Shibboleth Developers

The usual logistics:

The next  Shib  design  call is  scheduled for  monday, 11/12, from 
3-4 pm est, 2-3 cst, 12-1 pst.

Phone #:  800-998-2462
Pin #:  5601277

I'm probably going to miss most or all of the call, because of a 
commitment I made before we moved the call to monday's.

During last week's design call, the cmu team suggested a new approach 
to the coding/testing phases. (Tom just posted a note summarizing 
this proposal.) Tom suggested that each team start by implementing 
the basic framework (communications, participation in message 
exchanges) and only a shell of the components. (For instance, the 
Handle Server would be hardwired to generate only a single handle, 
the XML transport wouldn't worry at all about signing, etc.) We'd 
then do integration testing of the framework.

At this point, we're likely to have something that would be close to 
demo-able. Altho it certainly couldn't be used in a real situation. 
And each of the teams would have a simple framework they could use to 
test newer versions of their components (and thus could sidestep 
having to develop their own test harnesses.)

Everyone seems pretty comfortable with this proposal. At this point, 
I'd like to explore mapping this idea  onto a timeline. And, in 
particular, trying to do the initial integration before the end of 
the year holidays, and so much vacation begins. Here's a possible 
timeline:

(11/5, 11/12)  - Complete Design (Define v0.9 and v0.91, complete 
work breakdown, develop timeline, define interfaces/APIs)

(11/12, 11/19 (short), 11/26, 12/3)  - Implement the basic framework

12/10 - Integration Testing, basic framework. (completed by 12/15)

IBM is essentially on vacation starting 12/15, for two weeks. (I 
think Sridhar mentioned that he's actually gone as of 12/12.) This 
proposal calls for us to complete the integration testing of the 
basic framework before they start their holiday vacations. This means 
that between 12/15 and 1/4 any of the teams should be able to proceed 
independently on fleshing out the implementation of the components 
they are responsible for.

After 12/17 or so, each team would proceed to flesh out and finish 
the components they are responsible for. Previously, each team has 
developed estimates of how much programming they think is involved in 
their components. I'd  like to ask people to think about how those 
estimates might be affected by this currently proposed sequence.

This would be followed by final integration testing, packaging, and a 
QA install (try installing the package, following the directions, 
BEFORE we give it to a pilot site). I'm estimating this will take two 
weeks.

AGENDA - I'd like to suggest using the phone call to give each team 
an opportunity to identify what they need from the other teams in 
order to begin integration testing on dec 10 or so. And to move 
towards determining whether each team can 1) deliver what the others 
need in a timely fashion, and 2) deliver skeletal versions of their 
own components on this schedule.

--

------------------------------------------------------mace-shib-design-+
For list utilities, archives, subscribe, unsubscribe, etc. please visit the
ListProc web interface at 

   http://archives.internet2.edu/

------------------------------------------------------mace-shib-design--