Nice to Meet You App
Chris Messina
--
Chris Messina
Citizen-Participant &
Open Source Advocate-at-Large
factoryjoe.com # vidoop.com # diso-project.org
This email is: [X] bloggable [ ] ask first [ ] private
Joseph Holsten
> I wrote up a very simple and straightforward app that would leverage DiSo
> endpoints (hcard, OpenID, etc) that I'd love to see developed:
> http://factoryjoe.pbwiki.com/NiceToMeetYou
I'd like to see this that manage a group profile page, a linkroll, and
show feed reader modules. Maybe it could plug into a cms?
With that stuff, you can link to your group mailing list, code repo,
wiki, show the latest ten off your planet, flickr pool, magnolia tags,
&c. More than enough to supplant facebook groups. I'm not sure I can
count how many groups I'm in that could use this.
> ...Conferenceer app...
Doesn't seem to be online. Care to share the important bits?
http:// Joseph Holsten .com
Chris Messina
Chris Messina
Citizen-Participant &
Open Source Advocate-at-Large
citizenagency.com# vidoop.com
This email is: [ ] bloggable [X] ask first [ ] private
Julian Bond
>Steve, Stephen and I were talking about how it would be nice to be able
>to get to know other members of the DiSo Project/list without
>necessarily having one of those 1000 message "who are you?" threads.
>
>I wrote up a very simple and straightforward app that would leverage
>DiSo endpoints (hcard, OpenID, etc) that I'd love to see developed:
>
>http://factoryjoe.pbwiki.com/NiceToMeetYou
lagging. +1 Like it.
1 login with your OpenID
2 we'd import your profile/hcard/etc which we could tweak
Woah Joe! Stop right there! This is classic data portability stuff and
perhaps the primary use case. How do we get from the OpenID to the
profile/hcard/etc? got any ideas? I'm assuming you'd be expecting OpenID
delegation from a page that actually contained this or pointers to it.
Or perhaps that the XRDS-Simple for that page included links to the URIs
containing that data.
--
Julian Bond E&MSN: julian_bond at voidstar.com M: +44 (0)77 5907 2173
Webmaster: http://www.ecademy.com/ T: +44 (0)192 0412 433
Personal WebLog: http://www.voidstar.com/ skype:julian.bond?chat
Next Shopping Mall 75 Miles
Chris Messina
Chris Messina
Citizen-Participant &
Open Source Advocate-at-Large
Julian Bond
>I don't mean to be pedantic Julian, but it's not "data portability",
>it's what *should* happen! :)
(pokes Chris with a pointy stick ;) ) I promise not to mention those
words again. At least not here!
>1. present identifier to new site
>2. authenticate/prove that you own said identifier
>3. remote site will look at your identifier for a profile or for
>available services
>4. where applicable, the remote site will import/*subscribe* to your
>remote profile and ask you if you want to keep your profile up-to-date
>from the remote profile or if you want to create a new profile
>5. optionally, when the site looks at my identifier, I could assign the
>remote site a token that gives it permissions to write to an inbox or
>activity stream off of my identifier
>So, I think we just need an OpenID consumer that knows that look at an
>OpenID url and grab whatever hcard it finds. Indeed, if you use the
>DiSo Profile plugin, you should be set, as well as if you use your
>Flickr URL or host an hcard on your own OpenID URL.
I wonder if its a good idea to just assume that the page containing your
hcard profile (or FOAF or whatever) is on the same url as your OpenId.
Ignore for the moment people using Yahoo!, even if you use OpenID
delegation, your OpenID is probably going to be your blog home page
while your profile detail is on an "AboutMe" page. The same goes for
your contact list. Is it always and does it have to be on your blog home
URL?
It feels to me like we need to use some auto-discovery, either via a
<link> or via XRDS-Simple to get from the OpenID URL to the URLs
containing our profile and contact list info.
Stephen Paul Weber
> hcard profile (or FOAF or whatever) is on the same url as your OpenId.
> Ignore for the moment people using Yahoo!, even if you use OpenID
> delegation, your OpenID is probably going to be your blog home page
> while your profile detail is on an "AboutMe" page. The same goes for
> your contact list. Is it always and does it have to be on your blog home
> URL?
>
> It feels to me like we need to use some auto-discovery, either via a
> <link> or via XRDS-Simple to get from the OpenID URL to the URLs
> containing our profile and contact list info.
You have heard of rel=me? Google's Social Graph API? Hello,
autodiscovery. I usually tell people that they SHOULD have an hCard
on their OpenID page (since their OpenID SHOULD be delegated to their
main profile page) - but if they can't / don't want to a rel=me link
to their profile page keeps things working.
--
- Stephen Paul Weber (Singpolyma)
Web: http://singpolyma.net/
Twitter: http://twitter.com/singpolyma
IM: singp...@gmail.com
Chris Messina
or possible (not sure how that'd work with directed identity on Yahoo,
for example).
I think we need to provide guidance on "profile discovery using
OpenID" and offer some code samples on the wiki. If you own your own
domain, a rel-me is good, but for consumers, if a parser discovers 300
rel-me links, should it really curl them all and look for hcards? I
think we do need a hint -- either in XRDS or as an additional rel
value -- to indicate where to look for someone's definitive hcard.
This could even be set dynamically, as MyOpenID does with its persona
selector over SREG.
Point is, we can make this better and easier and more robust, so let's
think this through with the first test application being NiceToMeetYou!
Chris
Typed with fat fingers on a tiny keyboard. Sent by iPhone.
On May 29, 2008, at 6:32, "Stephen Paul Weber" <singp...@gmail.com>
wrote:
Steve Ivy
"representative hCard"[1] gotten clear enough to build on/around? If
so, we should declare it "best practice" and support it in
diso-profile (I think it already is - go Stephen) and in any profile
importing we do.
Also, we should start a page on the wiki for ideas for what we'd like
to see OpenID endpoints support. In this case, some sort of hCard
marked-up profile, or a facility for pointing to one (I'd prefer my
hCard stay on redmonk.net, for example).
--Steve
[1] http://microformats.org/wiki/representative-hcard
On Thu, May 29, 2008 at 6:32 AM, Stephen Paul Weber
<singp...@gmail.com> wrote:
> You have heard of rel=me? Google's Social Graph API? Hello,
> autodiscovery. I usually tell people that they SHOULD have an hCard
> on their OpenID page (since their OpenID SHOULD be delegated to their
> main profile page) - but if they can't / don't want to a rel=me link
> to their profile page keeps things working.
>
> --
> - Stephen Paul Weber (Singpolyma)
--
Steve Ivy
http://redmonk.net // http://diso-project.org
This email is: [ ] bloggable [x] ask first [ ] private
Jonathan Putnam
however I can.
Since this feature isn't available yet, I'll use this post to
introduce myself by briefly explaining how I got here.
I've loved working with computers ever since playing with Hypercard
and learning HTML on GeoCities in the early 90s. The Launch
interactive CD impressed me so much I decided that I wanted to do
stuff like that for a living. Trouble was, there were no cohesive
programs of study to follow when I went to college in 1996. I took a
couple of CS classes, bought books and diligently surfed the net in an
effort to learn as much as I could about interactive digital media.
Fast forward to 2005 >>>
I'd had a Friendster account for a couple of years and thought it was
pretty neat to reconnect with friends I hadn't spoken with in a while.
I didn't really use it that often, but it was nice to have and I saw
it had potential. Meanwhile, MySpace is starting to really blow up. I
resisted creating a profile for the standard reasons. I wasn't really
interested in reentering my profile data and "refriending" people.
Besides, the interface and graphic sensibility of the site was
atrocious! Alas, the gravitational pull of the Myspace vortex
eventually sucked me in. Not coincidentally, it was around this time
that I began to seriously research open social networking.
The first important piece of the puzzle Google led me to was <a
href="http://journal.planetwork.net/article.php?
lab=reed0704&page=3">The Social Web: Creating An Open Social Network
with XDI</a> I was ecstatic! Here were some people who had thought
through the situation in detail. Trust networks, persistent (granular)
identity access, intelligent filtering! Ideas which were loosely
formulated in my head had been mapped out and turned into a spec! I
felt we were getting really close to <a href="http://www.noosphere.org/
background/how_to">building a global brain.</a>
My enthusiasm was dampened a bit when I realized there would be no
foreseeable support for the spec in any browser (or even from the W3.)
In fact, the consensus seemed to be that XDI was a bloated solution
without a problem. The ambition of XDI created a barrier of
complexity the public wasn't ready to hurdle. I turned instead to
OpenID. Specifically, I was interested in using OpenID with Drupal
4.7, which appeared to be the most flexible web application framework.
I am not a coder by nature. I can read code. I understand programming
enough to tweak things to work. I once tried to write a equipment
checkout solution from scratch. It didn't take long for me to come to
grips with the fact that what I wanted well exceeded the scope of my
PHP chops. After a great deal of searching, I eventually found a
framework that I could modify instead. I was drawn to Drupal because
it is multi-user and there was a rapidly growing list of interesting
modules. Experiments with WordPress had made me realize how powerful a
modular system could be. I downloaded dozens of cool modules and
tried to glue them into a music sharing community. This turned out to
be horribly, horribly slow. Unusable. I scaled back features and put
things on hiatus. Money, or rather the lack of it, became a problem.
Feeling defeated and struggling to pay bills, I abandoned my ambitions
and only rarely logged on the web most of last year.
Which brings me to now. The rapid strides in interoperability have got
me excited again. Earlier this year I discovered the Data Portability
group by way of a thread on the Open Social Networks Group. The
swelling collective energy in these groups convinced me to register
for SXSWi at the last minute. The panels were that compelling. The
energy and focus in Austin motivated me to give it another shot. Right
now, web technology is mature enough to avoid endless headaches but
fresh enough to be combined in innovative ways. All these interop APIs
are like digital legos. The trick is to find the right pieces to build
a cool spaceship. (Figuratively for now, maybe for real in a few
years! ;)
There are a lot of things I'd like to work on. I am reacquainting
myself with WordPress theming and plugins for my personal blog. I
planned on waiting for summer to dive into Drupal 6 and lo, summer is
here. It will take me a little while to get up to speed on how the
various pieces fit together, but my goal is to create an installation
profile for Drupal geared towards artists and musicians. Social
networking and ecommerce capabilities figure heavily in my ideal
profile. I would like to make a similar package of bundled,
complimentary WordPress themes and plugins.
I will gladly help develop and test extensions that further make
Distributed Social a seamless feature of the web. Once I have got the
blog set up, I'll post some of my ideas on the topic in more detail. I
look forward to participating in this exciting discussion! Hope my
"brief" introduction didn't ramble on too long . . . (it seems like
I'm just scratching the surface!)
Cheers,
Jonathan
On May 29, 10:55 am, "Steve Ivy" <steve...@gmail.com> wrote:
> I agree that rel=me + hcard should be the trick. Has the definition of
> "representative hCard"[1] gotten clear enough to build on/around? If
> so, we should declare it "best practice" and support it in
> diso-profile (I think it already is - go Stephen) and in any profile
> importing we do.
>
> Also, we should start a page on the wiki for ideas for what we'd like
> to see OpenID endpoints support. In this case, some sort of hCard
> marked-up profile, or a facility for pointing to one (I'd prefer my
> hCard stay on redmonk.net, for example).
>
> --Steve
>
> [1]http://microformats.org/wiki/representative-hcard
>
> On Thu, May 29, 2008 at 6:32 AM, Stephen Paul Weber
>
> > You have heard of rel=me? Google's Social Graph API? Hello,
> > autodiscovery. I usually tell people that they SHOULD have an hCard
> > on their OpenID page (since their OpenID SHOULD be delegated to their
> > main profile page) - but if they can't / don't want to a rel=me link
> > to their profile page keeps things working.
>
> > --
> > - Stephen Paul Weber (Singpolyma)
>
> --
Julian Bond
>I think we need to provide guidance on "profile discovery using
>OpenID" and offer some code samples on the wiki. If you own your own
>domain, a rel-me is good, but for consumers, if a parser discovers 300
>rel-me links, should it really curl them all and look for hcards? I
>think we do need a hint -- either in XRDS or as an additional rel
>value -- to indicate where to look for someone's definitive hcard.
>This could even be set dynamically, as MyOpenID does with its persona
>selector over SREG.
>
>Point is, we can make this better and easier and more robust, so let's
>think this through with the first test application being NiceToMeetYou!
+1 to all this. I'm in favour of two standards one for <link> and one
for XRDS-Simple. And for two types of links
- Here's the URL of my definitive profile data
- Here's the URL of my definitive contact list data
And to Stephen, re Social Graph, OpenSocial people, Google Contacts,
foaf+vcard, etc etc etc, there are potentially many, or at least
several, possible standards for getting the data for each of the two use
cases. I do wish this would shake itself out and we could focus on just
one de facto standard of each but it shows no signs of doing that. And
unfortunately while hcard is easy to produce it's harder to consume. It
feels like profile and contact list data really ought to be more
formally structured in more precise xml or rdf-xml.
I keep worrying at this, but for me rel="me" just doesn't quite cut it
for this purpose, because as Chris said, it potentially forces the
consuming app into spidering your whole site to find all the possible
links. And rel="me" is getting overloaded in the process. This is
related a bit to the classic social network problem: is the person
looking the owner of the data or some else and so should I say "My
profile" or "Their Profile" and the classic problem of sites that let
you edit your profile but not see it as others see it. The point being
that rel="me" is being used for both, here's my data on this site, and
here's a profile I also have on another site. There's an implied rel="me
seeAlso" in some of them.
--
Julian Bond E&MSN: julian_bond at voidstar.com M: +44 (0)77 5907 2173
Webmaster: http://www.ecademy.com/ T: +44 (0)192 0412 433
Personal WebLog: http://www.voidstar.com/ skype:julian.bond?chat
Brought to you by the number 2, a dot and the letter Oh.
Stephen Paul Weber
build stuff.
--
Steve Ivy
On Thu, May 29, 2008 at 8:27 AM, Julian Bond <julia...@voidstar.com> wrote:
> +1 to all this. I'm in favour of two standards one for <link> and one
> for XRDS-Simple. And for two types of links
> - Here's the URL of my definitive profile data
> - Here's the URL of my definitive contact list data
I wonder if a <link rel="me" href="page/with/hcard"> would work?
> And to Stephen, re Social Graph, OpenSocial people, Google Contacts,
> foaf+vcard, etc etc etc, there are potentially many, or at least
> several, possible standards for getting the data for each of the two use
> cases. I do wish this would shake itself out and we could focus on just
> one de facto standard of each but it shows no signs of doing that. And
> unfortunately while hcard is easy to produce it's harder to consume. It
> feels like profile and contact list data really ought to be more
> formally structured in more precise xml or rdf-xml.
I try to err on the side of HTML as much as possible, so I'm going to
push for a rel="me" approach if at all possible. That said, when
saying "this is my representative profile" I can see the argument for
a more formalized approach. Read over the representative-hcard pages
(http://microformats.org/wiki/representative-hcard) and let's see if
we can decide on an approach that works. I think this problem is just
waiting for someone to build something, document the behavior, and
urge others to follow along.
All that said, there's also some contacts-api work going on that may
relate to this conversation; hopefully we'll hear more soon.
> I keep worrying at this, but for me rel="me" just doesn't quite cut it
> for this purpose, because as Chris said, it potentially forces the
> consuming app into spidering your whole site to find all the possible
> links. And rel="me" is getting overloaded in the process.
> This is related a bit to the classic social network problem: is the person
> looking the owner of the data or some else and so should I say "My
> profile" or "Their Profile" and the classic problem of sites that let
> you edit your profile but not see it as others see it. The point being
> that rel="me" is being used for both, here's my data on this site, and
> here's a profile I also have on another site.
Well, in both cases they are (or should be) public resources, so
there's no disparity in the kind or visiblity of information being
provided. This is the problem of identity consolidation
(http://microformats.org/wiki/identity_consolidation), and again, I
think that if we decide on an approach and implement it, others will
follow.
--Steve
--
Steve Ivy
http://redmonk.net // http://diso-project.org
Josh Patterson
saying theres a lot of ways to skin this cat, whats the best way to do
it? The thing is, right now there are a lot of different people saying
"do it this way" or "do it my way" --- and like I told some people
yesterday in a chat --- all of them are right AND wrong.
The thing is --- just do it.
In grad research if we think we know what is going to happen, and
assume that, we get in trouble, because we dont know what will happen,
only results can show us how its going to pan out. Here I think we
just do a "n + 1" solution under the best heuristic possible, and then
bang on it, and see what falls off --- rinse and repeat.
I'm keenly interested in data discovery for a lot of reasons, and I'm
up for whatever path people want to explore. I may try a few other
things on my own, but it doesn't matter, because in the end some will
get widespread adoption and we'll all move to that, and start the next
cycle.
I've talked discovery with a number of people, and I think Julian
brings up some interesting points --- cause its not clear on a best
way to do it, and it wont be until someone does something cool with it
and everyone goes "oh shi...!" --- and then it will be really clear in
a hurry. And I think we are all mostly in agreement here.
So lets just throw something (moderately intelligent) out there, maybe
cook up some fallback methods, sprinkle a little Occam's Razor on it,
and roll with the punches. (Cause I gotta find me some data!)
Josh
http://jpatterson.floe.tv
On May 29, 11:34 am, "Stephen Paul Weber" <singpol...@gmail.com>
wrote:
> build stuff.
>
>
>
>
> > Chris Messina <chris.mess...@gmail.com> Thu, 29 May 2008 07:49:38
Joseph A Holsten
rel="me" is getting used for two things these days, pointing to stuff related to me (claimid), and pointing to the canonical Me. So we've got two options: take back rel-me for our uses, or invent our own.
Julian Bond
>XRDS tends not to be about semantic webs, but instead about resource
>disco. Your highest priority hcard service type would imply it as
>canonical. We'll need a few new service types, but this should already
>be solved.
Like this?
hCard
seeAlso: http://microformats.org/wiki/profile-uris
Type: http://www.w3.org/2006/03/hcard
MediaType: text/html, application/xhtml+xml
XFN - Contact lists
Comment: A page with mainly links to contacts and friends
seeAlso: http://microformats.org/wiki/profile-uris
Type: http://gmpg.org/xfn/11#contact
MediaType: text/html, application/xhtml+xml
XFN - Profile URLs
Comment: A page with mainly links to external profiles
seeAlso: http://microformats.org/wiki/profile-uris
Type: http://gmpg.org/xfn/11#me
MediaType: text/html, application/xhtml+xml
Bob Ngu
> related to me (claimid), and pointing to the canonical Me. So we've
> got two options: take back rel-me for our uses, or invent our own.
the canonical me which is a different use case. Maybe the MF standards
committee (if there is such an entity) might consider enhancing MF to
include specs for cross-pollinating different MF standards. Here's an
example of what I mean
For "Me"
<div class="vcard"><a class="fn url" href="http://my.blog.com/"
rel="me">John Doe</a></div>
<a href="http://www.flickr.com/photos/myname/" rel="me">Photos</a>
<a href="http://www.twitter.com/myname/" rel="me">Twitter</a>
For my contacts
<li class="vcard"><a class="fn url" href="http://janesmith.com"
rel="met colleague friend">Jane Smith</a></li>
<li class="vcard"><a class="fn url" href="http://harrypotter.com"
rel="met colleague friend">Harry Potter</a></li>
So only one rel-me entry can have an associated vcard, none of the
other rel-me entries can. For each contact, there is one associated
vcard. I discussed this briefly with Kevin Marks and he thought this
might be a good feature to add to SGAPI. Is this something worth
exploring as an extension to MF standard? My concern with adding this
as one-off without it being a standard is that most parsers won't
follow the convention making it far less useful IMO.
Bob Ngu
> <div class="vcard"><a class="fn url" href="http://my.blog.com/"
> rel="me">John Doe</a></div>
> <a href="http://www.flickr.com/photos/myname/" rel="me">Photos</a>
> <a href="http://www.twitter.com/myname/" rel="me">Twitter</a>
> other rel-me entries can.
<div class="vcard"><a class="org" href="http://www.flickr.com/photos/
myname/" rel="me">Photos</a></div>
<div class="vcard"><a class="org" href="http://www.twitter.com/
myname/" rel="me">Twitter</a></div>
or should it be this instead (sorry, not a MF expert, can a property,
"org" in this case, be repeated within hCard?)
<a class="org" href="http://www.flickr.com/photos/myname/"
rel="me">Photos</a>
<a class="org" href="http://www.twitter.com/myname/" rel="me">Twitter</
a></div>
Reason for this is that I need to access a human readable name along
with a rel-me URL, it is an actual use case for my app. Thoughts?
Stephen Paul Weber
<div class="vcard">
<a class="fn url" href="http://my.blog.com/"> rel="me">John Doe</a>
<a class="url" href="http://www.flickr.com/photos/myname/" rel="me">Photos</a>
<a class="url" href="http://www.twitter.com/myname/" rel="me">Twitter</a>
</div>
and then if you're looking for the "main" hCard for a page use the rules
at <http://microformats.org/wiki/representative-hcard-parsing>
--
Bob Ngu
better.
Also, thanks for the link to the hCard parsing article, it spelled out
the following 2 ways to parse out a canonical me
1) After parsing the hCard(s) on a page, if there is one which has a
"url" property whose value is the url of the page (source) and is also
a "uid" property for the hCard, then that is the representative hCard
for the page.
2) If the aforementioned "url uid source" didn't find a representative
hCard, then look for an hCard with a "url" property that also has the
rel="me" relation. If you find such an hCard then you have found a
representative hCard for the page.
This is exactly the kind of best practice information I (and probably
other developers) are looking for on how to mix and match different MF
standards.
Joseph A Holsten
So only one rel-me entry can have an associated vcard, none of theother rel-me entries can.