Skip to content

Commit e0bd7bf

Browse files
committed
test: assert the partition-path parity write engages the native writer
`COMET_ICEBERG_NATIVE_WRITE_ENABLED` is off by default, so the parity test compared two JVM-written layouts and passed on the old iceberg-rust pin. Wrap the "native" insert in `assertNativeWriteEngages`, which enables the conf, enables the local-table-scan conversion the VALUES input needs, and asserts a `CometIcebergWriteExec` in the captured plan. The JVM twin now runs on the defaults instead of toggling the split-operator conf, matching the other parity tests.
1 parent f13e111 commit e0bd7bf

1 file changed

Lines changed: 3 additions & 3 deletions

File tree

‎spark/src/test/scala/org/apache/comet/CometIcebergWriteActionSuite.scala‎

Lines changed: 3 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -1187,10 +1187,10 @@ class CometIcebergWriteActionSuite
11871187
.map { case (region, i) => s"($i, '$region', $i.5)" }
11881188
.mkString(", ")
11891189

1190-
spark.sql(s"INSERT INTO $catalog.$ns.escaped_native VALUES $values")
1191-
withSQLConf(CometConf.COMET_ICEBERG_WRITE_SPLIT_OPERATOR_ENABLED.key -> "false") {
1192-
spark.sql(s"INSERT INTO $catalog.$ns.escaped_jvm VALUES $values")
1190+
assertNativeWriteEngages("escaped_native", regions.indices) {
1191+
spark.sql(s"INSERT INTO $catalog.$ns.escaped_native VALUES $values")
11931192
}
1193+
spark.sql(s"INSERT INTO $catalog.$ns.escaped_jvm VALUES $values")
11941194

11951195
// Every partition directory the native writer produced exists in the JVM writer's layout
11961196
// and vice versa, so the two tables are byte-for-byte compatible in their data locations.

0 commit comments

Comments
 (0)