rdblue commented on a change in pull request #24129: [SPARK-27190][SQL] add table capability for streaming URL: https://github.com/apache/spark/pull/24129#discussion_r276873657
########## File path: sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/v2/V2StreamingScanSupportCheck.scala ########## @@ -0,0 +1,51 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one or more + * contributor license agreements. See the NOTICE file distributed with + * this work for additional information regarding copyright ownership. + * The ASF licenses this file to You under the Apache License, Version 2.0 + * (the "License"); you may not use this file except in compliance with + * the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +package org.apache.spark.sql.execution.datasources.v2 + +import org.apache.spark.sql.AnalysisException +import org.apache.spark.sql.catalyst.plans.logical.LogicalPlan +import org.apache.spark.sql.execution.streaming.{StreamingRelation, StreamingRelationV2} +import org.apache.spark.sql.sources.v2.TableCapability.{CONTINUOUS_READ, MICRO_BATCH_READ} + +/** + * This rules adds some basic table capability check for streaming scan, without knowing the actual + * streaming execution mode. + */ +object V2StreamingScanSupportCheck extends (LogicalPlan => Unit) { + import DataSourceV2Implicits._ + + override def apply(plan: LogicalPlan): Unit = { Review comment: @cloud-fan, let's get the validation in now. I don't think that the fallback rule should be implemented as you describe. I think it should be done in 2 parts. The rule to fallback and update the plan, and a validation that all sources support streaming. Spark should not combine transform rules and validations. There are a couple of reasons for this principle: 1. Validations are used to ensure that the query is valid *and* to ensure that rules are run correctly. If the transform rule is added to the analyzer in a single-run batch, we want validation to catch that. These checks catch errors in Spark, too. 2. Rules should be as small as possible and focused on a single task. The fallback rule should not fail analysis if it doesn't know what to do because some other rule may be added later that does. For example, what if we build an adapter from continuous execution to micro-batch execution for a source? So we will need a validation rule either way. When the fallback rule runs and can't fix the problem, this check should be what fails the plan. ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
