Notes from the
Higgins Developers call on Thursday, August 28
Attendees
=========
* Duane Buss - Novell
* Brian Carrol - Serena
* Tom Doman -
Novell
* Andy Hodgkinson - Novell
* David
Kuehr-Mclaren - IBM
* Mike McIntosh - IBM
* Dale
Olds
- Novell
* Ernst Plassmann - IBM
* Drummond
Reed -
Cordance
* Bruce
Rich - IBM
* Mary
Ruddy
- Meristic/SocialPhysics
* Markus Sabedello - Parity
* Jim Sermersheim - Novell
* George Stanchev -
Serena
* Paul
Trevithick -
Parity/SocialPhysics
*
Brian
Walker -
Parity
* Mohamad Raji
Gani - Oracle
Agenda
1. [Brian] We are now
working on 1.1M4.
- [Brian] Milestone 4 is currently scheduled for
September 5th
- [Mike] Can we talk about
moving this out? Only 5 weeks since last build and we lost a week due to thee
SVN outage. I think it is more
important to have the items in the milestone than to stick to the date we sent
months ago.
- [Brian] That
comes…
- [Mike] I’m completely
focused on SOAP 1.2 and WS-Trust
1.3. I need more time for
testing, I wouldn’t feel comfortable having the milestone build go out without
more testing.
- [Brian] What is your plan
for completing the code?
[Mike] I want to check in my changes over the
weekend. Need time to test and
fix any problems
- [David] Mike are you
suggesting a week?
- [Mike] One week is
probably about right.
- [Brian]
OK
- [Mike] Depends on what
kinds of problems show up.
- [Brian] So push it out 1
week and then stay flexible based on feedback. I will take an action to update the
plan accordingly. We still have
56 items on the list. People need
to complete or move out (to m5) their items.
- [Action item] Brian to
update the Plan.
2. [Brian & David] Internationalization
- [Brian] I did send a note out. Haven’t heard anything yet. So If don’t hear, need to select a
victim.
- [David] There is a
function that allows you to identity strings. I can have links posted to how that
works in a week or so, but there is no automated way I can
find.
- [Brian] So tool
available, but manual in nature.
- [David ]
Yes…
- [Action item] David to
post link to tool.
3. [Mary] Microsoft
& IBM have published the
WS-Addressing Identity Extension
- See http://self-issued.info/?p=82.
- Can anyone from
IBM comment on this?
- [Mary] Microsoft has been
busy publishing two documents.
The first of these is the WS-Addressing Identity Extension. Can anyone from
IBM comment on
it?
- [Mike]It was initially
published 2.5 years ago and references by ISIP 1.0. It contains elements that were defined
in it. It is just a slight
revision.
- [Mary] So all is
fine?
- [Mike] No
impact.
- [Bruce] So while
WS-Policy talks about types of credentials, this also allows instance you’re
looking for. Particular certificate that matches what you want. It is a way to
crisp up that sort of thing. Not
a lot of interaction with infocard stuff.
4. [Mary]
ISIP v1.5 is published and
under the Microsoft OSP
- See http://self-issued.info/?p=80
- We have three selectors
to update to the new PPID algorithm:
- GTK/Cocoa – Andy [Done]
- AIR Selector – Brian &
Jeesmon
- RCP Selector – Tie Li,
Mike
- [Mary]The second item is
version 1.5 of the ISIP. The agenda talks about updating the
three selectors, but more importantly the STS’s need to be
updated.
- [Mike] I think we need to
make this code one shared service.
We tried before. I will
take a look at that when done with the current stuff.
- [Brian] Mike can we
tentatively plan from a milestone point of view?
- [Mike] There is a
critical path problem. If try to put it into M4, need two weeks, not one
week. I’m fine with that. Waiting
6 weeks is probably a long time.
- [Brain] That’s what we
thought on our side. We need to make appropriately server changes on our side.
So it sounds like people would be amenable to supporting the PPID change in
M4.
- [Dale] Wouldn’t it effect
all the selectors as well?
- [Mike] I agree as well.
The issue is that on some of the client side stuff, there is an embedded
STS.
- [Paul] Andy didn’t you
say that you had already done this?
- [Andy]
???
- [Paul] Are there plans to
make other changes in the RCP selector or not?
- [Mike] Yes. Whenever I say
STS, there is a client side
piece and the server side piece.
- [Paul] Brian, is there
anything needed on the AIR
selector?
- [Brian]
No.
- [Mike] I don’t fully
understand the AIR selector architecture. I
guess it uses a service to generate the STS
request.
- [Paul, Jeesmon]
Yes.
- [Mike] So you would need
to change that?
- [Jeesmon]
???
- [Paul] Mike to your point
about not understanding the AIR selector, that is partly
because the architecture diagram was missing up to the last 24 hours. It is now on the
wiki.
- [Paul] So, it sounds like
we will delay M4 to get these changes in also.
- [Brian] I will create
tickets.
- [Paul] The reason we were
quiet about this, is it was under NDA.
That is why we haven’t had discussion. Now we can create the bugzilla items,
etc.
- [Brian] So I will create
the bugzilla items for M4.
- [Paul] That is
great.
- [David]Before we go on,
I’d like not to see M4 moved out too far. Trying to sync up on
internationalization and bug fixes deferred from M3.
- [Brian] For now will put
a stake in the ground date of September 19.
- [David]
Thanks.
- [Action item] Brian to
enter bugzilla items.
5. [Mary] Eclipse
Summit Europe
- November 19-20 in
Ludwigsburg
Germany
- The program committee
requested/invited a talk about Higgins
- Any volunteers
?
- [Mary] Eclipse is having
a Summit on November 19-20 in
Germany. Eclipse has requested
that there be a Higgins presentation. Is there anyone who could give a
Higgins presentation in
Germany?
- [Silence]
- [Paul] We could volunteer
Markus.
- [Markus] It is not far
away.
- [Paul] We can talk off
line.
- [Drummond] He is a good
presenter too.
- [Action item] Paul, Mary,
Markus to arrange.
6. [Mary] Burton Group
ISWG (Identity Systems Working Group)
- Thoughts from companies
involved in the initiative?
- [Mary] The Burton Group
has formed an identity services working group that involves mostly banking
customers in North
America. There are use cases that they think
their existing vendors can’t serve.
Though it is at the early stages, they need an abstraction layer, it
sounds something like Higgins. There is a face-to-face
meeting in Toronto next week, that I don’t
have budget to attend. Has anyone
on the call been involved?
- [Dale] I’ve not been. But
are attempting to squeeze in
- …..
- [Mary] I was hoping that
some of the folks from a couple of other big companies would be on today’s
call.
- [Dale] It is really exciting – to hear what
the customer needs and have the vendors shut-up.
7. [Mary] Next Higgins
F2F
- [Mary] - Boston is the winning location,
with London second. It would be great if one of the
companies with offices in the area could provide us a room for the
meeting.
- [Mike] Maybe it would be
best to pick Boston and doodle for possible
dates.
- [Mike] Part of my
response will be to figure out when I and an office can be
available.
- [Mary] I will create a
doodle.
- [Action item] Mary to
create doodle.
- [Action item] Mike to
find room.
8. [Markus]
Relationship between Context ID and Context UDI
- [Mary] The next item is
the relationship between Context ID and Context UDI.
- [Markus] This topic was triggered by Jim a week
ago, by a question to the list.
Now I’ve checked in a new project that can do all this… Now we have two
ways to instantiate a context….
The other way is the configuration API. I think we don’t need iContextID
anymore.
- [Jim] I don’t see any
problem. One of us should try it
out locally. We have examples and JUNIT tests already set up. Someone would
need to do a pass to see what changes would have to be made.
- [Markus] Create context
method change needs to be done on the individual context factories. Wondering if anyone sees any problems
with that?
- [Jim] It sounds
reasonable to me. I haven’t
looked at all of the code yet.
So that would be my next thing.
I can’t think of any problems off the top of my head. That doesn’t mean there aren’t
any.
- [Markus] I don’t think
there would be much effect. Most
configure a context by … I think we should just try it out.
- [Jim]
Alright.
- [Markus] What does this
mean for an application with an entityID that is only unique for the context,
when they want to know the identifier of the context to …
- [Jim]
getContextID?
- ……
- [Jim] Right now it is a
URI. We didn’t push any changes in
from our last conversation.
- [Markus] I think it
should be a string. It could be a
URI.
- [Jim] Do we have use
cases where URI is too
restrictive?
- [Paul] I think string is
what we want there.
- [Drummond My recollection
is we agreed on string, and URI is a subset of
that.
- [Jim] We need to make
sure the Higgins data model, that is how
it is represented there, and IdAS should follow that.
- [Paul] I will double
check. It is my understanding
that that has already been changed…
- [Jim] Right now we need
to figure out how getContextFactory from the context registry using… Do we
need to change the registry getContextFactory method to take a
string….
- [Markus]
Right.
- [Paul] We added a link to
your wiki page to the agenda. Then Markus responded in an email. I haven’t
digested it yet.
- [Jim] I read it. I like
how it cleans things up.
The root is RDF.
- [Paul] I would say it
different. One of my objectives is to make the Higgins model a pure
superset. There are some
limitations in RDF I’m hedging my bet with the semantic web. There is still considerable investment
and energy going into that technology.
Even if all we did is use some of the tools coming out of it, it is a
win for us. We have been able to
clean up HOWL, and gradually correct any areas where we were a subset of
RDF. We are a pure superset
conceptually. IdAS is coming to
that from a different constraint. We aren’t the only ones who see impedance
miss-matches between Java and RDF.
So IdAS is adding value..
- [Paul] To summarize: The
Higgins model is something that
we are defining as needed as extensions to RDF semantics.
- [Drummond] So Markus is
proposing that the model description themselves would be in RDF and HOWL. Is
that right?
- [Markus]
Yes.
- [Drummond] So that is
consistent with what you said.
- [Jim] I like it and it
cleans up some of the issues I was having. ….
- [Jim] So if no one has
concerns, I can update the wiki page, go back to my drawing and throw that our
there. I can use the xml file
context provides to build out a sample CP that does this for its model. I think I can do that. It is pretty lenient. So I think I can build a context that
actually looks this way. Is anyone aware of any effort to write a generic
browser/editor tool?
- [Paul] We dream about
that at Parity. But haven’t made any progress on it.
- [Jim] I wonder if start
with an OpenLDAP tool and CP that out and add IdAS stuff.
- [Paul] Actually there are
a number of linked data browsers coming up. So it is good for us to be
compatible with linked data. That
is another way to do.
- [Jim] One other thought,
orthogonal, these models, a lot of them have a super type. If that attribute, subclass of, the
attribute value would be basically the entityID of what it is a subclass of,
if that can be de-referenced, then today could use this attribute value and
return an entity. If you look at
the attribute types, it would be nice to have another control or
extension. When you read
attributes, also include the values from the ….. Let’s say you have an
employee and they have a contact phone number…
- [Paul] So building a poor
man’s inference engine into it.
- [Paul] I highly recommend
that book The Semantic web for the Working Ontologist.
- [?] Could you repeat the
name?
- [Markus] Semantic web for
the Working Ontologist.
- [Drummond] It is easy to
find on Amazon. I’ve already got at and started reading it.
- [Paul]
OK.
- [Action item] Paul to
double check the Higgins data
model.
- [Action item] Jim to
update wiki page.
9. [Jim] IdAS Model
as Entities
10. [Mary] DIDW
OSIS Interop
- [Paul] Is there anything
we need to say? My understanding is that DigitalMe and
AIR selectors will be represented, in the
form of Parity’s version of the AIR
selector, and I don’t think IBM or anyone
is demonstrating the RCP selector.
- [Paul] That was all. Nothing
else.
End