18:23:09 [bblfish]
I suppose you control it too, since there are legal methods to change it...
18:23:29 [rszeno]
yes but this identify "workplace stuff" and "government stuff" not me
18:23:46 [bblfish]
you can be part of a larger agent
18:24:15 [rszeno]
true but is another agent
18:24:23 [bblfish]
no rszeno: you are many agents
18:25:19 [bblfish]
anyway, you can be rszeno@work, rszeno@freedombox ....
18:25:30 [rszeno]
not realy, at one moment i want to make the difference between multiple identities of same me
18:25:49 [bblfish]
ok, that's sidetracking
18:26:15 [bblfish]
what I want to know is with WebID there is the responsibility of the owner of the private key to take care of it...
18:26:26 [rszeno]
i can control some of them but not all
18:26:36 [bblfish]
is it such a responsibility that makes identity personal
18:26:42 [bblfish]
?
18:27:18 [bblfish]
it does not really matter what the WebID PRofile is
18:27:36 [bblfish]
if you have a private key and give it to a thief then you are in part responsible for something
18:28:33 [bblfish]
even if the private key now matches that WebID
18:28:38 [bblfish]
profile
18:28:43 [bblfish]
public key
18:29:33 [bblfish]
that private key can no longer be used to identify you. So when giving out a WebID there is a responsibility that you take on that you will take care of the private key - to some degree.
18:30:00 [rszeno]
yes but for legal considerations some of identities can't be controled. Like in your example the government can limit my ability to change things in what i have and they control so they must control the identity
18:30:38 [bblfish]
yes, but if they give you a gov WebID it will come with the provision that you take care of the private key, and notify them if it gets lost
18:31:13 [bblfish]
because otherwise the description on the web is no longer correct. It no longer identifies you, but the thief
18:31:49 [bblfish]
and the thief could use this to harm you or others.
18:32:07 [bblfish]
now say we use other auth systems,
18:32:15 [bblfish]
say http://en.wikipedia.org/wiki/Stephen_Fry#it describes where Stephen Fry lives
18:32:33 [bblfish]
and I can now authenticte steven by sending him a snail mail
18:32:40 [melvster]
melvster (~melvin@p5797EE95.dip.t-dialin.net) has joined #dig
18:32:42 [deiu]
bblfish, can we please discuss this during the teleconf on Fri? :)
18:32:57 [rszeno]
agree, but also the gov must have control of identity in case i'm not responsable, :)
18:33:04 [deiu]
I want to follow the discussion but I have to work at the same time
18:33:12 [bblfish]
deiu I added that to the topics for discussion
18:33:15 [melvster1]
melvster1 has quit (Ping timeout: 260 seconds)
18:33:48 [bblfish]
deiu: yes, I am not quite clear about this part... So if we come to a conclusion we can summarise it during teleconf
18:34:02 [deiu]
ok
18:34:51 [bblfish]
so now I need to work out how I can go from http://en.wikipedia.org/wiki/Stephen_Fry#it to address authentication
18:35:03 [bblfish]
perhaps openid auth is easier
18:35:21 [bblfish]
<http://en.wikipedia.org/wiki/Stephen_Fry#it> foaf:openid <http://stephenfry.org/
18:35:39 [rszeno]
from my point of view http://en.wikipedia.org/wiki/Stephen_Fry#it is only a description of a resource not something you can use for identification. Stephen Fry have limited control over what wikipedia editors say about him
18:36:18 [bblfish]
rszeno: I think there is something to that intuition.
18:36:25 [bblfish]
but it's difficult to work out what is missing.
18:36:34 [bblfish]
is it just control?
18:36:42 [bblfish]
is it responsibility?
18:36:55 [bblfish]
Does stephen fry have to know about his WebID somehow
18:37:02 [bblfish]
perhaps that is it.
18:37:23 [bblfish]
Just like knowledge requires belief of the knower, so an ID requires the referred person to know about his ID?
18:38:39 [bblfish]
S knows that P <-> p & S believes that P & if p were not the case then S would not believe that p
18:39:39 [bblfish]
so rszeno I think one could argue that for an ID the owner needs to at least know that ID
18:40:03 [bblfish]
because otherwise the whole privacy talk of "controlling your identity" is not going to make sense
18:41:52 [rszeno]
this i what i said, or indent to say, :)
18:42:18 [melvster]
rszeno a uri is a name or a reference, it is not a description, it is perfectly legitimate to use that name to identify someone or something
18:42:24 [rszeno]
ownership and control
18:42:40 [bblfish]
So a WebID is 1) a URL that refers to an agent determinately 2) the agent knows the webid to refer to himself
18:43:12 [rszeno]
melvster, we talk about webid and authentication?
18:43:35 [bblfish]
perhaps 3) the agent refered to can prove he his the referent of 3)
18:43:38 [scor]
bblfish: that means one cannot create a URI to denote someone else who does not have a WebID?
18:44:00 [bblfish]
scor: no you can create a URI to refer to someone, but then it would not be a WebID
18:44:22 [bblfish]
I am just trying to build in some notion of identity that privacy folks will understand
18:44:47 [scor]
bblfish: agreed
18:45:09 [scor]
because most likely no one will attempt to use that as WebID anyways
18:45:09 [bblfish]
perhaps we could have the folloiwing
18:47:17 [deiu]
I doubt it's worth considering an example in which you're dealing with public space, that everyone can control (limited in the case of Wikipedia)
18:47:52 [bblfish]
1) a WebID URI is a URL that refers to an agent determinately ( ie: distinguishes him from all other agents ) 2) the agent referred to should be able to prove his being the referent of that URI 3) the agent knows his WebID
18:48:09 [deiu]
the premise of WebID authentication is that user information can only be controlled by the user himself
18:48:35 [bblfish]
deiu, yes, but those examples danbri proposed are good ones, as they help us develop what a WebID is
18:49:02 [bblfish]
ie, our initial definition allows everything to be a WebID
18:49:04 [betehess]
silly question: what's the point of saying "URL" if the intent of retrieving some RDF is not there?
18:49:25 [bblfish]
betehess: I did not add all the rdf stuff up there cause I only have 10 fingers to type
18:49:26 [deiu]
indeed betehess
18:49:35 [betehess]
ah, ok
18:49:39 [bblfish]
:-)
18:49:39 [betehess]
just wanted to check
18:49:59 [bblfish]
I can't think without RDF anymore so don't worry about that :-)
18:50:27 [deiu]
it's pointless unless the authentication service dereferences the URL
18:51:14 [betehess]
sometimes it's good to let people walking in circles for a while, until they realize what's happening :-)
18:51:25 [bblfish]
yes, so the URL can be dereferenced. We are taking TPAC definition ( close to it) and adding some features to make it clear that there is a relation between the referred person and the URL where the referred to person in some sense controlls the id
18:52:06 [bblfish]
that was not in the TPAC definition
18:52:26 [deiu]
yes, I agree with that
18:52:56 [bblfish]
The TPAC definition did not even say what comes back in the RDF - I added earlier that the description at the URL has to be uniquely identifying
18:53:49 [bblfish]
so WebID - URL + Identifying description at URL Location + referent being an Agent that knows or controls something about the WebID
18:54:21 [betehess]
the "what" is not necessary, except for RDF
18:54:47 [betehess]
the ultimate plan is to s/Turtle/whatever LDP says/
18:54:54 [betehess]
just too soon
18:55:01 [scor]
betehess: +1
18:55:06 [bblfish]
yes, that's the syntax. But the content was not specified
18:55:07 [amy]
amy (~amy@30-6-207.wireless.csail.mit.edu) has joined #dig
18:55:11 [amy]
Joe?
18:55:17 [amy]
care to chat w/ TimBL?
18:56:12 [bblfish]
that is what does the WebID Profile return? Nothing ? Will an empty graph do ? Or somehting? What?
18:57:06 [bblfish]
so we used to have it that the graph contains the referents public key.
18:57:21 [bblfish]
now I propose to extend it to: "The WebID when dereferenced MUST return a document/representation that describes the URL referent uniquely."
18:57:28 [bblfish]
where uniquely is a definite description
18:58:22 [bblfish]
( a term that comes from philosophy of names btw )
18:58:25 [betehess]
the intent was already there, when we said "... denotes a Person/Agent"
18:58:30 [bblfish]
http://plato.stanford.edu/entries/names/
18:58:44 [deiu]
amy, Joe Presbrey?
18:58:55 [deiu]
he's on the couch
18:58:56 [rszeno]
or group?
18:59:00 [bblfish]
yes, the intent was there. but it is worth making explicit
18:59:01 [amy]
hi Andrei, great, tx :)
18:59:25 [bblfish]
because otherwise people will say: this uri denotes a person but nobody knows it or has any idea
18:59:29 [deiu]
amy, I passed the message
18:59:30 [bblfish]
and then any URI could be a WebID
19:00:13 [amy]
thanks! :)
19:00:21 [amy]
nice you're getting to hang w/ him :)
19:01:16 [bblfish]
so now you have a WebID URI that identifies an agent. But we still miss the point of the referent ( the identified agent ) knowing or controlling the identity.
19:01:36 [bblfish]
that was not in TPAC - I did bring it up though if you check the minutes
19:01:37 [deiu]
:-)
19:01:42 [betehess]
bblfish: what people wants to do with the URI (eg. authentication, authorization, etc.) does not matter
19:01:51 [betehess]
I don't understand your last point
19:02:04 [betehess]
" we still miss the point of the referent ( the identified agent ) knowing or controlling the identity"
19:02:09 [bblfish]
betehess: But then what has that got to do with an Identity system which privacy people speak about when they say you must control your identity
19:02:34 [betehess]
that's up to what people wants to do with it
19:02:37 [bblfish]
so you want to create the notion of an identity that most people who speak about identity have no use for?
19:02:45 [betehess]
if does not have to be about privacy
19:03:05 [bblfish]
so why bother calling it a WebID?
19:03:13 [bblfish]
Why not just say: a URL about a person
19:03:22 [betehess]
because it's only about identity?
19:03:28 [betehess]
on the Web (URL)
19:03:30 [bblfish]
every URI is about reference
19:03:35 [betehess]
sure
19:03:42 [bblfish]
and so why not create a URL for cheeses
19:03:43 [betehess]
I don't feel like I need to know more
19:04:01 [deiu]
WebID is a unique identifier, which only points to a single person/agent
19:04:02 [bblfish]
We could create a class of cheese urls and give it a name
19:04:05 [betehess]
well, we already have "denotes a Person/Agent"
19:04:24 [bblfish]
ok. So why bring in the RDF graph then?
19:04:26 [deiu]
hence the whole discussion about hash fragments
19:04:40 [betehess]
look, I'm just saying that the current definition already captures what we need
19:05:25 [betehess]
yes, the hash fragment makes explicit the difference between the abstraction and the document that you GET
19:05:37 [bblfish]
so then is <http://en.wikipedia.org/wiki/Stephen_Fry#it> a WebID for Stephen Fry?
19:05:50 [bblfish]
even if he does not know about it?
19:05:56 [bblfish]
or control it
19:05:57 [betehess]
that's an expectation that lets you speak about this Identity: it exists on the web through the document
19:06:03 [betehess]
so what?
19:06:17 [deiu]
by the latest definition it is
19:06:24 [betehess]
yes
19:06:27 [betehess]
I'm fine with it
19:06:29 [deiu]
it doesn't mean that it can be used for authentication
19:06:33 [betehess]
exactly
19:06:58 [deiu]
nor that I will trust it :-)
19:06:59 [bblfish]
so what have you gained by calling it a WebID
19:07:07 [bblfish]
why not just say a URL for stephen fry
19:07:31 [deiu]
because I know Stephen Fry is a person just by looking at the URL
19:07:45 [bblfish]
say I write down your name and a description of you on a piece of paper here. CAn I then say I have your id?
19:07:45 [betehess]
you solve http-range-14 per definition, and you set expectation on GETting some RDF
19:08:01 [betehess]
you don't have my id
19:08:07 [betehess]
you're just speaking about an id
19:08:08 [deiu]
you have _an_ id
19:08:10 [bblfish]
yes, I have a piece of paper that describes you
19:08:18 [betehess]
so you say
19:08:21 [betehess]
proove it :-)
19:08:24 [bblfish]
I don't even have AN ID
19:08:29 [betehess]
it does not matter
19:08:49 [deiu]
I think it's about _an_ id vs _the_ id
19:08:53 [betehess]
what you do with this id is not important at this point
19:08:58 [bblfish]
well I still have a question why don't you say URI
19:09:05 [bblfish]
you don't need a new concept
19:09:13 [bblfish]
you can say there is a URI for stephen fry
19:09:20 [bblfish]
and all is good
19:09:29 [bblfish]
why do you want to say WebID now?
19:09:54 [betehess]
bblfish, it's because I need the hash URI, I need GET some RDF. it's just easier to come up with a stable name for this concept, instead of using the whole definition
19:09:58 [deiu]
yes indeed, and the WebID is a subclass of URI, with specific semantics
19:10:10 [bblfish]
well then use URL
19:10:16 [betehess]
not enough
19:10:19 [bblfish]
there is a URL for stephen Fry
19:10:33 [betehess]
yes, but a machine can't learn anything about it
19:10:50 [betehess]
and a human can't know it's abstract, because of the missing hash
19:11:03 [bblfish]
so you need a concept of an RDF URL?
19:11:15 [deiu]
yes, specific for people/agents
19:11:28 [bblfish]
why specify to people agents. Let's start with the more generic case
19:11:31 [betehess]
yes, could be captured but LDP, but would be applied to people/agent
19:11:34 [bblfish]
an RDF URL for anything
19:11:40 [betehess]
s/but LDP/by LDP/
19:12:11 [bblfish]
the thing is betehess you need a lot more than the notion of a URL of a person that returns RDF in LDP
19:12:23 [bblfish]
you need all URLs that return RDF to be selected
19:12:29 [betehess]
not for WebId
19:12:30 [bblfish]
you could call those LD URLs
19:12:39 [betehess]
we're interested only about Identity here
19:12:45 [bblfish]
why are you?
19:13:03 [betehess]
because it enables discussions about AuthZ and AuthN
19:13:05 [bblfish]
Why is identity of agents more important to you than any other identity
19:13:12 [webr3]
i'm missing the point of this discussion
19:13:14 [deiu]
actually, the RDF returned by the # URL can contain a lot of things, but only the #me part of it refers to the person/agent
19:13:19 [betehess]
webr3, so am I
19:13:40 [betehess]
webr3, not sure why I'm discussing :-)
19:13:53 [bblfish]
the point is that betehess at TPAC said this is a marketing issue right
19:13:58 [betehess]
yes
19:14:09 [bblfish]
now we are using the notion of an ID to mean something that most people don't care about at all
19:14:19 [betehess]
?
19:14:23 [bblfish]
that has no relation to what they think of as an identifier
19:14:33 [bblfish]
like an ID is something a user controls
19:14:37 [bblfish]
owns
19:14:40 [bblfish]
does something with.
19:15:04 [bblfish]
In your definition I can create a WebID for you and you never know about it
19:15:09 [betehess]
we said we didn't want to solve the Identity problem for everybody, because we couldn't. The point is to make use of LDP as a platform.
19:15:18 [betehess]
yes, so what?
19:15:32 [betehess]
you keep saying things that are not an issue
19:15:35 [webr3]
the set of urls which refer to agents is the set of urls which have a type of :Agent . it's nothing in the url, in the name, it's in it's usage
19:15:58 [bblfish]
so now you are leaving yourself open to people saying: We have 1billion WebIDs, but nobody knows about them, so they are useless
19:16:12 [betehess]
I don't understand
19:16:19 [betehess]
why is it useless?
19:16:24 [betehess]
to you? maybe
19:16:36 [webr3]
if I have 1 billion screws, and nobody knows about them, they are not useless, they are just unknown at the present time
19:16:44 [betehess]
yes
19:16:57 [betehess]
and I don't see that as an issue
19:17:21 [betehess]
as webr3 said above, "it's in it's usage"
19:17:24 [bblfish]
If I say I have one billion screws, but I only have 1 because I know call 1 a "bliion" then it will just confuse people
19:17:28 [betehess]
I used the word "intent"
19:17:47 [betehess]
bblfish, I don't understand
19:18:04 [deiu]
there's a huge different between 1 billion facebook WebIDs and 1 billion Wikipedia URLs :-)
19:18:12 [bblfish]
people use the word ID to mean something they know about - that is something that people in the privacy world mean
19:18:30 [bblfish]
in your usage you are changing the meaning of a word
19:18:46 [bblfish]
( at least that is something one could argue - and that is my worry )
19:18:57 [betehess]
what's missing in the definition for the "people in the privacy world" then?
19:19:08 [bblfish]
it's that the agent knows about the WebID
19:19:12 [bblfish]
and has some control over it
19:19:21 [webr3]
deiu, there sure is, because fb keep all the good data back.. if it had all the relations in there, all the hidden by the silo stuff, it'd be as (if not more) useful than dbpedia
19:19:23 [betehess]
has he?
19:19:39 [bblfish]
yes, when I go to a web site I can choose my WebID
19:19:42 [bblfish]
or my ID
19:19:55 [bblfish]
I can use OpenID ot login, or WeBID, or something
19:20:06 [betehess]
I can have a private database of webids for my internal use, and you will never know about it
19:20:15 [betehess]
still ok
19:20:22 [bblfish]
under your definition
19:20:23 [betehess]
I just want to speak about these people
19:20:49 [webr3]
problem: I need/want to provide my webid to client side scripts - how?
19:21:08 [betehess]
you're conflating two notion: "speaking about an identity on the Web" and "using you Identity on the Web"
19:21:11 [bblfish]
well you give them the URL
19:21:14 [deiu]
define provide
19:21:28 [webr3]
... it's like client side login, make <x> script aware of my identity - "give them the URL" yes, but how?
19:21:43 [deiu]
as a param?
19:21:49 [bblfish]
I don't think it patters
19:21:51 [deiu]
-webid <URL>
19:21:52 [bblfish]
matters
19:22:21 [bblfish]
ok, got to go for dinner.
19:22:49 [webr3]
I feel it matters, I have an increasing amount of bookmarlets and client side js apps, all of which need to know my profile uri - and no common way to give it to them, or request it from the user
19:23:37 [bblfish]
thanks for the discussion...
19:24:29 [deiu]
webr3, so you have a discovery problem?
19:25:05 [deiu]
I wonder if you can export a system variable containing a WebID
19:25:43 [deiu]
OS login -> set WebID variable
19:27:11 [rszeno]
webr3 something using foaf: PersonalProfileDocument and foaf: primaryTopic ?
19:31:31 [timbl]
presbrey, grammar tools in n3 at http://www.w3.org/2000/10/swap/grammar
19:36:33 [timbl]
Q=../../../../2001/sw/DataAccess/rq23
19:36:33 [timbl]
sparql.n3 : $Q/grammar.yy
19:36:35 [timbl]
sed -f yy2n3.sed < $Q/grammar.yy > sparql.n3
19:36:52 [timbl]
in http://www.w3.org/2000/10/swap/grammar/Makefile
19:37:20 [timbl]
in the search for sparqlUpdate grammars
19:40:15 [tyteen4a03]
tyteen4a03 has quit (Read error: Connection reset by peer)
19:40:25 [tyteen4a03]
tyteen4a03 (~tyteen4a0@n218250225173.netvigator.com) has joined #dig
19:53:55 [danbri]
danbri has quit (Remote host closed the connection)
20:24:56 [Snuupy]
Snuupy has quit (Excess Flood)
20:25:12 [Snuupy]
Snuupy (~Snuupy@2607:5300:20:201::1:3077) has joined #dig
20:25:50 [Snuupy]
Snuupy has quit (Changing host)
20:25:50 [Snuupy]
Snuupy (~Snuupy@unaffiliated/snuupy) has joined #dig
20:36:17 [bblfish]
anyway, it's difficult to come to a conclusion on that one. by reducing WebID to a URI about a person we gain that we have something uncontroversial, making it easier to pass.
20:36:46 [bblfish]
through a standards process.
20:37:32 [bblfish]
And even then as we heard at the identity meeting on Wednesday some people find a URI that identifies peope should not be dereferenceable, but I think we can safely ignore those.
20:39:32 [bblfish]
then one has the notion of my WebID your WebID and well some WebID that describes me, but that is not the one you should link to.
20:40:03 [bblfish]
because I don't control it at all.
20:41:56 [scor]
scor has quit (Quit: scor)
20:51:57 [deiu]
yes, that's pretty much it
20:52:29 [deiu]
bblfish, do you have an example of POSTing turtle to an endpoint?
20:53:50 [bblfish]
deiu: I think I have examples of that in https://dvcs.w3.org/hg/read-write-web/
20:54:12 [deiu]
thanks, I'll take a look
20:54:48 [bblfish]
btw. looks like Colorado legalised Banana Diesel betehess :-)
20:55:44 [betehess]
don't know about banana, but I believe that it happened in MA as well
20:59:34 [melvster]
deiu: I think you can POST turtle to data.fm
21:00:46 [deiu]
melvster, that's not what I wanted :)
21:00:59 [deiu]
I want to see real examples of REST RWW using turtle
21:01:38 [deiu]
this should be part of the examples in the LDP specs, though only GET is presented so far
21:01:50 [deiu]
I guess I'll have to wait?
21:07:28 [deiu]
I think I'll have a talk with presbrey when he's around
21:14:08 [webr3]
deiu, what would you be looking for? (also LDP doesn't touch POST for normal LDPRs)
21:16:32 [deiu]
I suppose PUT would be better
21:17:58 [deiu]
webr3, section 4.3 says: "There are no additional requirements on HTTP POST for LDPRs."
21:18:16 [deiu]
it doesn't really say that POSTs are not supported
21:20:26 [RalphS]
RalphS has quit ()
21:20:42 [melvster]
deiu: welcome to decentralized information :D
21:21:08 [melvster]
it's the same in UNIX
21:21:24 [deiu]
which still sucks :)
21:22:20 [melvster]
PUT is powerful
21:22:23 [melvster]
use with care
21:22:59 [webr3]
sorry wrong term, POST isn't defined for LDP, but is for HTTP of course, ie no special actions and no constraints
21:23:53 [betehess]
webr3, deiu, we're still hoping that PATCH will actually be an HTTP POST
21:24:17 [betehess]
also, we want to be able to POST some RDF to an LDPR, and get that appended to what's already there
21:24:37 [deiu]
I thought that POST requests are only used to create new resources
21:24:48 [betehess]
for an LDPC
21:25:00 [betehess]
you mean, in general?
21:25:10 [deiu]
I mean in general
21:25:20 [betehess]
nothing against that in general, of course
21:25:30 [betehess]
I don't think this was ever discussed in the WG
21:26:02 [deiu]
is it worth bringing it up now?
21:26:32 [melvster]
dont forget btw rdf is a SET so append is not like a unix append, duplicates will be removed
21:26:47 [deiu]
removed/overwritten
21:26:59 [melvster]
same thing? :)
21:27:24 [betehess]
deiu, I'm still hopping that presbrey will jump in the WG to document that :-)
21:28:33 [deiu]
I really wish santa will bring me a "Best practice" book on implementing the LDP spec
21:29:30 [betehess]
well, it's still being defined
21:29:47 [deiu]
hehe, yes, that's why I said santa :)
21:30:37 [betehess]
:-)
21:46:18 [melvster1]
melvster1 (~melvin@p5797F097.dip.t-dialin.net) has joined #dig
21:47:51 [scor]
scor (~scor@c-98-216-39-127.hsd1.ma.comcast.net) has joined #dig
21:47:51 [scor]
scor has quit (Changing host)
21:47:51 [scor]
scor (~scor@drupal.org/user/52142/view) has joined #dig
21:48:05 [melvster]
melvster has quit (Ping timeout: 252 seconds)