## What changes were proposed in this pull request?
Binary operator requires its inputs to be of same type, but it should not consider nullability, e.g. `EqualTo` should be able to compare an element-nullable array and an element-non-nullable array.
## How was this patch tested?
a regression test in `DataFrameSuite`
Author: Wenchen Fan <wenchen@databricks.com>
Closes#15606 from cloud-fan/type-bug.
## What changes were proposed in this pull request?
Currently we always lowercase the partition columns of partition spec in parser, with the assumption that table partition columns are always lowercased.
However, this is not true for data source tables, which are case preserving. It's safe for now because data source tables don't store partition spec in metastore and don't support `ADD PARTITION`, `DROP PARTITION`, `RENAME PARTITION`, but we should make our code future-proof.
This PR makes partition spec case preserving at parser, and improve the `PreprocessTableInsertion` analyzer rule to normalize the partition columns in partition spec, w.r.t. the table partition columns.
## How was this patch tested?
existing tests.
Author: Wenchen Fan <wenchen@databricks.com>
Closes#15566 from cloud-fan/partition-spec.
## What changes were proposed in this pull request?
Simplify/cleanup TableFileCatalog:
1. pass a `CatalogTable` instead of `databaseName` and `tableName` into `TableFileCatalog`, so that we don't need to fetch table metadata from metastore again
2. In `TableFileCatalog.filterPartitions0`, DO NOT set `PartitioningAwareFileCatalog.BASE_PATH_PARAM`. According to the [classdoc](https://github.com/apache/spark/blob/master/sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/PartitioningAwareFileCatalog.scala#L189-L209), the default value of `basePath` already satisfies our need. What's more, if we set this parameter, we may break the case 2 which is metioned in the classdoc.
3. add `equals` and `hashCode` to `TableFileCatalog`
4. add `SessionCatalog.listPartitionsByFilter` which handles case sensitivity.
## How was this patch tested?
existing tests.
Author: Wenchen Fan <wenchen@databricks.com>
Closes#15568 from cloud-fan/table-file-catalog.
## What changes were proposed in this pull request?
The PR tries to fix [SPARK-18058](https://issues.apache.org/jira/browse/SPARK-18058) which refers to a bug that the column types are compared with the extra care about Nullability in Union and SetOperation.
This PR converts the columns types by setting all fields as nullable before comparison
## How was this patch tested?
regular unit test cases
Author: CodingCat <zhunansjtu@gmail.com>
Closes#15595 from CodingCat/SPARK-18058.
## What changes were proposed in this pull request?
Jira: https://issues.apache.org/jira/browse/SPARK-18035
In HiveInspectors, I saw that converting Java map to Spark's `ArrayBasedMapData` spent quite sometime in buffer copying : https://github.com/apache/spark/blob/master/sql/hive/src/main/scala/org/apache/spark/sql/hive/HiveInspectors.scala#L658
The reason being `map.toSeq` allocates a new buffer and copies the map entries to it: https://github.com/scala/scala/blob/2.11.x/src/library/scala/collection/MapLike.scala#L323
This copy is not needed as we get rid of it once we extract the key and value arrays.
Here is the call trace:
```
org.apache.spark.sql.hive.HiveInspectors$$anonfun$unwrapperFor$41.apply(HiveInspectors.scala:664)
scala.collection.AbstractMap.toSeq(Map.scala:59)
scala.collection.MapLike$class.toSeq(MapLike.scala:323)
scala.collection.AbstractMap.toBuffer(Map.scala:59)
scala.collection.MapLike$class.toBuffer(MapLike.scala:326)
scala.collection.AbstractTraversable.copyToBuffer(Traversable.scala:104)
scala.collection.TraversableOnce$class.copyToBuffer(TraversableOnce.scala:275)
scala.collection.mutable.ArrayBuffer.$plus$plus$eq(ArrayBuffer.scala:48)
scala.collection.mutable.ArrayBuffer.$plus$plus$eq(ArrayBuffer.scala:104)
scala.collection.generic.Growable$class.$plus$plus$eq(Growable.scala:59)
scala.collection.AbstractIterable.foreach(Iterable.scala:54)
scala.collection.IterableLike$class.foreach(IterableLike.scala:72)
scala.collection.AbstractIterator.foreach(Iterator.scala:1336)
scala.collection.Iterator$class.foreach(Iterator.scala:893)
scala.collection.generic.Growable$$anonfun$$plus$plus$eq$1.apply(Growable.scala:59)
scala.collection.generic.Growable$$anonfun$$plus$plus$eq$1.apply(Growable.scala:59)
```
Also, earlier code was populating keys and values arrays separately by iterating twice. The PR avoids double iteration of the map and does it in one iteration.
EDIT: During code review, there were several more places in the code which were found to do similar thing. The PR dedupes those instances and introduces convenient APIs which are performant and memory efficient
## Performance gains
The number is subjective and depends on how many map columns are accessed in the query and average entries per map. For one the queries that I tried out, I saw 3% CPU savings (end-to-end) for the query.
## How was this patch tested?
This does not change the end result produced so relying on existing tests.
Author: Tejas Patil <tejasp@fb.com>
Closes#15573 from tejasapatil/SPARK-18035_avoid_toSeq.
## What changes were proposed in this pull request?
`Array[T]()` -> `Array.empty[T]` to avoid allocating 0-length arrays.
Use regex `find . -name '*.scala' | xargs -i bash -c 'egrep "Array\[[A-Za-z]+\]\(\)" -n {} && echo {}'` to find modification candidates.
cc srowen
## How was this patch tested?
existing tests
Author: Zheng RuiFeng <ruifengz@foxmail.com>
Closes#15564 from zhengruifeng/avoid_0_length_array.
## What changes were proposed in this pull request?
In `PruneFileSourcePartitions`, we will replace the `LogicalRelation` with a pruned one. However, this replacement may change the output of the `LogicalRelation` if it doesn't have `expectedOutputAttributes`. This PR fixes it.
## How was this patch tested?
the new `PruneFileSourcePartitionsSuite`
Author: Wenchen Fan <wenchen@databricks.com>
Closes#15569 from cloud-fan/partition-bug.
## What changes were proposed in this pull request?
Add mapValues to KeyValueGroupedDataset
## How was this patch tested?
New test in DatasetSuite for groupBy function, mapValues, flatMap
Author: Koert Kuipers <koert@tresata.com>
Closes#13526 from koertkuipers/feat-keyvaluegroupeddataset-mapvalues.
## What changes were proposed in this pull request?
Jira : https://issues.apache.org/jira/browse/SPARK-17698
`ExtractEquiJoinKeys` is incorrectly using filter predicates as the join condition for joins. `canEvaluate` [0] tries to see if the an `Expression` can be evaluated using output of a given `Plan`. In case of filter predicates (eg. `a.id='1'`), the `Expression` passed for the right hand side (ie. '1' ) is a `Literal` which does not have any attribute references. Thus `expr.references` is an empty set which theoretically is a subset of any set. This leads to `canEvaluate` returning `true` and `a.id='1'` is treated as a join predicate. While this does not lead to incorrect results but in case of bucketed + sorted tables, we might miss out on avoiding un-necessary shuffle + sort. See example below:
[0] : https://github.com/apache/spark/blob/master/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/predicates.scala#L91
eg.
```
val df = (1 until 10).toDF("id").coalesce(1)
hc.sql("DROP TABLE IF EXISTS table1").collect
df.write.bucketBy(8, "id").sortBy("id").saveAsTable("table1")
hc.sql("DROP TABLE IF EXISTS table2").collect
df.write.bucketBy(8, "id").sortBy("id").saveAsTable("table2")
sqlContext.sql("""
SELECT a.id, b.id
FROM table1 a
FULL OUTER JOIN table2 b
ON a.id = b.id AND a.id='1' AND b.id='1'
""").explain(true)
```
BEFORE: This is doing shuffle + sort over table scan outputs which is not needed as both tables are bucketed and sorted on the same columns and have same number of buckets. This should be a single stage job.
```
SortMergeJoin [id#38, cast(id#38 as double), 1.0], [id#39, 1.0, cast(id#39 as double)], FullOuter
:- *Sort [id#38 ASC NULLS FIRST, cast(id#38 as double) ASC NULLS FIRST, 1.0 ASC NULLS FIRST], false, 0
: +- Exchange hashpartitioning(id#38, cast(id#38 as double), 1.0, 200)
: +- *FileScan parquet default.table1[id#38] Batched: true, Format: ParquetFormat, InputPaths: file:spark-warehouse/table1, PartitionFilters: [], PushedFilters: [], ReadSchema: struct<id:int>
+- *Sort [id#39 ASC NULLS FIRST, 1.0 ASC NULLS FIRST, cast(id#39 as double) ASC NULLS FIRST], false, 0
+- Exchange hashpartitioning(id#39, 1.0, cast(id#39 as double), 200)
+- *FileScan parquet default.table2[id#39] Batched: true, Format: ParquetFormat, InputPaths: file:spark-warehouse/table2, PartitionFilters: [], PushedFilters: [], ReadSchema: struct<id:int>
```
AFTER :
```
SortMergeJoin [id#32], [id#33], FullOuter, ((cast(id#32 as double) = 1.0) && (cast(id#33 as double) = 1.0))
:- *FileScan parquet default.table1[id#32] Batched: true, Format: ParquetFormat, InputPaths: file:spark-warehouse/table1, PartitionFilters: [], PushedFilters: [], ReadSchema: struct<id:int>
+- *FileScan parquet default.table2[id#33] Batched: true, Format: ParquetFormat, InputPaths: file:spark-warehouse/table2, PartitionFilters: [], PushedFilters: [], ReadSchema: struct<id:int>
```
## How was this patch tested?
- Added a new test case for this scenario : `SPARK-17698 Join predicates should not contain filter clauses`
- Ran all the tests in `BucketedReadSuite`
Author: Tejas Patil <tejasp@fb.com>
Closes#15272 from tejasapatil/SPARK-17698_join_predicate_filter_clause.
## What changes were proposed in this pull request?
This PR proposes to check the second argument, `ascendingOrder` rather than throwing `ClassCastException` exception message.
```sql
select sort_array(array('b', 'd'), '1');
```
**Before**
```
16/10/19 13:16:08 ERROR SparkSQLDriver: Failed in [select sort_array(array('b', 'd'), '1')]
java.lang.ClassCastException: org.apache.spark.unsafe.types.UTF8String cannot be cast to java.lang.Boolean
at scala.runtime.BoxesRunTime.unboxToBoolean(BoxesRunTime.java:85)
at org.apache.spark.sql.catalyst.expressions.SortArray.nullSafeEval(collectionOperations.scala:185)
at org.apache.spark.sql.catalyst.expressions.BinaryExpression.eval(Expression.scala:416)
at org.apache.spark.sql.catalyst.optimizer.ConstantFolding$$anonfun$apply$1$$anonfun$applyOrElse$1.applyOrElse(expressions.scala:50)
at org.apache.spark.sql.catalyst.optimizer.ConstantFolding$$anonfun$apply$1$$anonfun$applyOrElse$1.applyOrElse(expressions.scala:43)
at org.apache.spark.sql.catalyst.trees.TreeNode$$anonfun$3.apply(TreeNode.scala:292)
at org.apache.spark.sql.catalyst.trees.TreeNode$$anonfun$3.apply(TreeNode.scala:292)
at org.apache.spark.sql.catalyst.trees.CurrentOrigin$.withOrigin(TreeNode.scala:74)
at org.apache.spark.sql.catalyst.trees.TreeNode.transformDown(TreeNode.scala:291)
at org.apache.spark.sql.catalyst.trees.TreeNode$$anonfun$transformDown$1.apply(TreeNode.scala:297)
```
**After**
```
Error in query: cannot resolve 'sort_array(array('b', 'd'), '1')' due to data type mismatch: Sort order in second argument requires a boolean literal.; line 1 pos 7;
```
## How was this patch tested?
Unit test in `DataFrameFunctionsSuite`.
Author: hyukjinkwon <gurwls223@gmail.com>
Closes#15532 from HyukjinKwon/SPARK-17989.
## What changes were proposed in this pull request?
Unlike Hive, in Spark SQL, ALTER TABLE RENAME TO cannot move a table from one database to another(e.g. `ALTER TABLE db1.tbl RENAME TO db2.tbl2`), and will report error if the database in source table and destination table is different. So in #14955 , we forbid users to specify database of destination table in ALTER TABLE RENAME TO, to be consistent with other database systems and also make it easier to rename tables in non-current database, e.g. users can write `ALTER TABLE db1.tbl RENAME TO tbl2`, instead of `ALTER TABLE db1.tbl RENAME TO db1.tbl2`.
However, this is a breaking change. Users may already have queries that specify database of destination table in ALTER TABLE RENAME TO.
This PR reverts most of #14955 , and simplify the usage of ALTER TABLE RENAME TO by making database of source table the default database of destination table, instead of current database, so that users can still write `ALTER TABLE db1.tbl RENAME TO tbl2`, which is consistent with other databases like MySQL, Postgres, etc.
## How was this patch tested?
The added back tests and some new tests.
Author: Wenchen Fan <wenchen@databricks.com>
Closes#15434 from cloud-fan/revert.
### What changes were proposed in this pull request?
Dataset always does eager analysis now. Thus, `spark.sql.eagerAnalysis` is not used any more. Thus, we need to remove it.
This PR also outputs the plan. Without the fix, the analysis error is like
```
cannot resolve '`k1`' given input columns: [k, v]; line 1 pos 12
```
After the fix, the analysis error becomes:
```
org.apache.spark.sql.AnalysisException: cannot resolve '`k1`' given input columns: [k, v]; line 1 pos 12;
'Project [unresolvedalias(CASE WHEN ('k1 = 2) THEN 22 WHEN ('k1 = 4) THEN 44 ELSE 0 END, None), v#6]
+- SubqueryAlias t
+- Project [_1#2 AS k#5, _2#3 AS v#6]
+- LocalRelation [_1#2, _2#3]
```
### How was this patch tested?
N/A
Author: gatorsmile <gatorsmile@gmail.com>
Closes#15316 from gatorsmile/eagerAnalysis.
## What changes were proposed in this pull request?
Added a `prettyname` for current_database function.
## How was this patch tested?
Manually.
Before:
```
scala> sql("select current_database()").show
+-----------------+
|currentdatabase()|
+-----------------+
| default|
+-----------------+
```
After:
```
scala> sql("select current_database()").show
+------------------+
|current_database()|
+------------------+
| default|
+------------------+
```
Author: Weiqing Yang <yangweiqing001@gmail.com>
Closes#15506 from weiqingy/prettyName.
(This PR addresses https://issues.apache.org/jira/browse/SPARK-16980.)
## What changes were proposed in this pull request?
In a new Spark session, when a partitioned Hive table is converted to use Spark's `HadoopFsRelation` in `HiveMetastoreCatalog`, metadata for every partition of that table are retrieved from the metastore and loaded into driver memory. In addition, every partition's metadata files are read from the filesystem to perform schema inference.
If a user queries such a table with predicates which prune that table's partitions, we would like to be able to answer that query without consulting partition metadata which are not involved in the query. When querying a table with a large number of partitions for some data from a small number of partitions (maybe even a single partition), the current conversion strategy is highly inefficient. I suspect this scenario is not uncommon in the wild.
In addition to being inefficient in running time, the current strategy is inefficient in its use of driver memory. When the sum of the number of partitions of all tables loaded in a driver reaches a certain level (somewhere in the tens of thousands), their cached data exhaust all driver heap memory in the default configuration. I suspect this scenario is less common (in that not too many deployments work with tables with tens of thousands of partitions), however this does illustrate how large the memory footprint of this metadata can be. With tables with hundreds or thousands of partitions, I would expect the `HiveMetastoreCatalog` table cache to represent a significant portion of the driver's heap space.
This PR proposes an alternative approach. Basically, it makes four changes:
1. It adds a new method, `listPartitionsByFilter` to the Catalyst `ExternalCatalog` trait which returns the partition metadata for a given sequence of partition pruning predicates.
1. It refactors the `FileCatalog` type hierarchy to include a new `TableFileCatalog` to efficiently return files only for partitions matching a sequence of partition pruning predicates.
1. It removes partition loading and caching from `HiveMetastoreCatalog`.
1. It adds a new Catalyst optimizer rule, `PruneFileSourcePartitions`, which applies a plan's partition-pruning predicates to prune out unnecessary partition files from a `HadoopFsRelation`'s underlying file catalog.
The net effect is that when a query over a partitioned Hive table is planned, the analyzer retrieves the table metadata from `HiveMetastoreCatalog`. As part of this operation, the `HiveMetastoreCatalog` builds a `HadoopFsRelation` with a `TableFileCatalog`. It does not load any partition metadata or scan any files. The optimizer prunes-away unnecessary table partitions by sending the partition-pruning predicates to the relation's `TableFileCatalog `. The `TableFileCatalog` in turn calls the `listPartitionsByFilter` method on its external catalog. This queries the Hive metastore, passing along those filters.
As a bonus, performing partition pruning during optimization leads to a more accurate relation size estimate. This, along with c481bdf, can lead to automatic, safe application of the broadcast optimization in a join where it might previously have been omitted.
## Open Issues
1. This PR omits partition metadata caching. I can add this once the overall strategy for the cold path is established, perhaps in a future PR.
1. This PR removes and omits partitioned Hive table schema reconciliation. As a result, it fails to find Parquet schema columns with upper case letters because of the Hive metastore's case-insensitivity. This issue may be fixed by #14750, but that PR appears to have stalled. ericl has contributed to this PR a workaround for Parquet wherein schema reconciliation occurs at query execution time instead of planning. Whether ORC requires a similar patch is an open issue.
1. This PR omits an implementation of `listPartitionsByFilter` for the `InMemoryCatalog`.
1. This PR breaks parquet log output redirection during query execution. I can work around this by running `Class.forName("org.apache.spark.sql.execution.datasources.parquet.ParquetFileFormat$")` first thing in a Spark shell session, but I haven't figured out how to fix this properly.
## How was this patch tested?
The current Spark unit tests were run, and some ad-hoc tests were performed to validate that only the necessary partition metadata is loaded.
Author: Michael Allman <michael@videoamp.com>
Author: Eric Liang <ekl@databricks.com>
Author: Eric Liang <ekhliang@gmail.com>
Closes#14690 from mallman/spark-16980-lazy_partition_fetching.
Currently pyspark can only call the builtin java UDF, but can not call custom java UDF. It would be better to allow that. 2 benefits:
* Leverage the power of rich third party java library
* Improve the performance. Because if we use python UDF, python daemons will be started on worker which will affect the performance.
Author: Jeff Zhang <zjffdu@apache.org>
Closes#9766 from zjffdu/SPARK-11775.
## What changes were proposed in this pull request?
We are trying to resolve the attribute in sort by pulling up some column for grandchild into child, but that's wrong when the child is Distinct, because the added column will change the behavior of Distinct, we should not do that.
## How was this patch tested?
Added regression test.
Author: Davies Liu <davies@databricks.com>
Closes#15489 from davies/order_distinct.
## What changes were proposed in this pull request?
`HiveExternalCatalog` should be the only interface to talk to the hive metastore. In `MetastoreRelation` we can just use `ExternalCatalog` instead of `HiveClient` to interact with hive metastore, and add missing API in `ExternalCatalog`.
## How was this patch tested?
existing tests.
Author: Wenchen Fan <wenchen@databricks.com>
Closes#15460 from cloud-fan/relation.
## What changes were proposed in this pull request?
Value classes were unsupported because catalyst data types were
obtained through reflection on erased types, which would resolve to a
value class' wrapped type and hence lead to unavailable methods during
code generation.
E.g. the following class
```scala
case class Foo(x: Int) extends AnyVal
```
would be seen as an `int` in catalyst and will cause instance cast failures when generated java code tries to treat it as a `Foo`.
This patch simply removes the erasure step when getting data types for
catalyst.
## How was this patch tested?
Additional tests in `ExpressionEncoderSuite`.
Author: Jakob Odersky <jakob@odersky.com>
Closes#15284 from jodersky/value-classes.
## What changes were proposed in this pull request?
Metrics are needed for monitoring structured streaming apps. Here is the design doc for implementing the necessary metrics.
https://docs.google.com/document/d/1NIdcGuR1B3WIe8t7VxLrt58TJB4DtipWEbj5I_mzJys/edit?usp=sharing
Specifically, this PR adds the following public APIs changes.
### New APIs
- `StreamingQuery.status` returns a `StreamingQueryStatus` object (renamed from `StreamingQueryInfo`, see later)
- `StreamingQueryStatus` has the following important fields
- inputRate - Current rate (rows/sec) at which data is being generated by all the sources
- processingRate - Current rate (rows/sec) at which the query is processing data from
all the sources
- ~~outputRate~~ - *Does not work with wholestage codegen*
- latency - Current average latency between the data being available in source and the sink writing the corresponding output
- sourceStatuses: Array[SourceStatus] - Current statuses of the sources
- sinkStatus: SinkStatus - Current status of the sink
- triggerStatus - Low-level detailed status of the last completed/currently active trigger
- latencies - getOffset, getBatch, full trigger, wal writes
- timestamps - trigger start, finish, after getOffset, after getBatch
- numRows - input, output, state total/updated rows for aggregations
- `SourceStatus` has the following important fields
- inputRate - Current rate (rows/sec) at which data is being generated by the source
- processingRate - Current rate (rows/sec) at which the query is processing data from the source
- triggerStatus - Low-level detailed status of the last completed/currently active trigger
- Python API for `StreamingQuery.status()`
### Breaking changes to existing APIs
**Existing direct public facing APIs**
- Deprecated direct public-facing APIs `StreamingQuery.sourceStatuses` and `StreamingQuery.sinkStatus` in favour of `StreamingQuery.status.sourceStatuses/sinkStatus`.
- Branch 2.0 should have it deprecated, master should have it removed.
**Existing advanced listener APIs**
- `StreamingQueryInfo` renamed to `StreamingQueryStatus` for consistency with `SourceStatus`, `SinkStatus`
- Earlier StreamingQueryInfo was used only in the advanced listener API, but now it is used in direct public-facing API (StreamingQuery.status)
- Field `queryInfo` in listener events `QueryStarted`, `QueryProgress`, `QueryTerminated` changed have name `queryStatus` and return type `StreamingQueryStatus`.
- Field `offsetDesc` in `SourceStatus` was Option[String], converted it to `String`.
- For `SourceStatus` and `SinkStatus` made constructor private instead of private[sql] to make them more java-safe. Instead added `private[sql] object SourceStatus/SinkStatus.apply()` which are harder to accidentally use in Java.
## How was this patch tested?
Old and new unit tests.
- Rate calculation and other internal logic of StreamMetrics tested by StreamMetricsSuite.
- New info in statuses returned through StreamingQueryListener is tested in StreamingQueryListenerSuite.
- New and old info returned through StreamingQuery.status is tested in StreamingQuerySuite.
- Source-specific tests for making sure input rows are counted are is source-specific test suites.
- Additional tests to test minor additions in LocalTableScanExec, StateStore, etc.
Metrics also manually tested using Ganglia sink
Author: Tathagata Das <tathagata.das1565@gmail.com>
Closes#15307 from tdas/SPARK-17731.
## What changes were proposed in this pull request?
correct the expected type from Length function to be Int
## How was this patch tested?
Test runs on little endian and big endian platforms
Author: Pete Robbins <robbinspg@gmail.com>
Closes#15464 from robbinspg/SPARK-17827.
## What changes were proposed in this pull request?
minor doc fix for "getAnyValAs" in class Row
## How was this patch tested?
None.
(If this patch involves UI changes, please attach a screenshot; otherwise, remove this)
Author: buzhihuojie <ren.weiluo@gmail.com>
Closes#15452 from david-weiluo-ren/minorDocFixForRow.
## What changes were proposed in this pull request?
This change adds a check in castToInterval method of Cast expression , such that if converted value is null , then isNull variable should be set to true.
Earlier, the expression Cast(Literal(), CalendarIntervalType) was throwing NullPointerException because of the above mentioned reason.
## How was this patch tested?
Added test case in CastSuite.scala
jira entry for detail: https://issues.apache.org/jira/browse/SPARK-17884
Author: prigarg <prigarg@adobe.com>
Closes#15449 from priyankagargnitk/SPARK-17884.
## What changes were proposed in this pull request?
SQLConf is session-scoped and mutable. However, we do have the requirement for a static SQL conf, which is global and immutable, e.g. the `schemaStringThreshold` in `HiveExternalCatalog`, the flag to enable/disable hive support, the global temp view database in https://github.com/apache/spark/pull/14897.
Actually we've already implemented static SQL conf implicitly via `SparkConf`, this PR just make it explicit and expose it to users, so that they can see the config value via SQL command or `SparkSession.conf`, and forbid users to set/unset static SQL conf.
## How was this patch tested?
new tests in SQLConfSuite
Author: Wenchen Fan <wenchen@databricks.com>
Closes#15295 from cloud-fan/global-conf.
## What changes were proposed in this pull request?
Currently `Canonicalize` object doesn't support `And` and `Or`. So we can compare canonicalized form of predicates consistently. We should add the support.
## How was this patch tested?
Jenkins tests.
Author: Liang-Chi Hsieh <viirya@gmail.com>
Closes#15388 from viirya/canonicalize-and-or.
## What changes were proposed in this pull request?
The data type API has not been changed since Spark 1.3.0, and is ready for graduation. This patch marks them as stable APIs using the new InterfaceStability annotation.
This patch also looks at the various files in the catalyst module (not the "package") and marks the remaining few classes appropriately as well.
## How was this patch tested?
This is an annotation change. No functional changes.
Author: Reynold Xin <rxin@databricks.com>
Closes#15426 from rxin/SPARK-17864.
## What changes were proposed in this pull request?
address post hoc review comments for https://github.com/apache/spark/pull/14897
## How was this patch tested?
N/A
Author: Wenchen Fan <wenchen@databricks.com>
Closes#15424 from cloud-fan/global-temp-view.
## What changes were proposed in this pull request?
Global temporary view is a cross-session temporary view, which means it's shared among all sessions. Its lifetime is the lifetime of the Spark application, i.e. it will be automatically dropped when the application terminates. It's tied to a system preserved database `global_temp`(configurable via SparkConf), and we must use the qualified name to refer a global temp view, e.g. SELECT * FROM global_temp.view1.
changes for `SessionCatalog`:
1. add a new field `gloabalTempViews: GlobalTempViewManager`, to access the shared global temp views, and the global temp db name.
2. `createDatabase` will fail if users wanna create `global_temp`, which is system preserved.
3. `setCurrentDatabase` will fail if users wanna set `global_temp`, which is system preserved.
4. add `createGlobalTempView`, which is used in `CreateViewCommand` to create global temp views.
5. add `dropGlobalTempView`, which is used in `CatalogImpl` to drop global temp view.
6. add `alterTempViewDefinition`, which is used in `AlterViewAsCommand` to update the view definition for local/global temp views.
7. `renameTable`/`dropTable`/`isTemporaryTable`/`lookupRelation`/`getTempViewOrPermanentTableMetadata`/`refreshTable` will handle global temp views.
changes for SQL commands:
1. `CreateViewCommand`/`AlterViewAsCommand` is updated to support global temp views
2. `ShowTablesCommand` outputs a new column `database`, which is used to distinguish global and local temp views.
3. other commands can also handle global temp views if they call `SessionCatalog` APIs which accepts global temp views, e.g. `DropTableCommand`, `AlterTableRenameCommand`, `ShowColumnsCommand`, etc.
changes for other public API
1. add a new method `dropGlobalTempView` in `Catalog`
2. `Catalog.findTable` can find global temp view
3. add a new method `createGlobalTempView` in `Dataset`
## How was this patch tested?
new tests in `SQLViewSuite`
Author: Wenchen Fan <wenchen@databricks.com>
Closes#14897 from cloud-fan/global-temp-view.
## What changes were proposed in this pull request?
Currently we use the same rule to parse top level and nested data fields. For example:
```
create table tbl_x(
id bigint,
nested struct<col1:string,col2:string>
)
```
Shows both syntaxes. In this PR we split this rule in a top-level and nested rule.
Before this PR,
```
sql("CREATE TABLE my_tab(column1: INT)")
```
works fine.
After this PR, it will throw a `ParseException`:
```
scala> sql("CREATE TABLE my_tab(column1: INT)")
org.apache.spark.sql.catalyst.parser.ParseException:
no viable alternative at input 'CREATE TABLE my_tab(column1:'(line 1, pos 27)
```
## How was this patch tested?
Add new testcases in `SparkSqlParserSuite`.
Author: jiangxingbo <jiangxb1987@gmail.com>
Closes#15346 from jiangxb1987/cdt.
## What changes were proposed in this pull request?
The `quotedString` method in `TableIdentifier` and `FunctionIdentifier` produce an illegal (un-parseable) name when the name contains a backtick. For example:
```
import org.apache.spark.sql.catalyst.parser.CatalystSqlParser._
import org.apache.spark.sql.catalyst.TableIdentifier
import org.apache.spark.sql.catalyst.analysis.UnresolvedAttribute
val complexName = TableIdentifier("`weird`table`name", Some("`d`b`1"))
parseTableIdentifier(complexName.unquotedString) // Does not work
parseTableIdentifier(complexName.quotedString) // Does not work
parseExpression(complexName.unquotedString) // Does not work
parseExpression(complexName.quotedString) // Does not work
```
We should handle the backtick properly to make `quotedString` parseable.
## How was this patch tested?
Add new testcases in `TableIdentifierParserSuite` and `ExpressionParserSuite`.
Author: jiangxingbo <jiangxb1987@gmail.com>
Closes#15403 from jiangxb1987/backtick.
## What changes were proposed in this pull request?
In practice we cannot guarantee that an `InternalRow` is immutable. This makes the `MutableRow` almost redundant. This PR folds `MutableRow` into `InternalRow`.
The code below illustrates the immutability issue with InternalRow:
```scala
import org.apache.spark.sql.catalyst.InternalRow
import org.apache.spark.sql.catalyst.expressions.GenericMutableRow
val struct = new GenericMutableRow(1)
val row = InternalRow(struct, 1)
println(row)
scala> [[null], 1]
struct.setInt(0, 42)
println(row)
scala> [[42], 1]
```
This might be somewhat controversial, so feedback is appreciated.
## How was this patch tested?
Existing tests.
Author: Herman van Hovell <hvanhovell@databricks.com>
Closes#15333 from hvanhovell/SPARK-17761.
## What changes were proposed in this pull request?
Currently, Spark raises `RuntimeException` when creating a view with timestamp with INTERVAL arithmetic like the following. The root cause is the arithmetic expression, `TimeAdd`, was transformed into `timeadd` function as a VIEW definition. This PR fixes the SQL definition of `TimeAdd` and `TimeSub` expressions.
```scala
scala> sql("CREATE TABLE dates (ts TIMESTAMP)")
scala> sql("CREATE VIEW view1 AS SELECT ts + INTERVAL 1 DAY FROM dates")
java.lang.RuntimeException: Failed to analyze the canonicalized SQL: ...
```
## How was this patch tested?
Pass Jenkins with a new testcase.
Author: Dongjoon Hyun <dongjoon@apache.org>
Closes#15318 from dongjoon-hyun/SPARK-17750.
## What changes were proposed in this pull request?
The result of the `Last` function can be wrong when the last partition processed is empty. It can return `null` instead of the expected value. For example, this can happen when we process partitions in the following order:
```
- Partition 1 [Row1, Row2]
- Partition 2 [Row3]
- Partition 3 []
```
In this case the `Last` function will currently return a null, instead of the value of `Row3`.
This PR fixes this by adding a `valueSet` flag to the `Last` function.
## How was this patch tested?
We only used end to end tests for `DeclarativeAggregateFunction`s. I have added an evaluator for these functions so we can tests them in catalyst. I have added a `LastTestSuite` to test the `Last` aggregate function.
Author: Herman van Hovell <hvanhovell@databricks.com>
Closes#15348 from hvanhovell/SPARK-17758.
## What changes were proposed in this pull request?
This PR fixes the following NPE scenario in two ways.
**Reported Error Scenario**
```scala
scala> sql("EXPLAIN DESCRIBE TABLE x").show(truncate = false)
INFO SparkSqlParser: Parsing command: EXPLAIN DESCRIBE TABLE x
java.lang.NullPointerException
```
- **DESCRIBE**: Extend `DESCRIBE` syntax to accept `TABLE`.
- **EXPLAIN**: Prevent NPE in case of the parsing failure of target statement, e.g., `EXPLAIN DESCRIBE TABLES x`.
## How was this patch tested?
Pass the Jenkins test with a new test case.
Author: Dongjoon Hyun <dongjoon@apache.org>
Closes#15357 from dongjoon-hyun/SPARK-17328.
## What changes were proposed in this pull request?
Currently Spark SQL parses regular decimal literals (e.g. `10.00`) as decimals and scientific decimal literals (e.g. `10.0e10`) as doubles. The difference between the two confuses most users. This PR unifies the parsing behavior and also parses scientific decimal literals as decimals.
This implications in tests are limited to a single Hive compatibility test.
## How was this patch tested?
Updated tests in `ExpressionParserSuite` and `SQLQueryTestSuite`.
Author: Herman van Hovell <hvanhovell@databricks.com>
Closes#14828 from hvanhovell/SPARK-17258.
## What changes were proposed in this pull request?
Code generation including too many mutable states exceeds JVM size limit to extract values from `references` into fields in the constructor.
We should split the generated extractions in the constructor into smaller functions.
## How was this patch tested?
I added some tests to check if the generated codes for the expressions exceed or not.
Author: Takuya UESHIN <ueshin@happy-camper.st>
Closes#15275 from ueshin/issues/SPARK-17702.
## What changes were proposed in this pull request?
We currently only allow relatively simple expressions as the input for a value based case statement. Expressions like `case (a > 1) or (b = 2) when true then 1 when false then 0 end` currently fail. This PR adds support for such expressions.
## How was this patch tested?
Added a test to the ExpressionParserSuite.
Author: Herman van Hovell <hvanhovell@databricks.com>
Closes#15322 from hvanhovell/SPARK-17753.
## What changes were proposed in this pull request?
Generate basic column statistics for all the atomic types:
- numeric types: max, min, num of nulls, ndv (number of distinct values)
- date/timestamp types: they are also represented as numbers internally, so they have the same stats as above.
- string: avg length, max length, num of nulls, ndv
- binary: avg length, max length, num of nulls
- boolean: num of nulls, num of trues, num of falsies
Also support storing and loading these statistics.
One thing to notice:
We support analyzing columns independently, e.g.:
sql1: `ANALYZE TABLE src COMPUTE STATISTICS FOR COLUMNS key;`
sql2: `ANALYZE TABLE src COMPUTE STATISTICS FOR COLUMNS value;`
when running sql2 to collect column stats for `value`, we don’t remove stats of columns `key` which are analyzed in sql1 and not in sql2. As a result, **users need to guarantee consistency** between sql1 and sql2. If the table has been changed before sql2, users should re-analyze column `key` when they want to analyze column `value`:
`ANALYZE TABLE src COMPUTE STATISTICS FOR COLUMNS key, value;`
## How was this patch tested?
add unit tests
Author: Zhenhua Wang <wzh_zju@163.com>
Closes#15090 from wzhfy/colStats.
## What changes were proposed in this pull request?
Currently, Spark does not collapse adjacent windows with the same partitioning and sorting. This PR implements `CollapseWindow` optimizer to do the followings.
1. If the partition specs and order specs are the same, collapse into the parent.
2. If the partition specs are the same and one order spec is a prefix of the other, collapse to the more specific one.
For example:
```scala
val df = spark.range(1000).select($"id" % 100 as "grp", $"id", rand() as "col1", rand() as "col2")
// Add summary statistics for all columns
import org.apache.spark.sql.expressions.Window
val cols = Seq("id", "col1", "col2")
val window = Window.partitionBy($"grp").orderBy($"id")
val result = cols.foldLeft(df) { (base, name) =>
base.withColumn(s"${name}_avg", avg(col(name)).over(window))
.withColumn(s"${name}_stddev", stddev(col(name)).over(window))
.withColumn(s"${name}_min", min(col(name)).over(window))
.withColumn(s"${name}_max", max(col(name)).over(window))
}
```
**Before**
```scala
scala> result.explain
== Physical Plan ==
Window [max(col2#19) windowspecdefinition(grp#17L, id#14L ASC NULLS FIRST, RANGE BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) AS col2_max#234], [grp#17L], [id#14L ASC NULLS FIRST]
+- Window [min(col2#19) windowspecdefinition(grp#17L, id#14L ASC NULLS FIRST, RANGE BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) AS col2_min#216], [grp#17L], [id#14L ASC NULLS FIRST]
+- Window [stddev_samp(col2#19) windowspecdefinition(grp#17L, id#14L ASC NULLS FIRST, RANGE BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) AS col2_stddev#191], [grp#17L], [id#14L ASC NULLS FIRST]
+- Window [avg(col2#19) windowspecdefinition(grp#17L, id#14L ASC NULLS FIRST, RANGE BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) AS col2_avg#167], [grp#17L], [id#14L ASC NULLS FIRST]
+- Window [max(col1#18) windowspecdefinition(grp#17L, id#14L ASC NULLS FIRST, RANGE BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) AS col1_max#152], [grp#17L], [id#14L ASC NULLS FIRST]
+- Window [min(col1#18) windowspecdefinition(grp#17L, id#14L ASC NULLS FIRST, RANGE BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) AS col1_min#138], [grp#17L], [id#14L ASC NULLS FIRST]
+- Window [stddev_samp(col1#18) windowspecdefinition(grp#17L, id#14L ASC NULLS FIRST, RANGE BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) AS col1_stddev#117], [grp#17L], [id#14L ASC NULLS FIRST]
+- Window [avg(col1#18) windowspecdefinition(grp#17L, id#14L ASC NULLS FIRST, RANGE BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) AS col1_avg#97], [grp#17L], [id#14L ASC NULLS FIRST]
+- Window [max(id#14L) windowspecdefinition(grp#17L, id#14L ASC NULLS FIRST, RANGE BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) AS id_max#86L], [grp#17L], [id#14L ASC NULLS FIRST]
+- Window [min(id#14L) windowspecdefinition(grp#17L, id#14L ASC NULLS FIRST, RANGE BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) AS id_min#76L], [grp#17L], [id#14L ASC NULLS FIRST]
+- *Project [grp#17L, id#14L, col1#18, col2#19, id_avg#26, id_stddev#42]
+- Window [stddev_samp(_w0#59) windowspecdefinition(grp#17L, id#14L ASC NULLS FIRST, RANGE BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) AS id_stddev#42], [grp#17L], [id#14L ASC NULLS FIRST]
+- *Project [grp#17L, id#14L, col1#18, col2#19, id_avg#26, cast(id#14L as double) AS _w0#59]
+- Window [avg(id#14L) windowspecdefinition(grp#17L, id#14L ASC NULLS FIRST, RANGE BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) AS id_avg#26], [grp#17L], [id#14L ASC NULLS FIRST]
+- *Sort [grp#17L ASC NULLS FIRST, id#14L ASC NULLS FIRST], false, 0
+- Exchange hashpartitioning(grp#17L, 200)
+- *Project [(id#14L % 100) AS grp#17L, id#14L, rand(-6329949029880411066) AS col1#18, rand(-7251358484380073081) AS col2#19]
+- *Range (0, 1000, step=1, splits=Some(8))
```
**After**
```scala
scala> result.explain
== Physical Plan ==
Window [max(col2#5) windowspecdefinition(grp#3L, id#0L ASC NULLS FIRST, RANGE BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) AS col2_max#220, min(col2#5) windowspecdefinition(grp#3L, id#0L ASC NULLS FIRST, RANGE BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) AS col2_min#202, stddev_samp(col2#5) windowspecdefinition(grp#3L, id#0L ASC NULLS FIRST, RANGE BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) AS col2_stddev#177, avg(col2#5) windowspecdefinition(grp#3L, id#0L ASC NULLS FIRST, RANGE BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) AS col2_avg#153, max(col1#4) windowspecdefinition(grp#3L, id#0L ASC NULLS FIRST, RANGE BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) AS col1_max#138, min(col1#4) windowspecdefinition(grp#3L, id#0L ASC NULLS FIRST, RANGE BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) AS col1_min#124, stddev_samp(col1#4) windowspecdefinition(grp#3L, id#0L ASC NULLS FIRST, RANGE BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) AS col1_stddev#103, avg(col1#4) windowspecdefinition(grp#3L, id#0L ASC NULLS FIRST, RANGE BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) AS col1_avg#83, max(id#0L) windowspecdefinition(grp#3L, id#0L ASC NULLS FIRST, RANGE BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) AS id_max#72L, min(id#0L) windowspecdefinition(grp#3L, id#0L ASC NULLS FIRST, RANGE BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) AS id_min#62L], [grp#3L], [id#0L ASC NULLS FIRST]
+- *Project [grp#3L, id#0L, col1#4, col2#5, id_avg#12, id_stddev#28]
+- Window [stddev_samp(_w0#45) windowspecdefinition(grp#3L, id#0L ASC NULLS FIRST, RANGE BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) AS id_stddev#28], [grp#3L], [id#0L ASC NULLS FIRST]
+- *Project [grp#3L, id#0L, col1#4, col2#5, id_avg#12, cast(id#0L as double) AS _w0#45]
+- Window [avg(id#0L) windowspecdefinition(grp#3L, id#0L ASC NULLS FIRST, RANGE BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) AS id_avg#12], [grp#3L], [id#0L ASC NULLS FIRST]
+- *Sort [grp#3L ASC NULLS FIRST, id#0L ASC NULLS FIRST], false, 0
+- Exchange hashpartitioning(grp#3L, 200)
+- *Project [(id#0L % 100) AS grp#3L, id#0L, rand(6537478539664068821) AS col1#4, rand(-8961093871295252795) AS col2#5]
+- *Range (0, 1000, step=1, splits=Some(8))
```
## How was this patch tested?
Pass the Jenkins tests with a newly added testsuite.
Author: Dongjoon Hyun <dongjoon@apache.org>
Closes#15317 from dongjoon-hyun/SPARK-17739.
## What changes were proposed in this pull request?
There are many minor objects in references, which are extracted to the generated class field, e.g. `errMsg` in `GetExternalRowField` or `ValidateExternalType`, but number of fields in class is limited so we should reduce the number.
This pr adds unnamed version of `addReferenceObj` for these minor objects not to store the object into field but refer it from the `references` field at the time of use.
## How was this patch tested?
Existing tests.
Author: Takuya UESHIN <ueshin@happy-camper.st>
Closes#15276 from ueshin/issues/SPARK-17703.
## What changes were proposed in this pull request?
Currently for `Union [Distinct]`, a `Distinct` operator is necessary to be on the top of `Union`. Once there are adjacent `Union [Distinct]`, there will be multiple `Distinct` in the query plan.
E.g.,
For a query like: select 1 a union select 2 b union select 3 c
Before this patch, its physical plan looks like:
*HashAggregate(keys=[a#13], functions=[])
+- Exchange hashpartitioning(a#13, 200)
+- *HashAggregate(keys=[a#13], functions=[])
+- Union
:- *HashAggregate(keys=[a#13], functions=[])
: +- Exchange hashpartitioning(a#13, 200)
: +- *HashAggregate(keys=[a#13], functions=[])
: +- Union
: :- *Project [1 AS a#13]
: : +- Scan OneRowRelation[]
: +- *Project [2 AS b#14]
: +- Scan OneRowRelation[]
+- *Project [3 AS c#15]
+- Scan OneRowRelation[]
Only the top distinct should be necessary.
After this patch, the physical plan looks like:
*HashAggregate(keys=[a#221], functions=[], output=[a#221])
+- Exchange hashpartitioning(a#221, 5)
+- *HashAggregate(keys=[a#221], functions=[], output=[a#221])
+- Union
:- *Project [1 AS a#221]
: +- Scan OneRowRelation[]
:- *Project [2 AS b#222]
: +- Scan OneRowRelation[]
+- *Project [3 AS c#223]
+- Scan OneRowRelation[]
## How was this patch tested?
Jenkins tests.
Author: Liang-Chi Hsieh <viirya@gmail.com>
Closes#15238 from viirya/remove-extra-distinct-union.
Spark SQL has great support for reading text files that contain JSON data. However, in many cases the JSON data is just one column amongst others. This is particularly true when reading from sources such as Kafka. This PR adds a new functions `from_json` that converts a string column into a nested `StructType` with a user specified schema.
Example usage:
```scala
val df = Seq("""{"a": 1}""").toDS()
val schema = new StructType().add("a", IntegerType)
df.select(from_json($"value", schema) as 'json) // => [json: <a: int>]
```
This PR adds support for java, scala and python. I leveraged our existing JSON parsing support by moving it into catalyst (so that we could define expressions using it). I left SQL out for now, because I'm not sure how users would specify a schema.
Author: Michael Armbrust <michael@databricks.com>
Closes#15274 from marmbrus/jsonParser.
## What changes were proposed in this pull request?
This patch fixes a minor correctness issue impacting the pushdown of filters beneath aggregates. Specifically, if a filter condition references no grouping or aggregate columns (e.g. `WHERE false`) then it would be incorrectly pushed beneath an aggregate.
Intuitively, the only case where you can push a filter beneath an aggregate is when that filter is deterministic and is defined over the grouping columns / expressions, since in that case the filter is acting to exclude entire groups from the query (like a `HAVING` clause). The existing code would only push deterministic filters beneath aggregates when all of the filter's references were grouping columns, but this logic missed the case where a filter has no references. For example, `WHERE false` is deterministic but is independent of the actual data.
This patch fixes this minor bug by adding a new check to ensure that we don't push filters beneath aggregates when those filters don't reference any columns.
## How was this patch tested?
New regression test in FilterPushdownSuite.
Author: Josh Rosen <joshrosen@databricks.com>
Closes#15289 from JoshRosen/SPARK-17712.
## What changes were proposed in this pull request?
We added native versions of `collect_set` and `collect_list` in Spark 2.0. These currently also (try to) collect null values, this is different from the original Hive implementation. This PR fixes this by adding a null check to the `Collect.update` method.
## How was this patch tested?
Added a regression test to `DataFrameAggregateSuite`.
Author: Herman van Hovell <hvanhovell@databricks.com>
Closes#15208 from hvanhovell/SPARK-17641.
This patch ports changes from #15185 to Spark 2.x. In that patch, a correctness bug in Spark 1.6.x which was caused by an invalid `equals()` comparison between an `UnsafeRow` and another row of a different format. Spark 2.x is not affected by that specific correctness bug but it can still reap the error-prevention benefits of that patch's changes, which modify ``UnsafeRow.equals()` to throw an IllegalArgumentException if it is called with an object that is not an `UnsafeRow`.
Author: Josh Rosen <joshrosen@databricks.com>
Closes#15265 from JoshRosen/SPARK-17618-master.
## What changes were proposed in this pull request?
This PR introduces more compact representation for ```UnsafeArrayData```.
```UnsafeArrayData``` needs to accept ```null``` value in each entry of an array. In the current version, it has three parts
```
[numElements] [offsets] [values]
```
`Offsets` has the number of `numElements`, and represents `null` if its value is negative. It may increase memory footprint, and introduces an indirection for accessing each of `values`.
This PR uses bitvectors to represent nullability for each element like `UnsafeRow`, and eliminates an indirection for accessing each element. The new ```UnsafeArrayData``` has four parts.
```
[numElements][null bits][values or offset&length][variable length portion]
```
In the `null bits` region, we store 1 bit per element, represents whether an element is null. Its total size is ceil(numElements / 8) bytes, and it is aligned to 8-byte boundaries.
In the `values or offset&length` region, we store the content of elements. For fields that hold fixed-length primitive types, such as long, double, or int, we store the value directly in the field. For fields with non-primitive or variable-length values, we store a relative offset (w.r.t. the base address of the array) that points to the beginning of the variable-length field and length (they are combined into a long). Each is word-aligned. For `variable length portion`, each is aligned to 8-byte boundaries.
The new format can reduce memory footprint and improve performance of accessing each element. An example of memory foot comparison:
1024x1024 elements integer array
Size of ```baseObject``` for ```UnsafeArrayData```: 8 + 1024x1024 + 1024x1024 = 2M bytes
Size of ```baseObject``` for ```UnsafeArrayData```: 8 + 1024x1024/8 + 1024x1024 = 1.25M bytes
In summary, we got 1.0-2.6x performance improvements over the code before applying this PR.
Here are performance results of [benchmark programs](04d2e4b6db/sql/core/src/test/scala/org/apache/spark/sql/execution/benchmark/UnsafeArrayDataBenchmark.scala):
**Read UnsafeArrayData**: 1.7x and 1.6x performance improvements over the code before applying this PR
````
OpenJDK 64-Bit Server VM 1.8.0_91-b14 on Linux 4.4.11-200.fc22.x86_64
Intel Xeon E3-12xx v2 (Ivy Bridge)
Without SPARK-15962
Read UnsafeArrayData: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------
Int 430 / 436 390.0 2.6 1.0X
Double 456 / 485 367.8 2.7 0.9X
With SPARK-15962
Read UnsafeArrayData: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------
Int 252 / 260 666.1 1.5 1.0X
Double 281 / 292 597.7 1.7 0.9X
````
**Write UnsafeArrayData**: 1.0x and 1.1x performance improvements over the code before applying this PR
````
OpenJDK 64-Bit Server VM 1.8.0_91-b14 on Linux 4.0.4-301.fc22.x86_64
Intel Xeon E3-12xx v2 (Ivy Bridge)
Without SPARK-15962
Write UnsafeArrayData: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------
Int 203 / 273 103.4 9.7 1.0X
Double 239 / 356 87.9 11.4 0.8X
With SPARK-15962
Write UnsafeArrayData: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------
Int 196 / 249 107.0 9.3 1.0X
Double 227 / 367 92.3 10.8 0.9X
````
**Get primitive array from UnsafeArrayData**: 2.6x and 1.6x performance improvements over the code before applying this PR
````
OpenJDK 64-Bit Server VM 1.8.0_91-b14 on Linux 4.0.4-301.fc22.x86_64
Intel Xeon E3-12xx v2 (Ivy Bridge)
Without SPARK-15962
Get primitive array from UnsafeArrayData: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------
Int 207 / 217 304.2 3.3 1.0X
Double 257 / 363 245.2 4.1 0.8X
With SPARK-15962
Get primitive array from UnsafeArrayData: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------
Int 151 / 198 415.8 2.4 1.0X
Double 214 / 394 293.6 3.4 0.7X
````
**Create UnsafeArrayData from primitive array**: 1.7x and 2.1x performance improvements over the code before applying this PR
````
OpenJDK 64-Bit Server VM 1.8.0_91-b14 on Linux 4.0.4-301.fc22.x86_64
Intel Xeon E3-12xx v2 (Ivy Bridge)
Without SPARK-15962
Create UnsafeArrayData from primitive array: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------
Int 340 / 385 185.1 5.4 1.0X
Double 479 / 705 131.3 7.6 0.7X
With SPARK-15962
Create UnsafeArrayData from primitive array: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------
Int 206 / 211 306.0 3.3 1.0X
Double 232 / 406 271.6 3.7 0.9X
````
1.7x and 1.4x performance improvements in [```UDTSerializationBenchmark```](https://github.com/apache/spark/blob/master/mllib/src/test/scala/org/apache/spark/mllib/linalg/UDTSerializationBenchmark.scala) over the code before applying this PR
````
OpenJDK 64-Bit Server VM 1.8.0_91-b14 on Linux 4.4.11-200.fc22.x86_64
Intel Xeon E3-12xx v2 (Ivy Bridge)
Without SPARK-15962
VectorUDT de/serialization: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------
serialize 442 / 533 0.0 441927.1 1.0X
deserialize 217 / 274 0.0 217087.6 2.0X
With SPARK-15962
VectorUDT de/serialization: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------
serialize 265 / 318 0.0 265138.5 1.0X
deserialize 155 / 197 0.0 154611.4 1.7X
````
## How was this patch tested?
Added unit tests into ```UnsafeArraySuite```
Author: Kazuaki Ishizaki <ishizaki@jp.ibm.com>
Closes#13680 from kiszk/SPARK-15962.
## What changes were proposed in this pull request?
This pull request adds Scala/Java DataFrame API for null ordering (NULLS FIRST | LAST).
Also did some minor clean up for related code (e.g. incorrect indentation), and renamed "orderby-nulls-ordering.sql" to be consistent with existing test files.
## How was this patch tested?
Added a new test case in DataFrameSuite.
Author: petermaxlee <petermaxlee@gmail.com>
Author: Xin Wu <xinwu@us.ibm.com>
Closes#15123 from petermaxlee/SPARK-17551.
## What changes were proposed in this pull request?
We currently cannot execute an aggregate that contains a single distinct aggregate function and an one or more non-partially plannable aggregate functions, for example:
```sql
select grp,
collect_list(col1),
count(distinct col2)
from tbl_a
group by 1
```
This is a regression from Spark 1.6. This is caused by the fact that the single distinct aggregation code path assumes that all aggregates can be planned in two phases (is partially aggregatable). This PR works around this issue by triggering the `RewriteDistinctAggregates` in such cases (this is similar to the approach taken in 1.6).
## How was this patch tested?
Created `RewriteDistinctAggregatesSuite` which checks if the aggregates with distinct aggregate functions get rewritten into two `Aggregates` and an `Expand`. Added a regression test to `DataFrameAggregateSuite`.
Author: Herman van Hovell <hvanhovell@databricks.com>
Closes#15187 from hvanhovell/SPARK-17616.
