You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
[Feature] Export the pinned table's write Arrow schema through the C API #1328
I searched the issues and pull requests for a matching schema-export API.
Motivation
A C/C++ caller must construct Arrow batches matching the Paimon table's write
schema. The C API can create a table from resolved schema JSON but does not
expose its corresponding Arrow schema. Reimplementing the type mapping in
each caller can diverge from the Rust writer's validation; querying a catalog
again can also select a different table schema.
schema points to an initialized empty ArrowSchema. Reuse the core write
schema conversion and the Arrow C Data Interface, borrowing the table and
giving the output an independent release callback. Do not refresh catalog
metadata or change the core type rules. Reject null pointers and a live output
instead of overwriting it.
Anything else?
Cover type/order/nullability/field-ID mapping, a real write using the exported
schema, table/output lifetimes, release and reuse, and invalid pointers. This
proposal was prepared with AI assistance; it does not add another type mapping.
Activity