See also: IRC log
<trackbot> Date: 25 November 2009
<hhalpin> scribe: tinkster
<hhalpin> PROPOSED: to approve SWXG WG Weekly -- 18th November 2009 as a true record
<hhalpin> http://www.w3.org/2009/11/18-swxg-minutes.html
<yoshiaki> i see
hhalpin: Minutes need cleaning up, but everyone OK with them apart from that?
<hhalpin> RESOLVED: approved SWXG WG Weekly -- 18th November 2009 as a true record
<mischat> yoshiaki: http://www.w3.org/2001/12/zakim-irc-bot.html
<hhalpin> PROPOSED: to meet again Wed. Decemeber 2nd?
hhalpin: Meet again next week?
hhalpin: No speaker planned for next week. What should we do?
<hhalpin> Chris Messina next week.
<Adam> very nice
<mischat> http://activitystrea.ms/
petef: Chris Messina has been confirmed for next week - ActivityStrea.ms
<tpa> one action was closed
<tpa> I just closed it
hhalpin: Any progress since last week? We should probably just wrap them up, drop unfinished ones. Any issues with that?
<mischat> http://www.w3.org/2005/Incubator/socialweb/wiki/UserStories
<tpa> http://www.w3.org/2005/Incubator/socialweb/wiki/UserNarratives
hhalpin: It seems close to "done". We shouldn't spend too much more time on this.
<mischat> what is the relationship between the UserStories and the UserNarratives
<tpa> we talked about this last week
<Adam> think the groups were already there harry
hhalpin: This needs a strong editor. Henry, Oshani and Tim have sort of volunteered.
<mischat> http://www.w3.org/2005/Incubator/socialweb/wiki/UserStories#Groups is empty
<hhalpin> I am happy to step you through the process of making this is HTML document...
tpa: I was talking with Henry
about this, making them more readable. Dan A gave me an OK to
flesh out this idea.
... This is on the User Narratives wiki page.
<Adam> http://www.w3.org/2005/Incubator/socialweb/wiki/UserNarratives
hhalpin: What should a roll-your-own-Social-Network tool implement?
<hhalpin> Ah, I see now.
tpa: There are three types of users we want to address - end users, developer concerns, business concerns.
<hhalpin> This page is great!
<hhalpin> +1 tpa, the list is too long.
<Adam> my understanding was that they were a way to weave together the user scenarios in to a larger background
tpa: Aim is to take a big list and make it more readable/consumable/relatable.
hhalpin: What can we do to help?
tpa: We have four narratives. (Fourth one is about corporate usage.)
<bblfish> I can do a prison story if you like
<hhalpin> Which actions?
<hhalpin> Which action was done last week?
<mischat> hehe bblfish
tpa: End user story can be written by tpa. Need help with developer and business stories.
<bblfish> ;-)
<cperey> what do you mean business standpoi nt?
<Adam> i think tpa means from a startup standpoint
<cperey> by business stories do you mean corporate story?
tpa: This may be a matter of taking existing user stories and considering them more from a developer/business angle.
cperey: Different between corporate and business stories?
<cperey> can you put a URL to the third story?
<mischat> http://www.w3.org/2005/Incubator/socialweb/wiki/UserNarratives
<mischat> Zoe's ...
tpa: Corporate = social networks in a corporate context. Business = tech is not an end in itself; social networks are businesses.
<hhalpin> thinks maybe cperey could run with the business angle
<mischat> agree christine
cperey: Third story seems to be about how social networks can make money. Is that right?
<cperey> that's where I want to be so I'm happy to drive it!
tpa: Yes - precisely. I'd welcome your help on this cperey.
<mischat> i would like to help with http://www.w3.org/2005/Incubator/socialweb/wiki/UserNarratives#Trent_is_developing_Twitbook
<cperey> Yes
<hhalpin> Should these be actions?
tpa: Adam is taking lead on corporate (4th) story.
Adam: yes, that's right.
tpa: Can we talk more about this next week?
<Adam> peer pressure :)
<cperey> +1
<bblfish> sorry btw, to have been out of the loop for the last few weeks
hhalpin: We've got a backlog of actions. We'll put overdueness indicators beside each of them and start dropping the ones that seem static.
<hhalpin> Let's say end of Dec. or Jan.
<cperey> +1
<hhalpin> ACTION: cperey to flesh out corporate stories [recorded in http://www.w3.org/2009/11/25-swxg-minutes.html#action01]
<trackbot> Created ACTION-108 - Flesh out corporate stories [on Christine Perey - due 2009-12-02].
<cperey> Christine is not taking "Corporate stories"
mischat: User stories have one about Trent, which looks good for developer narrative.
<hhalpin> Oh, wait a sec..the "angel" story, right cperey?
<hhalpin> I.e. the "business" story?
<cperey> I'm taking Business strategies narrative (#3)
tobyink: I can have a look at that, yes.
<hhalpin> OK, will delete that previous action after meeting
<hhalpin> ACTION: cperey to take business narrative [recorded in http://www.w3.org/2009/11/25-swxg-minutes.html#action02]
<trackbot> Created ACTION-109 - Take business narrative [on Christine Perey - due 2009-12-02].
<hhalpin> ACTION: adam to take on corporate story [recorded in http://www.w3.org/2009/11/25-swxg-minutes.html#action03]
<trackbot> Created ACTION-110 - Take on corporate story [on Adam Boyet - due 2009-12-02].
<mischat> nice one tinkster
mischat: I'll make a start on that and liase with tobyink and tpa.
<hhalpin> ACTION: mischat to start on developer story [recorded in http://www.w3.org/2009/11/25-swxg-minutes.html#action04]
<trackbot> Sorry, couldn't find user - mischat
<tpa> ACTION: tanglade to take on developer narrative and user stories with mischat [recorded in http://www.w3.org/2009/11/25-swxg-minutes.html#action05]
<trackbot> Could not create new action (failed to parse response from server) - please contact sysreq with the details of what happened.
<trackbot> Could not create new action (unparseable data in server response: local variable 'd' referenced before assignment) - please contact sysreq with the details of what happened.
<mischat> hhalpin: my name on the w3c is mtiffiel
<hhalpin> ACTION: mtiffiel to start on developer story [recorded in http://www.w3.org/2009/11/25-swxg-minutes.html#action06]
<trackbot> Created ACTION-112 - Start on developer story [on Mischa Tuffield - due 2009-12-02].
<mischat> there you go mhausenblas we are on Final Report now ;)
<hajons> yes
<cperey> who?
<hhalpin> tim berners-lee
hhalpin: We have lots of writeups of speakers.
<cperey> ah, is there a URL?
<cperey> :-)
<mischat> when is it due ?
<cperey> we need to make recommendations
hhalpin: For final report, see if timbl can talk about his big vision; I'm happy to consolidate the various small pieces around that.
<hhalpin> ACTION: [CONTINUES] adam to write up the boeing use case for enterprise social networks [recorded in http://www.w3.org/2009/11/25-swxg-minutes.html#action07]
<hhalpin> ACTION: [CONTINUES] oshani to reframe the geolocation/intent/portability [recorded in http://www.w3.org/2009/11/25-swxg-minutes.html#action08]
<hhalpin> ACTION: [CONTINUES] bblfish to relabel data protection use case to be about controlled access and takedown to data "about" you [recorded in http://www.w3.org/2009/11/25-swxg-minutes.html#action09]
<hajons> Good idea. His vision was not entirely clear to me, so I think him writing something about it would be great
<hhalpin> ACTION: [CONTINUES] bblfish to merge Family and Group access usecases [recorded in http://www.w3.org/2009/11/25-swxg-minutes.html#action10]
<hhalpin> ACTION: [CONTINUES] hhalpin explain to henry and oshani doc editing process for usecases [recorded in http://www.w3.org/2009/11/25-swxg-minutes.html#action11]
<hhalpin> ACTION: [CONTINUES] DKA to write-up output of XG meeting next Tuesday [recorded in http://www.w3.org/2009/11/25-swxg-minutes.html#action12]
tpa?: His vision was changing the architecture for how this social data is stored and accessed.
<hhalpin> ACTION: [CONTINUES] DKA to send notes to camp to Social Web XG wiki and BarCamp.org [recorded in http://www.w3.org/2009/11/25-swxg-minutes.html#action13]
<hhalpin> ACTION: [CONTINUES] DKA to summarize OSLO and geoLocation conversation in order to spread knowledge of these efforts among W3C members. [recorded in http://www.w3.org/2009/11/25-swxg-minutes.html#action14]
<hhalpin> ACTION: [CONTINUES] Mischa to describe/implement a report of terms and conditions, and how they change between now and the end of the XG. [recorded in http://www.w3.org/2009/11/25-swxg-minutes.html#action15]
hhalpin: I'll try to get timbl involved.
<hhalpin> ACTION: [CONTINUES] Adam to write up Matt Lee's talk [recorded in http://www.w3.org/2009/11/25-swxg-minutes.html#action16]
<hhalpin> ACTION: [CONTINUES] mtuffied to put up wiki page about social networks deploying these technologies. [recorded in http://www.w3.org/2009/11/25-swxg-minutes.html#action17]
<hhalpin> ACTION: [CONTINUES] oshani to write up David Recordon and Luke Shepard talk [recorded in http://www.w3.org/2009/11/25-swxg-minutes.html#action18]
<hhalpin> ACTION: hhalpin to try to get TimBL to either talk or write-up his use-cases [recorded in http://www.w3.org/2009/11/25-swxg-minutes.html#action19]
<trackbot> Created ACTION-113 - Try to get TimBL to either talk or write-up his use-cases [on Harry Halpin - due 2009-12-02].
tpa?: Tim's vision is that the data is stored in the cloud and you'd use apps to process that.
<cperey> The final report: we need to make recommendations
hhalpin: Eran's gotten OAuth off the ground, simplified XRD, etc. Eran can you introduce your work?
<cperey> person who just entered
Eran: The basic premise is to bootstrap new social protocols on top of existing web arch; allow people to identify themselves using (interrupted)
<claudio> Zakin, mute me
Eran: OAuth takes care of
security/authorisation part.
... But how do we find out what services a person uses, their
preferences, their friends, etc?
... Started by looking at how to make OAuth easier.
... Has expanded into 5 or 6 specs across multiple
standardisation bodies.
... It's a big/common problem. Value in solving it
consistently.
<hhalpin> +1 to links being fundamental building block
Eran: Links are a fundamental
building block. If I've got a URI, I can get a representation
of that and find more links from it.
... Using HTML <link> elements we started to run into
some problems.
... Adding the links changes to documents and violates the
purity of the document.
... Retrieving the documents is a lot of work.
<PhilA2> See http://tools.ietf.org/html/draft-nottingham-http-link-header-06
Eran: So we've been looking at
the HTTP "Link" header which was taken out of HTTP 1.1 but
we're trying to bring back.
... Allows links in the HTTP header itself.
... HTTP also used to have LINK and UNLINK methods in addition
to GET, POST, etc.
<hhalpin> I vaguely remember hearing/reading from Roy that the removal of "Link" from HTTP was a cut-and-paste error :)
<mischat> http://www.w3.org/Protocols/9707-link-header.html
<jar> no, that was removal of Link: header, not LINK method
Eran: Still looking for ways to make this better. Sometimes resources don't have a representation at all, so GET/HEAD don't return anything useful.
<mischat> ah
<hhalpin> meant link header :)
Eran: Yadis/XRDS uses content negotiation to do this. You'd make a request for a URI and ask for XRDS and get a description.
<hhalpin> interesting point re misbehaving proxies for conneg.
Eran: But this doesn't seem to be
an appropriate use of conneg. And big consumers like Yahoo
refused to use it because of the risk of broken proxies.
... But how about using a pattern which maps between resource
URIs and description URIs. We spent 6 months trying to figure
out where to store that mapping pattern.
<hhalpin> Am a bit confused about relationship between host-meta and .well-known
Eran: We came up with two
solutions. One was a well-known document called first
site-meta, then host-meta, then .well-known.
... There isn't a good way of storing this information other
than well-known locations. (Other than DNS, more on that
later.)
... HTTP OPTIONS looked promising, but not cachable or widely
implemented/understood.
... LRDD tries to unify all these ideas. If you have a resource
identified by a URI, there are three paths to find descriptions
for it. (Not all of them will always be available.)
<PhilA2> Eran's excellent summary of this issue from September 2008 is at http://hueniverse.com/2008/09/discovery-and-http/
Eran: 1. <link> elements within document; 2. Link HTTP headers; 3. host-meta.
<hhalpin> +1 order of precedence
Eran: LRDD started as a very generic solution, but we plan on making it less generic because host-meta/.well-known provides alternative building-blocks for more flexible solutions.
<bblfish> I had a question earlier
<PhilA2> This document http://www.w3.org/2001/tag/doc/mime-respect suggests that HTTP Link should be the first/most authoritative
hhalpin: What if different sources of metadata contradict each other? What are the social web use cases?
Eran: If you use OpenID, you're starting with an identifier. The application then needs to figure out from that how to authenticate you.
<hhalpin> Exactly, that's where LRDD or some notion of "authoritative" metadata algorithm would be useful.
Eran: Because OpenID is a
security protocol, it's important that all apps use the same
approach - otherwise conflicting data could be very
problematic.
... OpenID thus has a very well-defined order of precedence
when looking for links.
... For LRDD, when you take a URI, you need to figure out which
description methods are available to you.
... Once you know your options, the issues are which do you try
first, and what happens if you get different response
codes?
... Link header is defined differently to other headers. A
"Link" header on an HTTP 401 page is a link from the error
page, not a link from the thing you asked for.
... If there's a 303, do you follow the redirection and use the
links at the destination; or both?
... Different people will want to take different
approaches.
... Some will want to slurp up as much data as possible; some
will want to stop as soon as they find what they need.
... LRDD right now is essentially an inventory of your
options.
... Another way we could have written LRDD would be to say
which order you should check descriptions.
... Right now we want to move to a very deterministic approach,
because the building blocks are still there for those who want
to do things differently.
<hhalpin> http://www.iana.org/assignments/relation/
<PhilA2> See http://www.iana.org/assignments/link-relations/link-relations.xhtml
hhalpin: <link> doesn't have a centralised registry. HTML5 has a wiki page. IANA might be starting a registry.
<hhalpin> all we have right now is ATOM :)
<PhilA2> (The link I posted was for Atom)
Eran: mnot started resurrecting Link and merging it with the Atom <link> registry.
<hhalpin> Maybe just update that Atom page?
<hhalpin> Or maybe not...
Eran: Some relationships have very specific meanings for particular media - e.g. rel="stylesheet" in HTML.
<PhilA2> HTTP Link Header draft establishes a new registry. See http://tools.ietf.org/html/draft-nottingham-http-link-header-06#section-6.2
<mischat> sounds like rdfa
Eran: People expect relationships to have semantic meanings like "parent" or "copyright". Some people want to restrict the media type / format of the linked object.
<hhalpin> well, this would be a way to give something like "foaf:Person" a shortname
<hhalpin> and solve the rather infamous namespace issue, i.e.what if people forget the namespace for certain well-known RDF vocabularies?
Eran: We'll probably have two types of relationship - some which are useful across different contexts - and these will be registered.
<Zakim> mhausenblas, you wanted to ask on XRD deployment (re update on http://linkeddata.deri.ie/tr/2009-discovery)
Eran: Other than that you can use a URI for a relationship type and you don't need to register it.
<mischat> sure, but the beauty of rdf is that you can say anything, why have to wait for a standard to be drawn out
<mischat> +1 tinkster
<PhilA2> yeay!
<hhalpin> +1 for having a centralized registries
Eran: For HTML5 we hope to avoid the need for different registry.
mhausenblas: Do you have any info on the XRD deployment base?
Eran: it's going through OASIS. We've got another change to align better with Atom's links that still needs to happen.
<mischat> mmm gmail foaf page?
Eran: In terms of deployment, it's used in Gmail/Yahoo for WebFinger. Not much data there. Gmail point to some FOAF.
<mischat> is that out already ?
Eran: We're waiting for some of the missing pieces to drop into place, especially LRDD.
hhalpin: What's the relationship between .well-known and host-meta?
<hhalpin> has one been deprecated?
Eran: Originally wanted to have
one document called site-meta which would be *the* place to
link to other documents. No more well-known URIs ever.
... It was then renamed to host-meta because site is not the
right term.
... But people said it would never work - it adds too much
indirection.
... Concerns that "/host-meta" might already be used on some
hosts.
<hhalpin> Ah, good point re someone getting /hoste-meta
Eran: So instead we decided that
any well-known paths are OK, so long as they start with
"/.well-known/" and are registered.
... But rather than require people to register well-known
documents, they can still use host-meta to store data if that's
more suitable.
... An HTTP server on port 80 also has authority to provide
host-meta data for other services on other ports (but the same
host name).
hhalpin: Any more questions?
<jar> TAG is trying to track this, but there needs to be more coordination
<mischat> are you there tinkster
<mischat> i can scribe
<jar> (actually I meant more coordination within W3C!)
Eran: I'm in constant touch with W3C TAG; make sure they're aware of this work and provide feedback.
<mischat> does the W3C need to track the LRDD, and similar outputs from the OpenID/OAuth world
Eran: Role of W3C should be
formats, more ways of describing metadata, making RDF easier to
use for descriptions.
... W3C doesn't seem agile enough to support these communities.
That's a bigger topic - how do we get the audience of the W3C
to participate in these ideas without slowing down spec
generation?
<PhilA2> Can't help noting that POWDER is all about making RDF about lots of resources easier to publish ;-)
<jar> among other things, W3C should decide how to advise use of these link mechanisms for nose-following... is POWDER the whole story?
<hajons> example host-meta: http://gmail.com/.well-known/host-meta
Eran: Maybe eventually we would want to turn to W3C or similar for standardisation once the specs are more stable.
<jar> I mean RDF nose-following specifically
<PhilA2> Even I wouldn't claim POWDER's the whole story jar!
Eran: This has been discussed extensively on TAG, so they're paying attention. TAG issues 62 and 63 relate to it.
<jar> (POWDER needs credit & standing. gets overlooked)
hhalpin: It's been discussed a lot recently what should be changed with RDF.
<mischat> interesting hajons
hhalpin: At TPAC we discussed
what the W3C could do to become more agile. Incubator groups
seem a good start, but what would be better?
... Can the W3C give advice on, e.g. how IP issues can be
resolved?
Eran: W3C participation is very
expensive.
... Open Web Foundation exists because a lot of people don't
want to work within established standards bodies.
... Standards bodies can address some of these issues, but
others cannot be addressed because of the very nature of
standards bodies.
... OAuth started as a bunch of people hacking at something to
try to make it work, but wanted a more "adult" spec, so turned
to IETF,
... If XRD hadn't started within OASIS, then I would have taken
XRD Simple to the W3C once it was getting mature.
<caribou> waiting until they are in stable state is not saving time
Eran: Little communication
between standards bodies and communities.
... Standards bodies need to look at how we can support this
work without getting jealous and competing.
hhalpin: Right now W3C is concentrating a lot on APIs.
<Zakim> PhilA, you wanted to chip in
hhalpin: We'd like to run our final report by you.
Eran: Sure.
PhilA: Device Application Working
Group has taken ideas from outside the W3C.
... I met Eran just over a year ago. If I'd seen XRD before we
started POWDER (or vice versa) we might not have ended up with
them both.
... Some competition is inevitable because not everybody
communicates with everybody else.
Eran: I meant more about gratuitous competing standards.
hhalpin: Academic journals - if you have a good paper, there's often only one journal it should be sent to.
<hhalpin> or is that problematic?
<mischat> i have to go now people, will follow on IRC
hhalpin: Perhaps standards should be endorsed by multiple different organisations. Perhaps W3C should be able to say, "for this use case, take a look at this spec".
Eran: W3C stamp means more than
"this is a good idea" - it has e.g. IP consequences.
... With XRD we spent a year throwing things out.
<hhalpin> easier to let people who start something finish it - like that point.
<hajons> have to go. thanks for a great talk
Eran: But spec writing doesn't
need to take long. It would be good if there was an easy path
for turning a community spec into a standard.
... The W3C already has the right expertise.
<hhalpin> well, we could imagine something like OWF or something jumping in where the XG process is right here.
Eran: I'm involved in this work because of my day-job with Yahoo.
hhalpin: Thanks Eran. We'll be
wrapping up now.
... Final Report should be in draft form in Jan/Feb. We'd like
to hear any ideas you've got.
... What would you suggest we say in our summary of
OAuth/WRAP?
<hhalpin> At some point, we should have serious discussion on maybe this topic - how to avoid rubber-stamping?
Eran: I'm happy to talk about this in the future. Now that it's 8am it's not so bad!
<cperey> bye all
<hhalpin> Meeting adjourned.
Fin.
<hhalpin> trackbot, end meeting
<cperey> very valuable