You are viewing a plain text version of this content. The canonical link for it is here.
Posted to reviews@spark.apache.org by GitBox <gi...@apache.org> on 2022/07/13 12:43:58 UTC

[GitHub] [spark] cloud-fan opened a new pull request, #37176: [SPARK-39764][SQL] Make PhysicalOperation the same as ScanOperation

cloud-fan opened a new pull request, #37176:
URL: https://github.com/apache/spark/pull/37176

   <!--
   Thanks for sending a pull request!  Here are some tips for you:
     1. If this is your first time, please read our contributor guidelines: https://spark.apache.org/contributing.html
     2. Ensure you have added or run the appropriate tests for your PR: https://spark.apache.org/developer-tools.html
     3. If the PR is unfinished, add '[WIP]' in your PR title, e.g., '[WIP][SPARK-XXXX] Your PR title ...'.
     4. Be sure to keep the PR description updated to reflect all changes.
     5. Please write your PR title to summarize what this PR proposes.
     6. If possible, provide a concise example to reproduce the issue for a faster review.
     7. If you want to add a new configuration, please read the guideline first for naming configurations in
        'core/src/main/scala/org/apache/spark/internal/config/ConfigEntry.scala'.
     8. If you want to add or modify an error type or message, please read the guideline first in
        'core/src/main/resources/error/README.md'.
   -->
   
   ### What changes were proposed in this pull request?
   <!--
   Please clarify what changes you are proposing. The purpose of this section is to outline the changes and how this PR fixes the issue. 
   If possible, please consider writing useful notes for better and faster reviews in your PR. See the examples below.
     1. If you refactor some codes with changing classes, showing the class hierarchy will help reviewers.
     2. If you fix some SQL features, you can provide some references of other DBMSes.
     3. If there is design documentation, please add the link.
     4. If there is a discussion in the mailing list, please add the link.
   -->
   This PR updates `PhysicalOperation` to make it the same as `ScanOperation`, then remove `ScanOperation` and replace all its usages with `PhysicalOperation`. It also adds a new pattern match `NodeWithOnlyDeterministicProjectAndFilter` and uses it in places where we only need to extract a relation, but no need to collect projects and filters.
   
   ### Why are the changes needed?
   <!--
   Please clarify why the changes are needed. For instance,
     1. If you propose a new API, clarify the use case for a new API.
     2. If you fix a bug, you can clarify why it is a bug.
   -->
   `PhysicalOperation` has known issues: it aggressively collapses projects and filters, which may lead to bad query plans. To fix this issue, we introduced `ScanOperation` and use it in a few critical places like scan strategies. To be conservative, we didn't replace `PhysicalOperation` with `ScanOperation` everywhere. However, `PhysicalOperation` has performance issues in itself when collapsing projects and merging expressions, if the query plan is very complicated. We should always follow the rule `CollpaseProjects` and stop merging expressions when we should not.
   
   ### Does this PR introduce _any_ user-facing change?
   <!--
   Note that it means *any* user-facing change including all aspects such as the documentation fix.
   If yes, please clarify the previous behavior and the change this PR proposes - provide the console output, description and/or an example to show the behavior difference if possible.
   If possible, please also clarify if this is a user-facing change compared to the released Spark versions or within the unreleased branches such as master.
   If no, write 'No'.
   -->
   No
   
   ### How was this patch tested?
   <!--
   If tests were added, say they were added here. Please make sure to add some test cases that check the changes thoroughly including negative and positive cases if possible.
   If it was tested in a way different from regular unit tests, please clarify how you tested step by step, ideally copy and paste-able, so that other reviewers can test and check, and descendants can verify in the future.
   If tests were not added, please describe why they were not added and/or why it was difficult to add.
   If benchmark tests were added, please run the benchmarks in GitHub Actions for the consistent environment, and the instructions could accord to: https://spark.apache.org/developer-tools.html#github-workflow-benchmarks.
   -->
   existing tests


-- 
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.

To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org


[GitHub] [spark] cloud-fan commented on pull request #37176: [SPARK-39764][SQL] Make PhysicalOperation the same as ScanOperation

Posted by GitBox <gi...@apache.org>.
cloud-fan commented on PR #37176:
URL: https://github.com/apache/spark/pull/37176#issuecomment-1207697625

   thanks for review, merging to master!


-- 
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.

To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org


[GitHub] [spark] viirya commented on a diff in pull request #37176: [SPARK-39764][SQL] Make PhysicalOperation the same as ScanOperation

Posted by GitBox <gi...@apache.org>.
viirya commented on code in PR #37176:
URL: https://github.com/apache/spark/pull/37176#discussion_r920465103


##########
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/planning/patterns.scala:
##########
@@ -113,23 +106,15 @@ trait OperationHelper extends AliasHelper with PredicateHelper {
 }
 
 /**
- * A pattern that matches any number of project or filter operations on top of another relational
- * operator.  All filter operators are collected and their conditions are broken up and returned
- * together with the top project operator.
- * [[org.apache.spark.sql.catalyst.expressions.Alias Aliases]] are in-lined/substituted if
- * necessary.
+ * A pattern that matches a plan node with only deterministic Project and/or Filter above it. It
+ * returns the input plan itself if there is no Project or Filter.
  */
-object PhysicalOperation extends OperationHelper with PredicateHelper {
-  override protected def legacyMode: Boolean = true
-}
-
-/**
- * A variant of [[PhysicalOperation]]. It matches any number of project or filter
- * operations even if they are non-deterministic, as long as they satisfy the
- * requirement of CollapseProject and CombineFilters.
- */
-object ScanOperation extends OperationHelper with PredicateHelper {
-  override protected def legacyMode: Boolean = false
+object NodeWithOnlyDeterministicProjectAndFilter {
+  def unapply(plan: LogicalPlan): Option[LogicalPlan] = plan match {
+    case Project(projectList, child) if projectList.forall(_.deterministic) => unapply(child)
+    case Filter(cond, child) if cond.deterministic => unapply(child)
+    case _ => Some(plan)

Review Comment:
   Hmm, if it returns `Some` for all other cases, doesn't it mean it matches all input plans?



-- 
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.

To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org


[GitHub] [spark] cloud-fan closed pull request #37176: [SPARK-39764][SQL] Make PhysicalOperation the same as ScanOperation

Posted by GitBox <gi...@apache.org>.
cloud-fan closed pull request #37176: [SPARK-39764][SQL] Make PhysicalOperation the same as ScanOperation
URL: https://github.com/apache/spark/pull/37176


-- 
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.

To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org


[GitHub] [spark] ulysses-you commented on pull request #37176: [SPARK-39764][SQL] Make PhysicalOperation the same as ScanOperation

Posted by GitBox <gi...@apache.org>.
ulysses-you commented on PR #37176:
URL: https://github.com/apache/spark/pull/37176#issuecomment-1183851038

   Follow the previous `PhysicalOperation` semantics, shall we also update this to `NodeWithOnlyDeterministicProjectAndFilter` ? we may miss inject runtime filter if the project of filter side holds some complex expressions.
   
   https://github.com/apache/spark/blob/899f6c90eb2de5b46a36710a131d7417010ce4b3/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/optimizer/InjectRuntimeFilter.scala#L119-L126
   
   otherwise looks good to me.


-- 
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.

To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org


[GitHub] [spark] cloud-fan commented on pull request #37176: [SPARK-39764][SQL] Make PhysicalOperation the same as ScanOperation

Posted by GitBox <gi...@apache.org>.
cloud-fan commented on PR #37176:
URL: https://github.com/apache/spark/pull/37176#issuecomment-1183178710

   cc @viirya @wangyum @ulysses-you 


-- 
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.

To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org


[GitHub] [spark] cloud-fan commented on a diff in pull request #37176: [SPARK-39764][SQL] Make PhysicalOperation the same as ScanOperation

Posted by GitBox <gi...@apache.org>.
cloud-fan commented on code in PR #37176:
URL: https://github.com/apache/spark/pull/37176#discussion_r920746311


##########
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/optimizer/InjectRuntimeFilter.scala:
##########
@@ -117,13 +117,24 @@ object InjectRuntimeFilter extends Rule[LogicalPlan] with PredicateHelper with J
    * do not add a subquery that might have an expensive computation
    */
   private def isSelectiveFilterOverScan(plan: LogicalPlan): Boolean = {
-    val ret = plan match {
-      case PhysicalOperation(_, filters, child) if child.isInstanceOf[LeafNode] =>

Review Comment:
   The old behavior was, collecting all the filters above a leaf node and checking if all filters are simple expressions and at least one of them is selective.
   
   I've rewritten the code to keep the old behavior but in a more efficient way: simply traverse the plan tree to check filter predicates, instead of merging expressions and collecting filters that may duplicate complicated expressions.



-- 
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.

To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org


[GitHub] [spark] cloud-fan commented on a diff in pull request #37176: [SPARK-39764][SQL] Make PhysicalOperation the same as ScanOperation

Posted by GitBox <gi...@apache.org>.
cloud-fan commented on code in PR #37176:
URL: https://github.com/apache/spark/pull/37176#discussion_r920622276


##########
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/planning/patterns.scala:
##########
@@ -113,23 +106,15 @@ trait OperationHelper extends AliasHelper with PredicateHelper {
 }
 
 /**
- * A pattern that matches any number of project or filter operations on top of another relational
- * operator.  All filter operators are collected and their conditions are broken up and returned
- * together with the top project operator.
- * [[org.apache.spark.sql.catalyst.expressions.Alias Aliases]] are in-lined/substituted if
- * necessary.
+ * A pattern that matches a plan node with only deterministic Project and/or Filter above it. It
+ * returns the input plan itself if there is no Project or Filter.
  */
-object PhysicalOperation extends OperationHelper with PredicateHelper {
-  override protected def legacyMode: Boolean = true
-}
-
-/**
- * A variant of [[PhysicalOperation]]. It matches any number of project or filter
- * operations even if they are non-deterministic, as long as they satisfy the
- * requirement of CollapseProject and CombineFilters.
- */
-object ScanOperation extends OperationHelper with PredicateHelper {
-  override protected def legacyMode: Boolean = false
+object NodeWithOnlyDeterministicProjectAndFilter {
+  def unapply(plan: LogicalPlan): Option[LogicalPlan] = plan match {
+    case Project(projectList, child) if projectList.forall(_.deterministic) => unapply(child)
+    case Filter(cond, child) if cond.deterministic => unapply(child)
+    case _ => Some(plan)

Review Comment:
   Yes, and I explicitly call it out in the classdoc. This is the same in `PhysicalOperation`, which always returns `Some`, and the caller side will specifiy the expected type.



##########
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/planning/patterns.scala:
##########
@@ -113,23 +106,15 @@ trait OperationHelper extends AliasHelper with PredicateHelper {
 }
 
 /**
- * A pattern that matches any number of project or filter operations on top of another relational
- * operator.  All filter operators are collected and their conditions are broken up and returned
- * together with the top project operator.
- * [[org.apache.spark.sql.catalyst.expressions.Alias Aliases]] are in-lined/substituted if
- * necessary.
+ * A pattern that matches a plan node with only deterministic Project and/or Filter above it. It
+ * returns the input plan itself if there is no Project or Filter.
  */
-object PhysicalOperation extends OperationHelper with PredicateHelper {
-  override protected def legacyMode: Boolean = true
-}
-
-/**
- * A variant of [[PhysicalOperation]]. It matches any number of project or filter
- * operations even if they are non-deterministic, as long as they satisfy the
- * requirement of CollapseProject and CombineFilters.
- */
-object ScanOperation extends OperationHelper with PredicateHelper {
-  override protected def legacyMode: Boolean = false
+object NodeWithOnlyDeterministicProjectAndFilter {
+  def unapply(plan: LogicalPlan): Option[LogicalPlan] = plan match {
+    case Project(projectList, child) if projectList.forall(_.deterministic) => unapply(child)
+    case Filter(cond, child) if cond.deterministic => unapply(child)
+    case _ => Some(plan)

Review Comment:
   Yes, and I explicitly call it out in the classdoc. This is the same in `PhysicalOperation`, which always returns `Some`, and the caller side will specify the expected type.



-- 
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.

To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org