Spark 3.1: fill ORC initial defaults on the row reader

[cbb330]

Summary

Opts the Spark 3.1 row ORC reader into default filling by supplying BaseDataReader's constant converter to the id-binding StructReader, so an absent field declaring an initial-default is materialized in Spark's in-memory representation.

• Mirrors the Generic reader from ORC: fill initial defaults for missing id-bound fields #263; the resolution logic itself is shared, not duplicated.
• The visitor call site in SparkOrcReader is unchanged and byte-identical to upstream — the converter is supplied by SparkOrcValueReaders.StructReader in its own constructor.
• BaseDataReader.convertConstant is made public so the reader package can reference it.
• BatchDataReader carries a follow-up note: vectorized reads are not wired, so a table whose schema declares defaults must stay on the row reader.

Provenance

Net new, like #263 — apache/main throws UnsupportedOperationException rather than reading ORC defaults. The design follows upstream's Parquet reader, where each engine supplies its own constant conversion.

Stack

1. ORC: bind struct fields by Iceberg field id in generic and Spark 3.1 readers #265 — ID-based StructReader backport (base: openhouse-1.2.0)
2. ORC: fill initial defaults for missing id-bound fields #263 — Generic ORC initial-default fill
3. This PR — Spark 3.1 ORC initial-default fill

Depends on #263's converter-aware constructor. Review this PR's own delta against its base branch chbush/oh120-orc-default-fill.

Supported scope

Spark 3.1 ORC row reader. Vectorized ORC, other Spark/Flink readers, and predicate pushdown on omitted default columns remain deferred.

Testing Done

• Spark 3.1 row tests: declared, undeclared, and nested defaults
• Spark 3.1 ORC suites pass: TestSparkOrcReaderDefaults (3), TestSparkOrcReader (13), TestSparkOrcReadMetadataColumns (6) — 22 tests, 0 failures
• Full forced rerun across the stack: 481 tests, 0 failures
• Spotless and checkstyle pass

---

Tests generated with unit-tests plugin

Made with Cursor