Attendees:
- Benjamin Oshrin
- David St Pierre Bantz
- Mark Rank
- Alan Buxey
- Keith Hazelton
- Albert Wu
- Additional ACAMP Participants, including Jon Miner, Scott Cantor, and several other attendees
Agenda
- Because this was an unconference session, there was no formal agenda. An ad hoc discussion suggested the following topics:
- Direction
- Status of Schema & Attribtues
- Work plan?
Notes
The notes from the ACAMP session are available here. Because a quorum was reached (5/7 current Schema Board Members attended) this session was declared an official Schema Board meeting, and the notes from the session are copied herein.
Topic 1: Direction → Not many requests.
- There’s a lack of organizational regularity to the schema board right now.
- Albert is willing to step up and help; considers what this group does as core / important;
- Get board back into a monthly / bi-monthly cadence (after holidays)
- Do this before TIIME (i.e. in January)
- Some logistics to do on getting Albert involved (wiki access, etc).
Topic 2: Status
- The schema board looks after several schemas - eduPerson, SHAC, VoPerson - there’s a bunch, some not even used anymore but still referenced
- e.g. no demand for new version of eduPerson;
- Big directory of values in eduPerson that aren’t used any more; can we clean some of this up?
- Is it time for eppn
- Albert: as we begin the journey of requiring standard attributes, “how to use”, “here they are” - some confusing things:
- multiple schemas;
- some attributes not yet included
- binding various protocols - some agnostic, and then specific protocol delivery
- Would be great if schema board took these problems on
- Schema board doesn’t need to set the MUST and MUST NOT, but can be enforced by implementer (I2)
- Would like to document from a protocol-specific lens; no ambiguity;
- Probably need a pure semantic description of elements, e.g. schema.org-like description of the elements
- Separating attribute definitions from schema definitions: there is a start on this already
- Deployers may not deal with abstraction very well.
- Behind the scenes, an abstraction may be helpful;
- Some disagreement here - esoteric concept may not be helpful; others believe the underpinnings need this.
- eduPerson as an example of the problem space
- Some agreement on identifiers, but that is thin; names? These are semantic: preferred name, legal name, etc - these are not protocol specific; they are a construct.
- Agreement met on what this means when understood that concepts need to be represented in data. e.g. defining what “Legal Name” means. → have come to “violent agreement” then.
- Note - this is required for automatic protocol translations.
- Example of problem space: OIDC claim email
- include in the schema how claims are intended to be used
- We don’t want these schemas to organically evolve - we want to guide this.
- Mapping exercise important / helpful, and where things pre-exist, don’t re-invent the wheel.
- OpenID space: we want to include mappings already in use to encourage adoption
- eduPerson essentially becomes the ldap-specific implementation
- What’s the prospect of merging the three specs?
- Usually these conversations result in “the less we touch them the better”
- Socially there are different communities - this is someone independent of different protocols, but you tend to get a different protocol in a different community
- There’s still some notion that eduPerson is US thing, SHAC is European thing
- Branding is important (i.e. communicating what’s happening, what these are)
- REFEDS data schema, REFEDS attribute registry (protocol specific)
- The public facing things are all protocol specific
- Is “name” an insoluble problem?
- Perhaps only “name” is the most difficult
- Many others seem reasonable to accept what exists?
- Some requests for new “name” values, but then when asked back what those were, no one responded for clarification …
- Values registry: there will be a need, but perhaps next gen
- Help guiding important e.g. from affiliation -> entitlement
- Discussion of library space
- Discussion of library space
- Multi-protocol → moving from SHAC to OpenID, need strong definitions;
- Allow for transforms
- Applications should handle localization
- Note: if we deprecate eppn, we deprecate R&S
- Interrelationships between specs and profiles: they depend on each other
- Much of this is about perception and organized change.
- So who is stewarding the migration - on a global level.
- REFEDS, eduGain …
Topic 3: next steps
- Albert / Ben → Flywheel; structural work
- Meeting in January
- TIIME, meetings
- Resources
- mailing list
- EduGain Slack
- REFEDS wiki
- (Slack? but not accessible)