[wsf-dev] Face-to-Face Recap

[Curtis]

Hi All:

I thought I'd recap this week's face-to-face meeting in DC for those who
were not able to attend.

The meeting was held on Wednesday 23-May-07 in a very nice conference
facility in the same Georgetown University building that houses Chad's
office.  We got started at 10:00 with the following attendees:

- Asa Hardcastle
- Curtis Jones
- Conor Cahill
- George Fletcher
- Chad La Joie

It was a free-flowing dialogue that included the following subjects (in
sequence):

1) Asa began a discussion of the somewhat parallel activities occurring in
the (very active) OpenID project and we explored the technical and community
issues around embracing OpenID under ID-WSF.  George mentioned some early
integration/translation work done by Paul a few years ago, but that has
apparently been on the shelf for some time.  We ultimately agreed that there
would definitely be value in having some support for/integration with
OpenID, but that we should focus our immediate energy on getting the working
WSC toolkit (with SAML bootstrapping) and contemplate OpenID afterwards.

2) We briefly discussed some tactical steps required to get the tangible
implementation fully rolling.  We talked about the chicken/egg scenario
around the WSP/WSC and some ideas for leveraging existing code (e.g.,
Conor's) to get past this.

3) Next, we talked about XML tooling and went into some depth around
marshalling and unmarshalling and the appropriate granularity of elements to
be marshalled/unmarshalled.  Chad explained some of the flexibility
available to the developer using his tooling to decide the appropriate
granularity for the application at hand.  We discussed the somewhat
tedious/burdensome nature of building the required sets of methods for each
element to be marshalled/unmarshalled -- and some future ideas for
minimizing this.  We also talked briefly about the required WSC-interface
"glue" that will be necessary for any new WSP -- and that it would seem
logical that the developer of the WSP would be responsible for at least a
basic implementation of this glue.

4) Next, at my request, Conor walked us through his C++ client
implementation (file/folder organization, source, scripts, environment
variables, etc.) and his Java service implementation.  This was enlightening
and segued into other subjects, including:

a) A discussion of library organization.  There seemed to be agreement on
exposing each distinct ID-WSF step at the WSC library level, and then
providing a high-level/abstracted interface that would likely bundle some of
these steps into logical "chunks".

b) A discussion of code generation techniques and their pros & cons,
including WSDLs and how they are supported/not supported in various
frameworks (e.g., Axis).

5) Finally, we had Chad give us a similar walk-through of his XML tooling
code/environment.  We discussed the 3 primary projects we'll be focusing on:
opensaml2, openws, and xmltooling -- with detailed discussion on the latter.
We talked about how data is fully separated from logic, with XML objects
simply serving as data stores and logic residing above this.  This
conversation also segued into ancillary discussions, including one on
containers and their related issues (e.g., related bugs in Axis).

At this point, we had reached 4:00 (having continued our meeting through
lunch) at which point we wrapped up.  Conor had a hard stop as he was about
to embark on a long drive to Florida.  To the other participants, please
feel free to make corrections and/or fill in any omissions in my summary.

All in all, it was a very productive and enlightening meeting.  Many thanks
to everyone who participated -- and particularly to Asa for pulling this
together and leading the agenda -- and to Chad for graciously hosting the
meeting.

- Curtis

Curtis Jones
curtis at upto11.com

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.openliberty.org/pipermail/wsf-dev/attachments/20070525/73136ff8/attachment.html