[Openid-specs-risc] Call notes

Atul Tulshibagwale atul at sgnl.ai
Wed Aug 30 00:51:27 UTC 2023


Hi all,
Here are the notes from today's call. They are also stored here
<https://hackmd.io/@oidf-wg-sse/wg-meeting-20230829>:

Atul

-- 

<https://sgnl.ai>

Atul Tulshibagwale

CTO

<https://linkedin.com/in/tulshi> <https://twitter.com/zirotrust>
<atul at sgnl.ai>

WG Meeting: 2023-08-29
<https://hackmd.io/nW6HWJ4WQfaX2ZAnm758Uw?view#Agenda>Agenda

   - Transmitter’s ability to update stream content
   <https://github.com/openid/sharedsignals/issues/103>
   - Empty events_requested field in Create Stream
   <https://github.com/openid/sharedsignals/pull/100>
   - Decoupling SSF metadata from OAuth
   <https://github.com/openid/sharedsignals/pull/101>
   - Conclude on format PR <https://github.com/openid/sharedsignals/pull/87>

<https://hackmd.io/nW6HWJ4WQfaX2ZAnm758Uw?view#Attendees>Attendees

   - Atul Tulshibagwale (SGNL)
   - Phillip Hunt (Independent ID)
   - Steve Venema (Forgerock)
   - Apoorva Deshpande (Okta)
   - Yair Sarig (VMWare)
   - Edumund Jay ()
   - Tim Cappalli (Microsoft)

<https://hackmd.io/nW6HWJ4WQfaX2ZAnm758Uw?view#Notes>Notes
<https://hackmd.io/nW6HWJ4WQfaX2ZAnm758Uw?view#Transmitter%E2%80%99s-ability-to-update-stream-content>Transmitter’s
ability to update stream content
<https://github.com/openid/sharedsignals/issues/103>

   - (Apoorva) I want to make sure that Transmitters have the ability to
   change the subjects included in the stream
   - (Atul) Should the Transmitter have to tell the Receiver that it honors
   an Add Subject
   - (Phil) That could lead to harvesting information from the Transmitter
   - (Yair) If I create a new stream and don’t make any add subject calls,
   does that mean all subjects are included are no subjects are included?
   - (Phil) The initial idea was that as a part of an initial authorization
   request, such add subject calls are required. In other cases, e.g. SCIM,
   all subjects are included by default
   - (Atul) I agree we don’t need to put strong normative language in the
   semantics of the Create Stream
   - (Phil) It could be a Transmitter discretion whether or not they should
   indicate whether an AddSubject succeeded or not
   - (Yair) There may be additional fields that indicate to the Transmitter
   whether all subjects should be included or not
   - (Phil) We could add a filter at the time of stream creation
   - (Steve) There’s a concrete use case for this: Our infosec would like
   to know about any “forgerock.com” identity for a subset of events that
   Google could notify us for. In our case, “forgerock.com” is not a
   separate tenant of Google
   - (Phil) Would a Receiver want to know the total universe of subjects in
   the stream?
   - (Apoorva / Steve) That sounds a bit like SCIM
   - (Steve) The filter seems to have value
   - (Phil) There are possibly two use cases:
      - User centric: User needs to consent to share events. The spec is
      well designed for this.
      - Enterprise use case: Consent belongs to the enterprise, so every
      account using a specific service must be included in the stream
   - (Phil) Does this lead to an interop issue
   - (Phil) In a typical SCIM use case all subjects are included by
   default. This is more of an admin issue than an interop issue
   - (Apoorva) How would a transmitter limit the audience of a stream to
   only a certain number of users
   - (Atul) Two issues here:
      - What is the semantics when the stream is created
      - How does the Transmitter add / remove subjects
   - (Yair) Filtering may be useful in many cases
   - (Phil) What if we just spec-ed that the Transmitter is always
   authoritative over what subjects it sends events about. The Transmitter may
   add or remove subjects as it desires.
   - (Phil) We could leave the spec unchanged, because it can be configured
   administratively
   - (Phil) We could leverage the SCIM filter language that is based on
   path. But what are you filtering against.
   - (Phil) Issue with adding a filter e.g. email ends with @okta.com, what
   is the Transmitter comparing that against?
   - (Steve) Events won’t be sent if they don’t match the filter
   specification
   - (Phil) We don’t want to force the Transmitter to look up a database
   before sending an event
   - (Yair) The device use cases, the id is only a device ID, and the
   Transmitter may not have this information
   - (Atul) Should we list this as a “known issue” in the spec?
   - (Phil) agree, but we should make sure that any normative language
   about subject content in the stream should be avoided. This will enable us
   to craft a more specific solution later that doesn’t break backward
   compatibility
   - (Phil) Even SCIM is trying to define how to define the scope of each
   domain

<https://hackmd.io/nW6HWJ4WQfaX2ZAnm758Uw?view#Decoupling-SSF-metadata-from-OAuth>Decoupling
SSF metadata from OAuth <https://github.com/openid/sharedsignals/pull/101>

   - (Atul) Value of the “scheme” field can be the URN
   -

<https://hackmd.io/nW6HWJ4WQfaX2ZAnm758Uw?view#Empty-events_requested-field-in-Create--Update-Stream>
Empty events_requested field in Create / Update Stream
<https://github.com/openid/sharedsignals/pull/100>

   - (Apoorva) What would be the use of creating a stream with no
   “events_requested”
   - (Phil) There’s a way around this: There are three params:
      - events_requested
      - events_supported
      - events_delivered
   - (Phil) If events_requested is missing, events_delivered should remain
   unaltered?
   - (Phil) If events_requested is not asserted, then it means that the
   Receiver is not specifying which events
   - (Phil) You could be requesting events A, B, and C. But you’ve only
   paid for A and B, so the Transmitter only includes A and B in the stream
   - (Apoorva) In the current spec, the events_delivered is a pure
   intersection of event_requested and events_supported
   - (Phil) We then need to update this “intersection” language
   - (Phil) supported describes technical capabilities, delivered means
   what you actually get
   - (Atul) supported could also mean just what is desired

<https://hackmd.io/nW6HWJ4WQfaX2ZAnm758Uw?view#Action-Items>Action Items
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openid.net/pipermail/openid-specs-risc/attachments/20230829/74bff50c/attachment-0001.html>


More information about the Openid-specs-risc mailing list