Further discussion - Chat Log from Skype
Chris Saad
If you want to be included in the chat just let me know your skype
account if I dont have it or ping me on skype and let me know.
[2:18:59 PM] Josh Patterson says: also, YADIS -- can anyone give me a
quick rundown on that one? im looking at the wikipeida on it
[2:19:05 PM] Josh Patterson says: it might just do a lot of what is
needed
[2:19:24 PM] Chris Saad says: Yeah YADIS is my find really - its
pretty obscure but i think it can be very powerful
[2:19:38 PM] Chris Saad says: it lets you describe a set of services
for scenarios - much like bluetooth scenarios
[2:19:58 PM] Josh Patterson says: does it handle data provider
"discovery", ala the web node functioanlity?
[2:20:10 PM] Josh Patterson says: er, "web inode"
[2:21:41 PM] Chris Saad says: a user could use it to describe, for
example:
For Authentication i use
1. OpenID from Engagd.com
2. OpenID from claimID.com
3. Yahoo OpenAuth
In that order
For Photos I use
1. Flickr
2. Picaso
3. iDrive
[2:21:44 PM] Chris Saad says: etc
[2:22:11 PM] Josh Patterson says: ok, so it does use do data
discovery, then... huh. that makes things simpler
[2:22:22 PM] Chris Saad says: yep
[2:22:30 PM] Chris Saad says: very powerful
[2:22:41 PM] Chris Saad says: we thought we would have to invent our
own format for it - but turns out not
[2:23:08 PM] Josh Patterson says: well, if it works that way, then
most of the stack is there
[2:23:33 PM] Josh Patterson says: http://yadis.org/wiki/What_is_Yadis
[2:24:27 PM] Chris Saad says: yep
[2:24:50 PM] Chris Saad says: they seem to only think of it in terms
of identity brokers - but i think it can be used for any sort of
service provider scenario
[2:24:54 PM] Chris Saad says: photos, video, bookmarks etc
[2:27:56 PM] Josh Patterson says: hrm. ill try and find time to digest
the YADIS specs tommorrow
[2:29:16 PM] Josh Patterson says: anyone know any of the YADIS guys?
I'd like to talk to them
[2:33:14 PM] Josh Patterson says: hrm. if Yaddis works as well as i
think it can, really, all the peices are in place to pull of dhcp for
data, other than a standardized way to address "data" at the "disk
block" level so as to do an "aggregation" query
[2:34:45 PM] Chris Saad says: Is it DHCP or DNS of data
[2:34:58 PM] Josh Patterson says: probabbly more so DNS
[2:35:03 PM] Josh Patterson says: i think DHCP sounded nice
[2:35:04 PM] Chris Saad says: DHCP assigns a dynamic IP address from
server to client
[2:35:04 PM] Josh Lewis says: I believe its a little of both
[2:35:17 PM] Chris Saad says: DNS resolves a URL to an IP
[2:35:27 PM] Josh Patterson says: possibly, when i originally said it,
i mean it moreso as an expression
[2:35:28 PM] Chris Saad says: we are talking about resolving an OpenID
to Data/Services
[2:35:43 PM] Ashley says: DHCP does include an Address to a DNS though
[2:35:56 PM] Ashley says: just saying
[2:36:05 PM] Josh Lewis says: Exactly a little of both
[2:36:11 PM] Ashley says: yes
[2:36:24 PM] Ashley says: i can c we're going to get along fabulously
Lweis.
[2:36:25 PM] Ashley says: ;)
[2:36:32 PM] Josh Lewis says: lol
[2:36:36 PM] Josh Lewis says: agreed
[2:36:42 PM] Josh Patterson says: discovery, in the end, they are both
about discovery, which is the function of the abstract concept from
file systems, "web inode", and im getting a good feeling that YADIS
will give us that
[2:37:01 PM] Ashley says: DHCP give you an identifier and a pointer to
various network services which apply to a network.
[2:37:03 PM] Josh Lewis says: I will try to read it tomorrow.
[2:37:07 PM] Chris Saad says: so... us marketing people need to know -
is it DHCP or DNS :)
[2:37:27 PM] Ashley says: since i havnt read all of the above - i cant
tell you :S
[2:37:49 PM] Ashley says: but from my vague understanding i'll vote
DHCP
[2:37:55 PM] Ashley says: which by the way stands for :
[2:37:57 PM] Josh Lewis says: We will work that out this week. As well
as what to call the different abstraction layers
[2:38:05 PM] Josh Patterson says: yeah, i think thats part of YADIS's
problem, no one knows what the hell it is, so it hasnt gotten big
uptake, plus its only one layer of an abstraction, so it needs to be
"shipped" in a working stack
[2:38:07 PM] Ashley says: Dynamic Host Configuration Protocol
[2:38:29 PM] Josh Patterson says: Dynamic Data Configuration Protocol?
[2:38:32 PM] Josh Patterson says: DDCP?
[2:38:34 PM] Josh Patterson says: lol
[2:41:14 PM] Josh Patterson says: well, from a marketing standpoint,
none of them are out of the box hits, but at the same time, you mesh
great utility with an ok-to-decent name, and it wil take off
[2:41:46 PM] Josh Patterson says: yeah, i like the kooky-old school
names from the linux world, lol, but at the same time, look at YADIS
-- no one knows what that is
[2:42:17 PM] Josh Patterson says: i think this stuff absolutely has to
be marketed as a "stack"
[2:42:41 PM] Josh Patterson says: YADIS by itself is like buying a car
transmission. what are you going to do with it by itself? nothing.
[2:42:46 PM] Josh Patterson says: you need the whole damn car
[2:43:01 PM] Chris Saad says: yeah but that's already covered with
datapotability.org
[2:43:11 PM] Chris Saad says: we just need to add the last remaining
blocks to the stack there
[2:43:17 PM] Chris Saad says: and give it a name
[2:43:18 PM] Josh Patterson says: yeah, definitely
[2:43:44 PM] Josh Patterson says: what i mean, though, is give
developers a "turn key" way to get on board, ie, they can download a
kit that has a prebuilt set of layers for the stack
[2:44:03 PM] Josh Patterson says: javascript apis, query translators,
sample webservices, and starter api endpoints for the identity
[2:44:10 PM] Chris Saad says: yes - agreed
[2:45:24 PM] Josh Patterson says: and me and josh talked about that
tonite, basically, use the WRFS idea as a type of "Reference Model",
and allow people to plug in a range of tech for the layers, but in the
end, it has to work just like the OSI Model for TCP/IP --- layers have
to take care of their responsibility, but can be interchanged as long
as they do the protocol
[2:46:18 PM] Josh Patterson says: anyone know Johannes Ernst ?
[2:47:22 PM] Josh Patterson says: apparently zooomr is already using
yadis... we might link floe.tv up with them as an experiment
[2:48:13 PM] Chris Saad says: Wher eis Johannes Ersnt from
[2:48:15 PM] Chris Saad says: rings a bell
[2:48:20 PM] Josh Patterson says: he did YADIS
[2:48:23 PM] Chris Saad says: OpenID uses Yadis internally somehow
[2:56:53 PM] Josh Lewis says: ok back
[3:00:40 PM] Josh Lewis says: Wow you guys have been busy. I should
take phone calls more often. This thing will get hammered out in know
time.
[3:01:57 PM] Josh Patterson says: so, actually, we are closer to "dhcp
for data" than i had previously though
[3:01:58 PM] Josh Patterson says: thought
[3:02:53 PM] Josh Patterson says: its more of executing a working
stack at this point, well, that and figuring out how to standardize
data access at the server/"disk block" metaphor --- ie, what is going
to be the "bmap" of WRFS?
[3:03:41 PM] Josh Patterson says: that, and the XML data stuff dialect
{ RDF, SIOC, XML, DOAP, OIL, ? }
[3:03:47 PM] Josh Patterson says: FOAF
[3:04:22 PM] Chris Saad says: well there are few considerations there:
[3:04:45 PM] Josh Lewis says: I think your getting a little acronym
happy there.
[3:05:06 PM] Chris Saad says: 1. Do we want do generic data blocks
only? Or do we want to treat meme types in native ways - that is a
photo is different from a video because right now there are filetype
specific services around as well as generic cloud based storage
services
[3:05:22 PM] Josh Patterson says: yeah, thats definitely an area of
consideration
[3:05:43 PM] Chris Saad says: Mime*
[3:05:49 PM] Josh Patterson says: and that might be our contribution
to the stack implementaion
[3:07:00 PM] Chris Saad says: Right now we have in our labs
1. A generic notification model for notifying about change events
2. A generic diff description model for describing changes to XML data
(such as changes to an APML file or a YADIS file - could just as
easily be a 'Image Format XML File')
[3:07:55 PM] Chris Saad says: 3. An actual implementation of an XML
versioning repository system implements those generic notification and
diff description models as well as manages permissions to who can and
can not get access to the diff stream
[3:08:23 PM] Josh Patterson says: so you want basically do a
"lifecasting" service off of something like that?
[3:09:52 PM] Chris Saad says: i am moreso thinking about.... a middle
layer that allows users to make a change anywhere, which in-turn
raises an event that notifies all the other services
[3:10:22 PM] Chris Saad says: so if i change my address in service A,
it should reflect in Service B, C and D
[3:10:42 PM] Chris Saad says: if I upload a photo to Flickr, a
notification of that should be sent to Myspace, Facebook and others
[3:10:55 PM] Josh Lewis says: Thats cool.
[3:11:01 PM] Josh Lewis says: Really cool
[3:11:27 PM] Josh Patterson says: the converse of that would be to
have the service poll via something like "dhcp for data", and refresh
every so often, but then again, i dont know how you want to struture
it --- "on demand", or "push"
[3:13:03 PM] Chris Saad says: our system can do both
[3:14:12 PM] Chris Saad says: So u change your APML file, and our
system detects the change, describes the change and then notifies all
the services in any way they like
1. RSS
2. Jabber
3. Our service calls THEIR API
4. They can call OUR API for a list of changes since [time/date]
[3:17:03 PM] Josh Patterson says: boys its been fun, but i gotta get
up and ... ugh... write more nesC code, and prep for this masters
algorithm exam. almost done. cya tommorrow