May 20-21 Seattle F2F Agenda

Higgins face-to-face meeting in Seattle, Washington, May 20-21, 2008.

Logistics

• Location:
• Time: The event will start Tuesday at 9:00 AM and end Wednesday at noon.
• Action Items are indicated by the presence of "(AI)" on a line and the owner of the action item is in bold.
  • (AI) Jim will turn the action items into bugzilla entries as appropriate

Expected Attendees

1. Jim Sermersheim - Novell
2. Paul Trevithick - Parity (by phone on Tuesday, in person on Wed morning)
3. Drummond Reed - Cordance/Parity
4. Michael McIntosh - IBM

Attending by phone

1. Bruce Rich
2. Paul Trevithick (tue)
3. Tom Doman (tue)
4. Daniel Sanders (wed)

Participant Passcode: 685641
US toll free: 877-421-0038
US: 770-615-1254

Attending on IRC

1. Tom Doman

Tuesday

We will track at least the agenda on the #Higgins IRC channel. If you wish to call in for an agenda item, please let us know on the #higgins IRC channel

9:30-10:00 Welcome, Introductions, Agenda [Mike, Jim]

• Introductions
• Eclipse ground rules
• Agenda bashing

10:45 - 11:15

• M1 Status
  • Complete -- release is tagged complete
  • Two components were failing the build.
    • Duane fixed the failure introduced by vkokhan
    • (AI) Brian will have Peter review these
• Higgins_1.1M2 Planning
  • 20080613 is target date
  • [1] is the project plan page
  • How should we deal with the seemingly snowballing effect we get with simply pushing unaddressed items from release to release?
    • Maybe make a milestone called "backlog" or something. That way we can keep bugs associated with a release to a reasonable number, and more well prioritized.
      • (AI) Brian will create this release tag
      • (AI) All People should put all their bugs into that backlog that aren't tied to a specific release
• Automated Testing Process Planning
  • Eclipse has some test reporting tools
  • Eclipse also has servers available for testing
  • Need to set up at least one example of a project that does nightly testing
    • (AI) Mike volunteers to help with this, but needs guidance.
      • (AI) Paul will find out how other eclipse projects do this (starting with Bjorn)
    • Where do the test projects reside?
      • Mike proposes they should go under test (peer to plugins)
        • (AI) All People should move their test projects over there

11:25 - 11:35 Development Processes

• Mike suggested we need more doc about our development processes
  • unit tests --the .test projects should be in the "test" folder in SVN (not where they currently are)
  • who gets notified when a nightly build fails? (where is this documented as to how this is changed)
  • starting to use the voting mechanisms in Bugzilla to prioritize
• Here's all the doc we have now: Higgins Committers and Contributors

11:35 - 12:00 SAML2

• SAML 2.0 IdP (Markus and Paul?)
  • (AI) Mike will set up a phone call with Markus, Daniel, Mike, and his colleague.

1:30 - 3:10

• Access Control in IdAS
• http://dev.eclipse.org/mhonarc/lists/higgins-dev/msg04472.html David's email
  • Interested in discussing: (Jim, Mike, Bruce, Paul, Drummond, David Kuer-McLaren, Tom, Duane)
  • AuthZ as Entities
  • (AI) Jim update wiki notes (Done 20080323. See Access Control Teleconf 20080520)

3:15 - 4:30

• Prioritization of IdAS backlog
  • Need AuthZ for R-Cards
  • Complex attrs tends to change many of the other backlog items, so let's get it done first.
• IdAS: Complex values as Entities
  • (AI) Jim update wiki notes (20080602 Done)
  • (AI) Jim update Paul with decisions. Done

4:30 - 5:00

• Coming up with global identifiers for all idas elements
  • Paul, Markus, et al. came up with a UDI which includes context ID, Entity ID, and Attribute ID.
    • Can we extend this so we can address:
      • a specific attribute value?
      • an element of a specific complex attribute?
      • element through an attribute value which is a reference to another entity (and lower)
  • Drummond talked about Cool URIs
  • (AI) Paul update us on the status of the UDI spec.

Wednesday

9:00 - 9:45

• ISIP Interop Issues

9:45AM...

Review access control discussion from yesterday:

• Conclusion: Both subject and resource must be either some kind identifier (Entity or Attribute) OR an expression
  • Resource Identifiers (entire Entity or an Attribute and all values):
    • http://www.parity.com/spec/udi/udi-syntax.html provides a way to identify either an Entity (called a Resource) or an Attribute --but what we mean by Attribute is the Attribute-type and all of the values of the attribute
      • e.g. GET means all values can be read
      • e.g. ADD means a new Attribute and at least its min-cardinality set of values (remember in Higgins min-cardinality = 1)
      • e.g. DEL means that all values can be deleted in one operation OR any subset of the set of values can be deleted
      • e.g. MOD means that any or all values can be modified
  • Expresssions: Perhaps we should allow wildcards & and regex-like patterns in the UDI ?
    • allows partial specification of a UDI
    • Allows one "UDI-expression" to resolve to multiple UDIs
    • ...this is a bit like UDI XPath; should also have a look at SPARQL; the text representation of IdAS Filters

11:00-11:30 Complex Valued Attributes

Conclusions from yesterday

• Values of attributes can be:
  1. XSD Type Literal
  2. IBlankEntity or subclass
  3. (Context, Entity, or Attribute) UDI Literal
  4. Local EntityId String Literal
• IAttributeValue.getValue() will have a new param or something that will, as a convenience, attempt to dereference the Context, Entity or Attribute identifier

Ideas to put on the agenda

• IdAS Model as Entities
• IdAS Changing metadata to attributes