Minutes of 2026-01-14 Meeting #538

[HeikoTheissen]

1. Roll call [8:00]
   Meeting link
   Attendees:

   • George Ericson (Dell)
   • Stefan Hagen (Individual)
   • Ralf Handl (SAP)
   • Hubert Heijkers (IBM)
   • Ling Jin (IBM)
   • Gerald Krause (SAP)
   • Mike Pizzo (Microsoft)
   • Christof Sprenger (Microsoft)
   • Heiko Theißen (SAP)

   Quorum is achieved (9 out of 9 voting members)

2. Approve agenda [8:05]

   Agenda is approved.

3. Approve minutes from previous meeting(s) [8:10]
   Minutes of 2026-01-07 Meeting #537

   Minutes are approved.

4. Issues [8:15]
   Issues - OData TC

   Clarification requested: What does the spec mean by "non-null path"? #2151
   Heiko moves to merge PR Model element paths and null #2155 and close the issue as applied. Christof and Gerald second. No objections, motion passes.

   Access a property of the result of a non-composable function #400
   TC agrees that only composable functions can be indexed by key. Mike notes that Capabilities.ComputeSupported should also apply to singletons and will create an issue for that. (AppliesTo="Singleton" comprises also single-valued navigation properties, see Annotations targeting contained singletons #509.)
   Mike suggests that, in the absence of capabilities annotations, composable functions SHOULD support system query options like entity sets do, whereas non-composable functions and actions SHOULD NOT be assumed to support them. The target of these capabilities annotations can be the entity set, collection or singleton which the operation returns, or the target can start with a container and end with the function or action. He will update the PR Define Composability and use of query options for actions/functions (Fixes OData-400) #2158 accordingly. Mike suggests that actions and functions should also be targeted as schema elements, and will create a separate issue for that.
   Before making a decision, Heiko will check whether the assumed non-support of system query options after non-composable functions and actions is a breaking change for SAP's implementation.

   It is not clear if the entities in a collection returned by a function always have an entity-id #376
   Heiko points out that Inconsistencies for transient entities handling in JSON format #1928 introduced the possibility for a function to return generated entity ids odata:transient:{some-generated-identifier-unique-within-the-response} in order for entities to reference one another. Mike proposes to allow functions to return non-persisted entities (whose context URL is a type, not a collection) with ids that are stable across invocations of the function with the same parameters. This would allow clients to permanently address these entities through the function invocation followed by a key segment. Discussion will be continued in the next meeting.

5. Next meetings [9:50]

   • Wednesday, January 21, 2026, during 8-10 PST (17:00-19:00 CET) - Mike, Gerald (and George) out, Ralf will chair
   • Wednesday, January 28, 2026, during 8-10 PST (17:00-19:00 CET)
   • Wednesday, February 4, 2026, during 8-10 PST (17:00-19:00 CET) - Heiko out
   • Wednesday, February 11, 2026, during 8-10 PST (17:00-19:00 CET)

6. AOB and wrap up [9:55]
   Meeting is adjourned.
   Find these minutes also at Minutes of 2026-01-14 Meeting #538 #2157

[HeikoTheissen]

Permalink