Github user rdblue commented on a diff in the pull request:
https://github.com/apache/spark/pull/23208#discussion_r239613722
--- Diff:
sql/core/src/main/java/org/apache/spark/sql/sources/v2/SupportsBatchWrite.java
---
@@ -25,14 +25,14 @@
import org.apache.spark.sql.types.StructType;
/**
- * A mix-in interface for {@link DataSourceV2}. Data sources can implement
this interface to
+ * A mix-in interface for {@link Table}. Data sources can implement this
interface to
* provide data writing ability for batch processing.
*
* This interface is used to create {@link BatchWriteSupport} instances
when end users run
* {@code Dataset.write.format(...).option(...).save()}.
*/
@Evolving
-public interface BatchWriteSupportProvider extends DataSourceV2 {
+public interface SupportsBatchWrite extends Table {
--- End diff --
`Table` exposes `newScanBuilder` without an interface. Why should the write
side be different? Doesn't Spark support sources that are read-only and
write-only?
I think that both reads and writes should use interfaces to mix support
into `Table` or both should be exposed by `Table` and throw
`UnsupportedOperationException` by default, not a mix of the two options.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
