[SPARK-56650][SDP][CONNECT] Add AutoCDC Spark Connect APIs by anew · Pull Request #55589 · apache/spark

anew · 2026-04-28T21:40:12Z

What changes were proposed in this pull request?

Add the spark-connect protos to support AutoCDC flows. This is mainly two new protos:

AutoCdcFlowDetails: similar to the existing WriteRelationFlowDetails, describes an AutoCDC flow.
SCDType an enum to specify the type of SCD (slow changing dimension) storage type.

For now, throw "not implemented" for the new flow type.

Why are the changes needed?

This new API is needed to specify an AutoCDC flow in a declarative pipeline.

Does this PR introduce any user-facing change?

It introduces a new Spark Connect API, however, no implementation yet, and also no language wrappers (e.g. Python), yet.

How was this patch tested?

No tests, as this API is unimplemented for now.

Was this patch authored or co-authored using generative AI tooling?

Generated-by: Claude Sonnet 4.6

AnishMahto

Couple small questions

AnishMahto · 2026-05-20T17:55:39Z

+      // Optional condition applied to source and target for optimizations like partition pruning.
+      optional Expression where = 4;


Let's drop this as it's not an API we've agreed on for yet in the SPIP? We can just reserve the proto number if that's important.

good point, this can be a future improvement

AnishMahto · 2026-05-20T17:59:35Z

+      // Column indicating which user columns to update or ignore.
+      optional Expression columns_to_update = 16;


Same thing here, is this an API we reached consensus on in the SPIP?

ok, we can drop this, too.

AnishMahto · 2026-05-20T18:01:05Z

+      // Whether to ignore null values in source data updates.
+      optional bool ignore_null_updates = 5;


IIRC in the SPIP we decided to drop the boolean ignore null configuration altogether in favor of the ignore null columns list?

makes sense

AnishMahto · 2026-05-20T18:06:48Z


 DESCRIPTOR = _descriptor_pool.Default().AddSerializedFile(
-    b'\n\x1dspark/connect/pipelines.proto\x12\rspark.connect\x1a\x19google/protobuf/any.proto\x1a\x1fgoogle/protobuf/timestamp.proto\x1a\x1aspark/connect/common.proto\x1a\x1dspark/connect/relations.proto\x1a\x19spark/connect/types.proto"\xa4\'\n\x0fPipelineCommand\x12h\n\x15\x63reate_dataflow_graph\x18\x01 \x01(\x0b\x32\x32.spark.connect.PipelineCommand.CreateDataflowGraphH\x00R\x13\x63reateDataflowGraph\x12R\n\rdefine_output\x18\x02 \x01(\x0b\x32+.spark.connect.PipelineCommand.DefineOutputH\x00R\x0c\x64\x65\x66ineOutput\x12L\n\x0b\x64\x65\x66ine_flow\x18\x03 \x01(\x0b\x32).spark.connect.PipelineCommand.DefineFlowH\x00R\ndefineFlow\x12\x62\n\x13\x64rop_dataflow_graph\x18\x04 \x01(\x0b\x32\x30.spark.connect.PipelineCommand.DropDataflowGraphH\x00R\x11\x64ropDataflowGraph\x12\x46\n\tstart_run\x18\x05 \x01(\x0b\x32\'.spark.connect.PipelineCommand.StartRunH\x00R\x08startRun\x12r\n\x19\x64\x65\x66ine_sql_graph_elements\x18\x06 \x01(\x0b\x32\x35.spark.connect.PipelineCommand.DefineSqlGraphElementsH\x00R\x16\x64\x65\x66ineSqlGraphElements\x12\xa1\x01\n*get_query_function_execution_signal_stream\x18\x07 \x01(\x0b\x32\x44.spark.connect.PipelineCommand.GetQueryFunctionExecutionSignalStreamH\x00R%getQueryFunctionExecutionSignalStream\x12\x88\x01\n!define_flow_query_function_result\x18\x08 \x01(\x0b\x32<.spark.connect.PipelineCommand.DefineFlowQueryFunctionResultH\x00R\x1d\x64\x65\x66ineFlowQueryFunctionResult\x12\x65\n\x14\x65xecute_output_flows\x18\t \x01(\x0b\x32\x31.spark.connect.PipelineCommand.ExecuteOutputFlowsH\x00R\x12\x65xecuteOutputFlows\x12\x35\n\textension\x18\xe7\x07 \x01(\x0b\x32\x14.google.protobuf.AnyH\x00R\textension\x1a\xb4\x02\n\x13\x43reateDataflowGraph\x12,\n\x0f\x64\x65\x66\x61ult_catalog\x18\x01 \x01(\tH\x00R\x0e\x64\x65\x66\x61ultCatalog\x88\x01\x01\x12.\n\x10\x64\x65\x66\x61ult_database\x18\x02 \x01(\tH\x01R\x0f\x64\x65\x66\x61ultDatabase\x88\x01\x01\x12Z\n\x08sql_conf\x18\x05 \x03(\x0b\x32?.spark.connect.PipelineCommand.CreateDataflowGraph.SqlConfEntryR\x07sqlConf\x1a:\n\x0cSqlConfEntry\x12\x10\n\x03key\x18\x01 \x01(\tR\x03key\x12\x14\n\x05value\x18\x02 \x01(\tR\x05value:\x02\x38\x01\x42\x12\n\x10_default_catalogB\x13\n\x11_default_database\x1aZ\n\x11\x44ropDataflowGraph\x12/\n\x11\x64\x61taflow_graph_id\x18\x01 \x01(\tH\x00R\x0f\x64\x61taflowGraphId\x88\x01\x01\x42\x14\n\x12_dataflow_graph_id\x1a\x92\n\n\x0c\x44\x65\x66ineOutput\x12/\n\x11\x64\x61taflow_graph_id\x18\x01 \x01(\tH\x01R\x0f\x64\x61taflowGraphId\x88\x01\x01\x12$\n\x0boutput_name\x18\x02 \x01(\tH\x02R\noutputName\x88\x01\x01\x12?\n\x0boutput_type\x18\x03 \x01(\x0e\x32\x19.spark.connect.OutputTypeH\x03R\noutputType\x88\x01\x01\x12\x1d\n\x07\x63omment\x18\x04 \x01(\tH\x04R\x07\x63omment\x88\x01\x01\x12X\n\x14source_code_location\x18\x05 \x01(\x0b\x32!.spark.connect.SourceCodeLocationH\x05R\x12sourceCodeLocation\x88\x01\x01\x12_\n\rtable_details\x18\x06 \x01(\x0b\x32\x38.spark.connect.PipelineCommand.DefineOutput.TableDetailsH\x00R\x0ctableDetails\x12\\\n\x0csink_details\x18\x07 \x01(\x0b\x32\x37.spark.connect.PipelineCommand.DefineOutput.SinkDetailsH\x00R\x0bsinkDetails\x12\x35\n\textension\x18\xe7\x07 \x01(\x0b\x32\x14.google.protobuf.AnyH\x00R\textension\x1a\xc0\x03\n\x0cTableDetails\x12x\n\x10table_properties\x18\x01 \x03(\x0b\x32M.spark.connect.PipelineCommand.DefineOutput.TableDetails.TablePropertiesEntryR\x0ftableProperties\x12%\n\x0epartition_cols\x18\x02 \x03(\tR\rpartitionCols\x12\x1b\n\x06\x66ormat\x18\x03 \x01(\tH\x01R\x06\x66ormat\x88\x01\x01\x12\x43\n\x10schema_data_type\x18\x04 \x01(\x0b\x32\x17.spark.connect.DataTypeH\x00R\x0eschemaDataType\x12%\n\rschema_string\x18\x05 \x01(\tH\x00R\x0cschemaString\x12-\n\x12\x63lustering_columns\x18\x06 \x03(\tR\x11\x63lusteringColumns\x1a\x42\n\x14TablePropertiesEntry\x12\x10\n\x03key\x18\x01 \x01(\tR\x03key\x12\x14\n\x05value\x18\x02 \x01(\tR\x05value:\x02\x38\x01\x42\x08\n\x06schemaB\t\n\x07_format\x1a\xd1\x01\n\x0bSinkDetails\x12^\n\x07options\x18\x01 \x03(\x0b\x32\x44.spark.connect.PipelineCommand.DefineOutput.SinkDetails.OptionsEntryR\x07options\x12\x1b\n\x06\x66ormat\x18\x02 \x01(\tH\x00R\x06\x66ormat\x88\x01\x01\x1a:\n\x0cOptionsEntry\x12\x10\n\x03key\x18\x01 \x01(\tR\x03key\x12\x14\n\x05value\x18\x02 \x01(\tR\x05value:\x02\x38\x01\x42\t\n\x07_formatB\t\n\x07\x64\x65tailsB\x14\n\x12_dataflow_graph_idB\x0e\n\x0c_output_nameB\x0e\n\x0c_output_typeB\n\n\x08_commentB\x17\n\x15_source_code_location\x1a\xff\x06\n\nDefineFlow\x12/\n\x11\x64\x61taflow_graph_id\x18\x01 \x01(\tH\x01R\x0f\x64\x61taflowGraphId\x88\x01\x01\x12 \n\tflow_name\x18\x02 \x01(\tH\x02R\x08\x66lowName\x88\x01\x01\x12\x33\n\x13target_dataset_name\x18\x03 \x01(\tH\x03R\x11targetDatasetName\x88\x01\x01\x12Q\n\x08sql_conf\x18\x04 \x03(\x0b\x32\x36.spark.connect.PipelineCommand.DefineFlow.SqlConfEntryR\x07sqlConf\x12 \n\tclient_id\x18\x05 \x01(\tH\x04R\x08\x63lientId\x88\x01\x01\x12X\n\x14source_code_location\x18\x06 \x01(\x0b\x32!.spark.connect.SourceCodeLocationH\x05R\x12sourceCodeLocation\x88\x01\x01\x12x\n\x15relation_flow_details\x18\x07 \x01(\x0b\x32\x42.spark.connect.PipelineCommand.DefineFlow.WriteRelationFlowDetailsH\x00R\x13relationFlowDetails\x12\x35\n\textension\x18\xe7\x07 \x01(\x0b\x32\x14.google.protobuf.AnyH\x00R\textension\x12\x17\n\x04once\x18\x08 \x01(\x08H\x06R\x04once\x88\x01\x01\x1a:\n\x0cSqlConfEntry\x12\x10\n\x03key\x18\x01 \x01(\tR\x03key\x12\x14\n\x05value\x18\x02 \x01(\tR\x05value:\x02\x38\x01\x1a\x61\n\x18WriteRelationFlowDetails\x12\x38\n\x08relation\x18\x01 \x01(\x0b\x32\x17.spark.connect.RelationH\x00R\x08relation\x88\x01\x01\x42\x0b\n\t_relation\x1a:\n\x08Response\x12 \n\tflow_name\x18\x01 \x01(\tH\x00R\x08\x66lowName\x88\x01\x01\x42\x0c\n\n_flow_nameB\t\n\x07\x64\x65tailsB\x14\n\x12_dataflow_graph_idB\x0c\n\n_flow_nameB\x16\n\x14_target_dataset_nameB\x0c\n\n_client_idB\x17\n\x15_source_code_locationB\x07\n\x05_once\x1a\xe4\x02\n\x12\x45xecuteOutputFlows\x12U\n\rdefine_output\x18\x01 \x01(\x0b\x32+.spark.connect.PipelineCommand.DefineOutputH\x00R\x0c\x64\x65\x66ineOutput\x88\x01\x01\x12L\n\x0c\x64\x65\x66ine_flows\x18\x02 \x03(\x0b\x32).spark.connect.PipelineCommand.DefineFlowR\x0b\x64\x65\x66ineFlows\x12&\n\x0c\x66ull_refresh\x18\x03 \x01(\x08H\x01R\x0b\x66ullRefresh\x88\x01\x01\x12\x1d\n\x07storage\x18\x04 \x01(\tH\x02R\x07storage\x88\x01\x01\x12\x33\n\textension\x18\xe7\x07 \x03(\x0b\x32\x14.google.protobuf.AnyR\textensionB\x10\n\x0e_define_outputB\x0f\n\r_full_refreshB\n\n\x08_storage\x1a\xc2\x02\n\x08StartRun\x12/\n\x11\x64\x61taflow_graph_id\x18\x01 \x01(\tH\x00R\x0f\x64\x61taflowGraphId\x88\x01\x01\x12\x34\n\x16\x66ull_refresh_selection\x18\x02 \x03(\tR\x14\x66ullRefreshSelection\x12-\n\x10\x66ull_refresh_all\x18\x03 \x01(\x08H\x01R\x0e\x66ullRefreshAll\x88\x01\x01\x12+\n\x11refresh_selection\x18\x04 \x03(\tR\x10refreshSelection\x12\x15\n\x03\x64ry\x18\x05 \x01(\x08H\x02R\x03\x64ry\x88\x01\x01\x12\x1d\n\x07storage\x18\x06 \x01(\tH\x03R\x07storage\x88\x01\x01\x42\x14\n\x12_dataflow_graph_idB\x13\n\x11_full_refresh_allB\x06\n\x04_dryB\n\n\x08_storage\x1a\xc7\x01\n\x16\x44\x65\x66ineSqlGraphElements\x12/\n\x11\x64\x61taflow_graph_id\x18\x01 \x01(\tH\x00R\x0f\x64\x61taflowGraphId\x88\x01\x01\x12\'\n\rsql_file_path\x18\x02 \x01(\tH\x01R\x0bsqlFilePath\x88\x01\x01\x12\x1e\n\x08sql_text\x18\x03 \x01(\tH\x02R\x07sqlText\x88\x01\x01\x42\x14\n\x12_dataflow_graph_idB\x10\n\x0e_sql_file_pathB\x0b\n\t_sql_text\x1a\x9e\x01\n%GetQueryFunctionExecutionSignalStream\x12/\n\x11\x64\x61taflow_graph_id\x18\x01 \x01(\tH\x00R\x0f\x64\x61taflowGraphId\x88\x01\x01\x12 \n\tclient_id\x18\x02 \x01(\tH\x01R\x08\x63lientId\x88\x01\x01\x42\x14\n\x12_dataflow_graph_idB\x0c\n\n_client_id\x1a\xc6\x02\n\x1d\x44\x65\x66ineFlowQueryFunctionResult\x12$\n\tflow_name\x18\x01 \x01(\tB\x02\x18\x01H\x00R\x08\x66lowName\x88\x01\x01\x12O\n\x0f\x66low_identifier\x18\x04 \x01(\x0b\x32!.spark.connect.ResolvedIdentifierH\x01R\x0e\x66lowIdentifier\x88\x01\x01\x12/\n\x11\x64\x61taflow_graph_id\x18\x02 \x01(\tH\x02R\x0f\x64\x61taflowGraphId\x88\x01\x01\x12\x38\n\x08relation\x18\x03 \x01(\x0b\x32\x17.spark.connect.RelationH\x03R\x08relation\x88\x01\x01\x42\x0c\n\n_flow_nameB\x12\n\x10_flow_identifierB\x14\n\x12_dataflow_graph_idB\x0b\n\t_relationB\x0e\n\x0c\x63ommand_type"\xf0\x05\n\x15PipelineCommandResult\x12\x81\x01\n\x1c\x63reate_dataflow_graph_result\x18\x01 \x01(\x0b\x32>.spark.connect.PipelineCommandResult.CreateDataflowGraphResultH\x00R\x19\x63reateDataflowGraphResult\x12k\n\x14\x64\x65\x66ine_output_result\x18\x02 \x01(\x0b\x32\x37.spark.connect.PipelineCommandResult.DefineOutputResultH\x00R\x12\x64\x65\x66ineOutputResult\x12\x65\n\x12\x64\x65\x66ine_flow_result\x18\x03 \x01(\x0b\x32\x35.spark.connect.PipelineCommandResult.DefineFlowResultH\x00R\x10\x64\x65\x66ineFlowResult\x1a\x62\n\x19\x43reateDataflowGraphResult\x12/\n\x11\x64\x61taflow_graph_id\x18\x01 \x01(\tH\x00R\x0f\x64\x61taflowGraphId\x88\x01\x01\x42\x14\n\x12_dataflow_graph_id\x1a\x85\x01\n\x12\x44\x65\x66ineOutputResult\x12W\n\x13resolved_identifier\x18\x01 \x01(\x0b\x32!.spark.connect.ResolvedIdentifierH\x00R\x12resolvedIdentifier\x88\x01\x01\x42\x16\n\x14_resolved_identifier\x1a\x83\x01\n\x10\x44\x65\x66ineFlowResult\x12W\n\x13resolved_identifier\x18\x01 \x01(\x0b\x32!.spark.connect.ResolvedIdentifierH\x00R\x12resolvedIdentifier\x88\x01\x01\x42\x16\n\x14_resolved_identifierB\r\n\x0bresult_type"I\n\x13PipelineEventResult\x12\x32\n\x05\x65vent\x18\x01 \x01(\x0b\x32\x1c.spark.connect.PipelineEventR\x05\x65vent"t\n\rPipelineEvent\x12\x38\n\ttimestamp\x18\x01 \x01(\x0b\x32\x1a.google.protobuf.TimestampR\ttimestamp\x12\x1d\n\x07message\x18\x02 \x01(\tH\x00R\x07message\x88\x01\x01\x42\n\n\x08_message"\xf1\x01\n\x12SourceCodeLocation\x12 \n\tfile_name\x18\x01 \x01(\tH\x00R\x08\x66ileName\x88\x01\x01\x12$\n\x0bline_number\x18\x02 \x01(\x05H\x01R\nlineNumber\x88\x01\x01\x12,\n\x0f\x64\x65\x66inition_path\x18\x03 \x01(\tH\x02R\x0e\x64\x65\x66initionPath\x88\x01\x01\x12\x33\n\textension\x18\xe7\x07 \x03(\x0b\x32\x14.google.protobuf.AnyR\textensionB\x0c\n\n_file_nameB\x0e\n\x0c_line_numberB\x12\n\x10_definition_path"\x97\x01\n$PipelineQueryFunctionExecutionSignal\x12!\n\nflow_names\x18\x01 \x03(\tB\x02\x18\x01R\tflowNames\x12L\n\x10\x66low_identifiers\x18\x02 \x03(\x0b\x32!.spark.connect.ResolvedIdentifierR\x0f\x66lowIdentifiers"\xf0\x02\n\x17PipelineAnalysisContext\x12/\n\x11\x64\x61taflow_graph_id\x18\x01 \x01(\tH\x00R\x0f\x64\x61taflowGraphId\x88\x01\x01\x12,\n\x0f\x64\x65\x66inition_path\x18\x02 \x01(\tH\x01R\x0e\x64\x65\x66initionPath\x88\x01\x01\x12$\n\tflow_name\x18\x03 \x01(\tB\x02\x18\x01H\x02R\x08\x66lowName\x88\x01\x01\x12O\n\x0f\x66low_identifier\x18\x04 \x01(\x0b\x32!.spark.connect.ResolvedIdentifierH\x03R\x0e\x66lowIdentifier\x88\x01\x01\x12\x33\n\textension\x18\xe7\x07 \x03(\x0b\x32\x14.google.protobuf.AnyR\textensionB\x14\n\x12_dataflow_graph_idB\x12\n\x10_definition_pathB\x0c\n\n_flow_nameB\x12\n\x10_flow_identifier*i\n\nOutputType\x12\x1b\n\x17OUTPUT_TYPE_UNSPECIFIED\x10\x00\x12\x15\n\x11MATERIALIZED_VIEW\x10\x01\x12\t\n\x05TABLE\x10\x02\x12\x12\n\x0eTEMPORARY_VIEW\x10\x03\x12\x08\n\x04SINK\x10\x04\x42\x36\n\x1eorg.apache.spark.connect.protoP\x01Z\x12internal/generatedb\x06proto3'
+    b'\n\x1dspark/connect/pipelines.proto\x12\rspark.connect\x1a\x19google/protobuf/any.proto\x1a\x1fgoogle/protobuf/timestamp.proto\x1a\x1aspark/connect/common.proto\x1a\x1fspark/connect/expressions.proto\x1a\x1dspark/connect/relations.proto\x1a\x19spark/connect/types.proto"\xbb\x32\n\x0fPipelineCommand\x12h\n\x15\x63reate_dataflow_graph\x18\x01 \x01(\x0b\x32\x32.spark.connect.PipelineCommand.CreateDataflowGraphH\x00R\x13\x63reateDataflowGraph\x12R\n\rdefine_output\x18\x02 \x01(\x0b\x32+.spark.connect.PipelineCommand.DefineOutputH\x00R\x0c\x64\x65\x66ineOutput\x12L\n\x0b\x64\x65\x66ine_flow\x18\x03 \x01(\x0b\x32).spark.connect.PipelineCommand.DefineFlowH\x00R\ndefineFlow\x12\x62\n\x13\x64rop_dataflow_graph\x18\x04 \x01(\x0b\x32\x30.spark.connect.PipelineCommand.DropDataflowGraphH\x00R\x11\x64ropDataflowGraph\x12\x46\n\tstart_run\x18\x05 \x01(\x0b\x32\'.spark.connect.PipelineCommand.StartRunH\x00R\x08startRun\x12r\n\x19\x64\x65\x66ine_sql_graph_elements\x18\x06 \x01(\x0b\x32\x35.spark.connect.PipelineCommand.DefineSqlGraphElementsH\x00R\x16\x64\x65\x66ineSqlGraphElements\x12\xa1\x01\n*get_query_function_execution_signal_stream\x18\x07 \x01(\x0b\x32\x44.spark.connect.PipelineCommand.GetQueryFunctionExecutionSignalStreamH\x00R%getQueryFunctionExecutionSignalStream\x12\x88\x01\n!define_flow_query_function_result\x18\x08 \x01(\x0b\x32<.spark.connect.PipelineCommand.DefineFlowQueryFunctionResultH\x00R\x1d\x64\x65\x66ineFlowQueryFunctionResult\x12\x65\n\x14\x65xecute_output_flows\x18\t \x01(\x0b\x32\x31.spark.connect.PipelineCommand.ExecuteOutputFlowsH\x00R\x12\x65xecuteOutputFlows\x12\x35\n\textension\x18\xe7\x07 \x01(\x0b\x32\x14.google.protobuf.AnyH\x00R\textension\x1a\xb4\x02\n\x13\x43reateDataflowGraph\x12,\n\x0f\x64\x65\x66\x61ult_catalog\x18\x01 \x01(\tH\x00R\x0e\x64\x65\x66\x61ultCatalog\x88\x01\x01\x12.\n\x10\x64\x65\x66\x61ult_database\x18\x02 \x01(\tH\x01R\x0f\x64\x65\x66\x61ultDatabase\x88\x01\x01\x12Z\n\x08sql_conf\x18\x05 \x03(\x0b\x32?.spark.connect.PipelineCommand.CreateDataflowGraph.SqlConfEntryR\x07sqlConf\x1a:\n\x0cSqlConfEntry\x12\x10\n\x03key\x18\x01 \x01(\tR\x03key\x12\x14\n\x05value\x18\x02 \x01(\tR\x05value:\x02\x38\x01\x42\x12\n\x10_default_catalogB\x13\n\x11_default_database\x1aZ\n\x11\x44ropDataflowGraph\x12/\n\x11\x64\x61taflow_graph_id\x18\x01 \x01(\tH\x00R\x0f\x64\x61taflowGraphId\x88\x01\x01\x42\x14\n\x12_dataflow_graph_id\x1a\x92\n\n\x0c\x44\x65\x66ineOutput\x12/\n\x11\x64\x61taflow_graph_id\x18\x01 \x01(\tH\x01R\x0f\x64\x61taflowGraphId\x88\x01\x01\x12$\n\x0boutput_name\x18\x02 \x01(\tH\x02R\noutputName\x88\x01\x01\x12?\n\x0boutput_type\x18\x03 \x01(\x0e\x32\x19.spark.connect.OutputTypeH\x03R\noutputType\x88\x01\x01\x12\x1d\n\x07\x63omment\x18\x04 \x01(\tH\x04R\x07\x63omment\x88\x01\x01\x12X\n\x14source_code_location\x18\x05 \x01(\x0b\x32!.spark.connect.SourceCodeLocationH\x05R\x12sourceCodeLocation\x88\x01\x01\x12_\n\rtable_details\x18\x06 \x01(\x0b\x32\x38.spark.connect.PipelineCommand.DefineOutput.TableDetailsH\x00R\x0ctableDetails\x12\\\n\x0csink_details\x18\x07 \x01(\x0b\x32\x37.spark.connect.PipelineCommand.DefineOutput.SinkDetailsH\x00R\x0bsinkDetails\x12\x35\n\textension\x18\xe7\x07 \x01(\x0b\x32\x14.google.protobuf.AnyH\x00R\textension\x1a\xc0\x03\n\x0cTableDetails\x12x\n\x10table_properties\x18\x01 \x03(\x0b\x32M.spark.connect.PipelineCommand.DefineOutput.TableDetails.TablePropertiesEntryR\x0ftableProperties\x12%\n\x0epartition_cols\x18\x02 \x03(\tR\rpartitionCols\x12\x1b\n\x06\x66ormat\x18\x03 \x01(\tH\x01R\x06\x66ormat\x88\x01\x01\x12\x43\n\x10schema_data_type\x18\x04 \x01(\x0b\x32\x17.spark.connect.DataTypeH\x00R\x0eschemaDataType\x12%\n\rschema_string\x18\x05 \x01(\tH\x00R\x0cschemaString\x12-\n\x12\x63lustering_columns\x18\x06 \x03(\tR\x11\x63lusteringColumns\x1a\x42\n\x14TablePropertiesEntry\x12\x10\n\x03key\x18\x01 \x01(\tR\x03key\x12\x14\n\x05value\x18\x02 \x01(\tR\x05value:\x02\x38\x01\x42\x08\n\x06schemaB\t\n\x07_format\x1a\xd1\x01\n\x0bSinkDetails\x12^\n\x07options\x18\x01 \x03(\x0b\x32\x44.spark.connect.PipelineCommand.DefineOutput.SinkDetails.OptionsEntryR\x07options\x12\x1b\n\x06\x66ormat\x18\x02 \x01(\tH\x00R\x06\x66ormat\x88\x01\x01\x1a:\n\x0cOptionsEntry\x12\x10\n\x03key\x18\x01 \x01(\tR\x03key\x12\x14\n\x05value\x18\x02 \x01(\tR\x05value:\x02\x38\x01\x42\t\n\x07_formatB\t\n\x07\x64\x65tailsB\x14\n\x12_dataflow_graph_idB\x0e\n\x0c_output_nameB\x0e\n\x0c_output_typeB\n\n\x08_commentB\x17\n\x15_source_code_location\x1a\x96\x12\n\nDefineFlow\x12/\n\x11\x64\x61taflow_graph_id\x18\x01 \x01(\tH\x01R\x0f\x64\x61taflowGraphId\x88\x01\x01\x12 \n\tflow_name\x18\x02 \x01(\tH\x02R\x08\x66lowName\x88\x01\x01\x12\x33\n\x13target_dataset_name\x18\x03 \x01(\tH\x03R\x11targetDatasetName\x88\x01\x01\x12Q\n\x08sql_conf\x18\x04 \x03(\x0b\x32\x36.spark.connect.PipelineCommand.DefineFlow.SqlConfEntryR\x07sqlConf\x12 \n\tclient_id\x18\x05 \x01(\tH\x04R\x08\x63lientId\x88\x01\x01\x12X\n\x14source_code_location\x18\x06 \x01(\x0b\x32!.spark.connect.SourceCodeLocationH\x05R\x12sourceCodeLocation\x88\x01\x01\x12x\n\x15relation_flow_details\x18\x07 \x01(\x0b\x32\x42.spark.connect.PipelineCommand.DefineFlow.WriteRelationFlowDetailsH\x00R\x13relationFlowDetails\x12q\n\x15\x61uto_cdc_flow_details\x18\n \x01(\x0b\x32<.spark.connect.PipelineCommand.DefineFlow.AutoCdcFlowDetailsH\x00R\x12\x61utoCdcFlowDetails\x12\x35\n\textension\x18\xe7\x07 \x01(\x0b\x32\x14.google.protobuf.AnyH\x00R\textension\x12\x17\n\x04once\x18\x08 \x01(\x08H\x06R\x04once\x88\x01\x01\x1a:\n\x0cSqlConfEntry\x12\x10\n\x03key\x18\x01 \x01(\tR\x03key\x12\x14\n\x05value\x18\x02 \x01(\tR\x05value:\x02\x38\x01\x1a\x61\n\x18WriteRelationFlowDetails\x12\x38\n\x08relation\x18\x01 \x01(\x0b\x32\x17.spark.connect.RelationH\x00R\x08relation\x88\x01\x01\x42\x0b\n\t_relation\x1a\xdc\t\n\x12\x41utoCdcFlowDetails\x12\x1b\n\x06source\x18\x01 \x01(\tH\x00R\x06source\x88\x01\x01\x12-\n\x04keys\x18\x02 \x03(\x0b\x32\x19.spark.connect.ExpressionR\x04keys\x12?\n\x0bsequence_by\x18\x03 \x01(\x0b\x32\x19.spark.connect.ExpressionH\x01R\nsequenceBy\x88\x01\x01\x12\x34\n\x05where\x18\x04 \x01(\x0b\x32\x19.spark.connect.ExpressionH\x02R\x05where\x88\x01\x01\x12\x33\n\x13ignore_null_updates\x18\x05 \x01(\x08H\x03R\x11ignoreNullUpdates\x88\x01\x01\x12H\n\x10\x61pply_as_deletes\x18\x06 \x01(\x0b\x32\x19.spark.connect.ExpressionH\x04R\x0e\x61pplyAsDeletes\x88\x01\x01\x12L\n\x12\x61pply_as_truncates\x18\x07 \x01(\x0b\x32\x19.spark.connect.ExpressionH\x05R\x10\x61pplyAsTruncates\x88\x01\x01\x12:\n\x0b\x63olumn_list\x18\x08 \x03(\x0b\x32\x19.spark.connect.ExpressionR\ncolumnList\x12G\n\x12\x65xcept_column_list\x18\t \x03(\x0b\x32\x19.spark.connect.ExpressionR\x10\x65xceptColumnList\x12^\n\x12stored_as_scd_type\x18\n \x01(\x0e\x32\x31.spark.connect.PipelineCommand.DefineFlow.SCDTypeR\x0fstoredAsScdType\x12T\n\x19track_history_column_list\x18\x0b \x03(\x0b\x32\x19.spark.connect.ExpressionR\x16trackHistoryColumnList\x12\x61\n track_history_except_column_list\x18\x0c \x03(\x0b\x32\x19.spark.connect.ExpressionR\x1ctrackHistoryExceptColumnList\x12_\n\x1fignore_null_updates_column_list\x18\x0e \x03(\x0b\x32\x19.spark.connect.ExpressionR\x1bignoreNullUpdatesColumnList\x12l\n&ignore_null_updates_except_column_list\x18\x0f \x03(\x0b\x32\x19.spark.connect.ExpressionR!ignoreNullUpdatesExceptColumnList\x12J\n\x11\x63olumns_to_update\x18\x10 \x01(\x0b\x32\x19.spark.connect.ExpressionH\x06R\x0f\x63olumnsToUpdate\x88\x01\x01\x42\t\n\x07_sourceB\x0e\n\x0c_sequence_byB\x08\n\x06_whereB\x16\n\x14_ignore_null_updatesB\x13\n\x11_apply_as_deletesB\x15\n\x13_apply_as_truncatesB\x14\n\x12_columns_to_update\x1a:\n\x08Response\x12 \n\tflow_name\x18\x01 \x01(\tH\x00R\x08\x66lowName\x88\x01\x01\x42\x0c\n\n_flow_name"C\n\x07SCDType\x12\x18\n\x14SCD_TYPE_UNSPECIFIED\x10\x00\x12\x0e\n\nSCD_TYPE_1\x10\x01\x12\x0e\n\nSCD_TYPE_2\x10\x02\x42\t\n\x07\x64\x65tailsB\x14\n\x12_dataflow_graph_idB\x0c\n\n_flow_nameB\x16\n\x14_target_dataset_nameB\x0c\n\n_client_idB\x17\n\x15_source_code_locationB\x07\n\x05_once\x1a\xe4\x02\n\x12\x45xecuteOutputFlows\x12U\n\rdefine_output\x18\x01 \x01(\x0b\x32+.spark.connect.PipelineCommand.DefineOutputH\x00R\x0c\x64\x65\x66ineOutput\x88\x01\x01\x12L\n\x0c\x64\x65\x66ine_flows\x18\x02 \x03(\x0b\x32).spark.connect.PipelineCommand.DefineFlowR\x0b\x64\x65\x66ineFlows\x12&\n\x0c\x66ull_refresh\x18\x03 \x01(\x08H\x01R\x0b\x66ullRefresh\x88\x01\x01\x12\x1d\n\x07storage\x18\x04 \x01(\tH\x02R\x07storage\x88\x01\x01\x12\x33\n\textension\x18\xe7\x07 \x03(\x0b\x32\x14.google.protobuf.AnyR\textensionB\x10\n\x0e_define_outputB\x0f\n\r_full_refreshB\n\n\x08_storage\x1a\xc2\x02\n\x08StartRun\x12/\n\x11\x64\x61taflow_graph_id\x18\x01 \x01(\tH\x00R\x0f\x64\x61taflowGraphId\x88\x01\x01\x12\x34\n\x16\x66ull_refresh_selection\x18\x02 \x03(\tR\x14\x66ullRefreshSelection\x12-\n\x10\x66ull_refresh_all\x18\x03 \x01(\x08H\x01R\x0e\x66ullRefreshAll\x88\x01\x01\x12+\n\x11refresh_selection\x18\x04 \x03(\tR\x10refreshSelection\x12\x15\n\x03\x64ry\x18\x05 \x01(\x08H\x02R\x03\x64ry\x88\x01\x01\x12\x1d\n\x07storage\x18\x06 \x01(\tH\x03R\x07storage\x88\x01\x01\x42\x14\n\x12_dataflow_graph_idB\x13\n\x11_full_refresh_allB\x06\n\x04_dryB\n\n\x08_storage\x1a\xc7\x01\n\x16\x44\x65\x66ineSqlGraphElements\x12/\n\x11\x64\x61taflow_graph_id\x18\x01 \x01(\tH\x00R\x0f\x64\x61taflowGraphId\x88\x01\x01\x12\'\n\rsql_file_path\x18\x02 \x01(\tH\x01R\x0bsqlFilePath\x88\x01\x01\x12\x1e\n\x08sql_text\x18\x03 \x01(\tH\x02R\x07sqlText\x88\x01\x01\x42\x14\n\x12_dataflow_graph_idB\x10\n\x0e_sql_file_pathB\x0b\n\t_sql_text\x1a\x9e\x01\n%GetQueryFunctionExecutionSignalStream\x12/\n\x11\x64\x61taflow_graph_id\x18\x01 \x01(\tH\x00R\x0f\x64\x61taflowGraphId\x88\x01\x01\x12 \n\tclient_id\x18\x02 \x01(\tH\x01R\x08\x63lientId\x88\x01\x01\x42\x14\n\x12_dataflow_graph_idB\x0c\n\n_client_id\x1a\xc6\x02\n\x1d\x44\x65\x66ineFlowQueryFunctionResult\x12$\n\tflow_name\x18\x01 \x01(\tB\x02\x18\x01H\x00R\x08\x66lowName\x88\x01\x01\x12O\n\x0f\x66low_identifier\x18\x04 \x01(\x0b\x32!.spark.connect.ResolvedIdentifierH\x01R\x0e\x66lowIdentifier\x88\x01\x01\x12/\n\x11\x64\x61taflow_graph_id\x18\x02 \x01(\tH\x02R\x0f\x64\x61taflowGraphId\x88\x01\x01\x12\x38\n\x08relation\x18\x03 \x01(\x0b\x32\x17.spark.connect.RelationH\x03R\x08relation\x88\x01\x01\x42\x0c\n\n_flow_nameB\x12\n\x10_flow_identifierB\x14\n\x12_dataflow_graph_idB\x0b\n\t_relationB\x0e\n\x0c\x63ommand_type"\xf0\x05\n\x15PipelineCommandResult\x12\x81\x01\n\x1c\x63reate_dataflow_graph_result\x18\x01 \x01(\x0b\x32>.spark.connect.PipelineCommandResult.CreateDataflowGraphResultH\x00R\x19\x63reateDataflowGraphResult\x12k\n\x14\x64\x65\x66ine_output_result\x18\x02 \x01(\x0b\x32\x37.spark.connect.PipelineCommandResult.DefineOutputResultH\x00R\x12\x64\x65\x66ineOutputResult\x12\x65\n\x12\x64\x65\x66ine_flow_result\x18\x03 \x01(\x0b\x32\x35.spark.connect.PipelineCommandResult.DefineFlowResultH\x00R\x10\x64\x65\x66ineFlowResult\x1a\x62\n\x19\x43reateDataflowGraphResult\x12/\n\x11\x64\x61taflow_graph_id\x18\x01 \x01(\tH\x00R\x0f\x64\x61taflowGraphId\x88\x01\x01\x42\x14\n\x12_dataflow_graph_id\x1a\x85\x01\n\x12\x44\x65\x66ineOutputResult\x12W\n\x13resolved_identifier\x18\x01 \x01(\x0b\x32!.spark.connect.ResolvedIdentifierH\x00R\x12resolvedIdentifier\x88\x01\x01\x42\x16\n\x14_resolved_identifier\x1a\x83\x01\n\x10\x44\x65\x66ineFlowResult\x12W\n\x13resolved_identifier\x18\x01 \x01(\x0b\x32!.spark.connect.ResolvedIdentifierH\x00R\x12resolvedIdentifier\x88\x01\x01\x42\x16\n\x14_resolved_identifierB\r\n\x0bresult_type"I\n\x13PipelineEventResult\x12\x32\n\x05\x65vent\x18\x01 \x01(\x0b\x32\x1c.spark.connect.PipelineEventR\x05\x65vent"t\n\rPipelineEvent\x12\x38\n\ttimestamp\x18\x01 \x01(\x0b\x32\x1a.google.protobuf.TimestampR\ttimestamp\x12\x1d\n\x07message\x18\x02 \x01(\tH\x00R\x07message\x88\x01\x01\x42\n\n\x08_message"\xf1\x01\n\x12SourceCodeLocation\x12 \n\tfile_name\x18\x01 \x01(\tH\x00R\x08\x66ileName\x88\x01\x01\x12$\n\x0bline_number\x18\x02 \x01(\x05H\x01R\nlineNumber\x88\x01\x01\x12,\n\x0f\x64\x65\x66inition_path\x18\x03 \x01(\tH\x02R\x0e\x64\x65\x66initionPath\x88\x01\x01\x12\x33\n\textension\x18\xe7\x07 \x03(\x0b\x32\x14.google.protobuf.AnyR\textensionB\x0c\n\n_file_nameB\x0e\n\x0c_line_numberB\x12\n\x10_definition_path"\x97\x01\n$PipelineQueryFunctionExecutionSignal\x12!\n\nflow_names\x18\x01 \x03(\tB\x02\x18\x01R\tflowNames\x12L\n\x10\x66low_identifiers\x18\x02 \x03(\x0b\x32!.spark.connect.ResolvedIdentifierR\x0f\x66lowIdentifiers"\xf0\x02\n\x17PipelineAnalysisContext\x12/\n\x11\x64\x61taflow_graph_id\x18\x01 \x01(\tH\x00R\x0f\x64\x61taflowGraphId\x88\x01\x01\x12,\n\x0f\x64\x65\x66inition_path\x18\x02 \x01(\tH\x01R\x0e\x64\x65\x66initionPath\x88\x01\x01\x12$\n\tflow_name\x18\x03 \x01(\tB\x02\x18\x01H\x02R\x08\x66lowName\x88\x01\x01\x12O\n\x0f\x66low_identifier\x18\x04 \x01(\x0b\x32!.spark.connect.ResolvedIdentifierH\x03R\x0e\x66lowIdentifier\x88\x01\x01\x12\x33\n\textension\x18\xe7\x07 \x03(\x0b\x32\x14.google.protobuf.AnyR\textensionB\x14\n\x12_dataflow_graph_idB\x12\n\x10_definition_pathB\x0c\n\n_flow_nameB\x12\n\x10_flow_identifier*i\n\nOutputType\x12\x1b\n\x17OUTPUT_TYPE_UNSPECIFIED\x10\x00\x12\x15\n\x11MATERIALIZED_VIEW\x10\x01\x12\t\n\x05TABLE\x10\x02\x12\x12\n\x0eTEMPORARY_VIEW\x10\x03\x12\x08\n\x04SINK\x10\x04\x42\x36\n\x1eorg.apache.spark.connect.protoP\x01Z\x12internal/generatedb\x06proto3'


Is there a particular python version you generated the stubs with? IIRC stub generation can be python verison dependent.

Just as a sanity check, if you were to regenerate these stubs using the same python environment on master, is it generating a no-diff stub?

I am not sure how I would generate these stubs on master because the protos won't be there. But I rebased the branch on latest master and regenerated the protos now.

I checked out master and ran the check:

[arca.ssh] ~/spark (master) $ python3 dev/check-protos.py ... Finish checking the generated codes in pyspark-streaming: SUCCESS

My Python version is 3.12, which should be correct.

anew · 2026-05-20T19:47:41Z

+              objectName = Option(flowIdentifier.unquotedString),
+              language = Some(Python()))))
+      case proto.PipelineCommand.DefineFlow.DetailsCase.AUTO_CDC_FLOW_DETAILS =>
+        throw new UnsupportedOperationException("AutoCdcFlowDetails is not yet implemented.")


this will be implemented as soon as the internal dataflow graph classes are merged.

AnishMahto · 2026-05-20T19:48:04Z

      optional spark.connect.Relation relation = 1;
    }

+    // Details for Apply Changes Into (ACI) flows.


nit on comment: AutoCDC flow, not Apply Changes Into flow

szehon-ho

are we ok with skipping the numbers for those fields we dropped ? if we are sure we will add those extra fields soon , it is ok, but otherwise it is a bit strange for future readers

szehon-ho · 2026-05-20T20:26:01Z

+    // Details for Apply Changes Into (ACI) flows.
+    message AutoCdcFlowDetails {
+      // The name of the CDC source to stream from.
+      optional string source = 1;


just to double check, we are ok with string here and not modeled Relation , right?

yes this is on purpose, the source should be a dataset name, which may be a view in order to represent any streaming query.

anew · 2026-05-20T21:44:39Z

are we ok with skipping the numbers for those fields we dropped ?
Yes, we are planning to add these fields in the future.

szehon-ho

thanks!

…2 from proto Remove track_history_column_list, track_history_except_column_list fields and the SCD_TYPE_2 enum value from AutoCdcFlowDetails in pipelines.proto. Co-authored-by: Isaac

Regenerated after removing track_history_column_list, track_history_except_column_list, and SCD_TYPE_2 from AutoCdcFlowDetails. Co-authored-by: Isaac

szehon-ho

chat offline with @anew , only change from the last is remove SCD-2 fields for now, as we target SCD-1 first

HyukjinKwon changed the title ~~[SPARK-56650][SDP|CONNECT] Add AutoCDC Spark Connect APIs~~ [SPARK-56650][SDP][CONNECT] Add AutoCDC Spark Connect APIs Apr 28, 2026

anew force-pushed the add-sc-apis branch 3 times, most recently from d35a707 to 1a71da4 Compare April 29, 2026 04:51

anew added 3 commits May 20, 2026 17:16

[SPARK-56650][SDP|CONNECT] Add AutoCDC Spark Connect APIs

0af0827

generate python protos

3599611

[spark tests]

bc48375

anew force-pushed the add-sc-apis branch from eb45993 to bc48375 Compare May 20, 2026 17:16

AnishMahto reviewed May 20, 2026

View reviewed changes

adress comments and regenerate protos

5cbfa07

AnishMahto approved these changes May 20, 2026

View reviewed changes

anew commented May 20, 2026

View reviewed changes

AnishMahto reviewed May 20, 2026

View reviewed changes

fix comment

a0d15f7

szehon-ho reviewed May 20, 2026

View reviewed changes

szehon-ho approved these changes May 20, 2026

View reviewed changes

gengliangwang approved these changes May 20, 2026

View reviewed changes

anew added 2 commits May 20, 2026 23:09

[SPARK-56650][SDP|CONNECT] Remove track history columns and SCD_TYPE_…

a8927f5

…2 from proto Remove track_history_column_list, track_history_except_column_list fields and the SCD_TYPE_2 enum value from AutoCdcFlowDetails in pipelines.proto. Co-authored-by: Isaac

[SPARK-56650][SDP|CONNECT] Regenerate Python proto stubs for pipelines

806bf52

Regenerated after removing track_history_column_list, track_history_except_column_list, and SCD_TYPE_2 from AutoCdcFlowDetails. Co-authored-by: Isaac

szehon-ho approved these changes May 20, 2026

View reviewed changes

		// Optional condition applied to source and target for optimizations like partition pruning.
		optional Expression where = 4;

		// Column indicating which user columns to update or ignore.
		optional Expression columns_to_update = 16;

		// Whether to ignore null values in source data updates.
		optional bool ignore_null_updates = 5;

Conversation

anew commented Apr 28, 2026

What changes were proposed in this pull request?

Why are the changes needed?

Does this PR introduce any user-facing change?

How was this patch tested?

Was this patch authored or co-authored using generative AI tooling?

Uh oh!

AnishMahto left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

szehon-ho left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

anew commented May 20, 2026

Uh oh!

szehon-ho left a comment

Choose a reason for hiding this comment

Uh oh!

szehon-ho left a comment

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants