FaunaDB FQL v10 UDFs and client driver: serializability guarantees across regions

Designing a multi-region read path that invokes server-side User Defined Functions through a client driver requires clarity on where consistency is enforced.

FaunaDB provides a distributed transactional model with strict serializability and exposes FQL via a REST-based API that language SDKs abstract. UDFs encapsulate logic on the server side, while the document and index model governs retrieval cost. FQL syntax differs between legacy v4 and v10, and index design affects compute usage.

For an integration that relies on the documented serializable guarantees, it is unclear how the driver surfaces transaction metadata and whether any client-side configuration can affect the consistency applied to a UDF invocation.

Does the REST response for a UDF call include headers or fields that indicate the serializability level applied? Is the consistency model for UDF execution fixed server-side in FQL v10, or can driver options influence it? How should index-based retrieval inside a UDF be reasoned about with respect to global read latency?