### What changes were proposed in this pull request?
We only need to do aggregate evaluation once per group in `UnboundedWindowFunctionFrame`
### Why are the changes needed?
Currently, in `UnboundedWindowFunctionFrame.write`，it re-evaluate the processor for each row in a group, which is not necessary in fact which I'll address later. It hurts performance when the evaluation is time-consuming (for example, Percentile's eval need to sort its buffer and do some calculation). In our production, there is a percentile with window operation sql, it costs more than 10 hours in SparkSQL while 10min in Hive.
In fact, `UnboundedWindowFunctionFrame` can be treated as `SlidingWindowFunctionFrame` with `lbound = UnboundedPreceding` and `ubound = UnboundedFollowing`, just as its comments. In that case, `SlidingWindowFunctionFrame` also only do evaluation once for each group.
The performance issue can be reproduced by running the follow scripts in local spark-shell
```
spark.range(100*100).map(i => (i, "India")).toDF("uv", "country").createOrReplaceTempView("test")
sql("select uv, country, percentile(uv, 0.95) over (partition by country) as ptc95 from test").collect.foreach(println)
```
Before this patch, the sql costs **128048 ms**.
With this patch, the sql costs **3485 ms**.
If we increase the data size to 1000*1000 for example, then spark cannot even produce result without this patch(I'v waited for several hours).
### Does this PR introduce any user-facing change?
NO
### How was this patch tested?
Existing UT
Closes#27558 from WangGuangxin/windows.
Authored-by: wangguangxin.cn <wangguangxin.cn@gmail.com>
Signed-off-by: herman <herman@databricks.com>
### What changes were proposed in this pull request?
Define a new enumeration `LegacyBehaviorPolicy` in SQLConf, it will be used as the common value for result change configs.
### Why are the changes needed?
During API auditing for the 3.0 release, we found several new approaches that will change the results silently. For these features, we need a common three-value config.
### Does this PR introduce any user-facing change?
Yes, original config `spark.sql.legacy.ctePrecedence.enabled` change to `spark.sql.legacy.ctePrecedencePolicy`.
### How was this patch tested?
Existing UT.
Closes#27579 from xuanyuanking/SPARK-30829.
Authored-by: Yuanjian Li <xyliyuanjian@gmail.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>
### What changes were proposed in this pull request?
See JIRA: https://issues.apache.org/jira/browse/SPARK-29089
Mailing List: http://apache-spark-developers-list.1001551.n3.nabble.com/DataFrameReader-bottleneck-in-DataSource-checkAndGlobPathIfNecessary-when-reading-S3-files-td27828.html
When using DataFrameReader#csv to read many files on S3, globbing and fs.exists on DataSource#checkAndGlobPathIfNecessary becomes a bottleneck.
From the mailing list discussions, an improvement that can be made is to parallelize the blocking FS calls:
> - have SparkHadoopUtils differentiate between files returned by globStatus(), and which therefore exist, and those which it didn't glob for -it will only need to check those.
> - add parallel execution to the glob and existence checks
### Why are the changes needed?
Verifying/globbing files happens on the driver, and if this operations take a long time (for example against S3), then the entire cluster has to wait, potentially sitting idle. This change hopes to make this process faster.
### Does this PR introduce any user-facing change?
No
### How was this patch tested?
I added a test suite `DataSourceSuite` - open to suggestions for better naming.
See [here](https://github.com/apache/spark/pull/25899#issuecomment-534380034) and [here](https://github.com/apache/spark/pull/25899#issuecomment-534069194) for some measurements
Closes#25899 from cozos/master.
Lead-authored-by: Arwin Tio <Arwin.tio@adroll.com>
Co-authored-by: Arwin Tio <arwin.tio@hotmail.com>
Co-authored-by: Arwin Tio <arwin.tio@adroll.com>
Signed-off-by: Sean Owen <srowen@gmail.com>
### What changes were proposed in this pull request?
In the PR, I propose to replace `%` by `Math.floorMod` in `DateTimeUtils.truncTimestamp` for the `SECOND` and `MINUTE` levels.
### Why are the changes needed?
This fixes the issue of incorrect truncation of timestamps before the epoch `1970-01-01T00:00:00.000000Z` to the `SECOND` and `MINUTE` levels. For example, timestamps after the epoch are truncated by cutting off the rest part of the timestamp:
```sql
spark-sql> select date_trunc('SECOND', '2020-02-11 00:01:02.123');
2020-02-11 00:01:02
```
but seconds in the truncated timestamp before the epoch are increased by 1:
```sql
spark-sql> select date_trunc('SECOND', '1960-02-11 00:01:02.123');
1960-02-11 00:01:03
```
### Does this PR introduce any user-facing change?
Yes. After the changes, the example above outputs correct result:
```sql
spark-sql> select date_trunc('SECOND', '1960-02-11 00:01:02.123');
1960-02-11 00:01:02
```
### How was this patch tested?
Added new tests to `DateFunctionsSuite`.
Closes#27543 from MaxGekk/fix-second-minute-truc.
Authored-by: Maxim Gekk <max.gekk@gmail.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>
### What changes were proposed in this pull request?
This is a follow-up for #23124, add a new config `spark.sql.legacy.allowDuplicatedMapKeys` to control the behavior of removing duplicated map keys in build-in functions. With the default value `false`, Spark will throw a RuntimeException while duplicated keys are found.
### Why are the changes needed?
Prevent silent behavior changes.
### Does this PR introduce any user-facing change?
Yes, new config added and the default behavior for duplicated map keys changed to RuntimeException thrown.
### How was this patch tested?
Modify existing UT.
Closes#27478 from xuanyuanking/SPARK-25892-follow.
Authored-by: Yuanjian Li <xyliyuanjian@gmail.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>
### What changes were proposed in this pull request?
1. Rewrite DateTimeUtils methods `getHours()`, `getMinutes()`, `getSeconds()`, `getSecondsWithFraction()`, `getMilliseconds()` and `getMicroseconds()` using Java 8 time APIs. This will automatically switch the `Hour`, `Minute`, `Second` and `DatePart` expressions on Proleptic Gregorian calendar.
2. Remove unused methods and constant of DateTimeUtils - `to2001`, `YearZero `, `toYearZero` and `absoluteMicroSecond()`.
3. Remove unused value `timeZone` from `TimeZoneAwareExpression` since all expressions have been migrated to Java 8 time API, and legacy instance of `TimeZone` is not needed any more.
4. Change signatures of modified DateTimeUtils methods, and pass `ZoneId` instead of `TimeZone`. This will allow to avoid unnecessary conversions `TimeZone` -> `String` -> `ZoneId`.
5. Modify tests in `DateTimeUtilsSuite` and in `DateExpressionsSuite` to pass `ZoneId` instead of `TimeZone`. Correct the tests, to pass tested zone id instead of None.
### Why are the changes needed?
The changes fix the issue of wrong results returned by the `hour()`, `minute()`, `second()`, `date_part('millisecond', ...)` and `date_part('microsecond', ....)`, see example in [SPARK-30843](https://issues.apache.org/jira/browse/SPARK-30843).
### Does this PR introduce any user-facing change?
Yes. After the changes, the results of examples from SPARK-30843:
```sql
spark-sql> select hour(timestamp '0010-01-01 00:00:00');
0
spark-sql> select minute(timestamp '0010-01-01 00:00:00');
0
spark-sql> select second(timestamp '0010-01-01 00:00:00');
0
spark-sql> select date_part('milliseconds', timestamp '0010-01-01 00:00:00');
0.000
spark-sql> select date_part('microseconds', timestamp '0010-01-01 00:00:00');
0
```
### How was this patch tested?
- By existing test suites `DateTimeUtilsSuite`, `DateExpressionsSuite` and `DateFunctionsSuite`.
- Add new tests to `DateExpressionsSuite` and `DateTimeUtilsSuite` for 10 year, like:
```scala
input = date(10, 1, 1, 0, 0, 0, 0, zonePST)
assert(getHours(input, zonePST) === 0)
```
- Re-run `DateTimeBenchmark` using Amazon EC2.
| Item | Description |
| ---- | ----|
| Region | us-west-2 (Oregon) |
| Instance | r3.xlarge |
| AMI | ami-06f2f779464715dc5 (ubuntu/images/hvm-ssd/ubuntu-bionic-18.04-amd64-server-20190722.1) |
| Java | OpenJDK8/11 |
Closes#27596 from MaxGekk/localtimestamp-greg-cal.
Lead-authored-by: Maxim Gekk <max.gekk@gmail.com>
Co-authored-by: Max Gekk <max.gekk@gmail.com>
Co-authored-by: Ubuntu <ubuntu@ip-172-31-1-30.us-west-2.compute.internal>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>
### What changes were proposed in this pull request?
No v2 command supports temp views and the `ResolveCatalogs`/`ResolveSessionCatalog` framework is designed with this assumption.
However, `ResolveSessionCatalog` needs to fallback to v1 commands, which do support temp views (e.g. CACHE TABLE). To work around it, we add a hack in `CatalogAndIdentifier`, which does not expand the given identifier with current namespace if the catalog is session catalog.
This works fine in most cases, as temp views should take precedence over tables during lookup. So if `CatalogAndIdentifier` returns a single name "t", the v1 commands can still resolve it to temp views correctly, or resolve it to table "default.t" if temp view doesn't exist.
However, if users write `spark_catalog.t`, it shouldn't be resolved to temp views as temp views don't belong to any catalog. `CatalogAndIdentifier` can't distinguish between `spark_catalog.t` and `t`, so the caller side may mistakenly resolve `spark_catalog.t` to a temp view.
This PR proposes to fix this issue by
1. remove the hack in `CatalogAndIdentifier`, and clearly document that this shouldn't be used to resolve temp views.
2. update `ResolveSessionCatalog` to explicitly look up temp views first before calling `CatalogAndIdentifier`, for v1 commands that support temp views.
### Why are the changes needed?
To avoid releasing a behavior that we should not support.
Removing the hack also fixes the problem we hit in https://github.com/apache/spark/pull/27532/files#diff-57b3d87be744b7d79a9beacf8e5e5eb2R937
### Does this PR introduce any user-facing change?
yes, now it's not allowed to refer to a temp view with `spark_catalog` prefix.
### How was this patch tested?
new tests
Closes#27550 from cloud-fan/ns.
Authored-by: Wenchen Fan <wenchen@databricks.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>
### What changes were proposed in this pull request?
In the PR, I propose to convert the attribute name of `StringStartsWith` pushed down to the Parquet datasource to column reference via the `nameToParquetField` map. Similar conversions are performed for other source filters pushed down to parquet.
### Why are the changes needed?
This fixes the bug described in [SPARK-30826](https://issues.apache.org/jira/browse/SPARK-30826). The query from an external table:
```sql
CREATE TABLE t1 (col STRING)
USING parquet
OPTIONS (path '$path')
```
created on top of written parquet files by `Seq("42").toDF("COL").write.parquet(path)` returns wrong empty result:
```scala
spark.sql("SELECT * FROM t1 WHERE col LIKE '4%'").show
+---+
|col|
+---+
+---+
```
### Does this PR introduce any user-facing change?
Yes. After the changes the result is correct for the example above:
```scala
spark.sql("SELECT * FROM t1 WHERE col LIKE '4%'").show
+---+
|col|
+---+
| 42|
+---+
```
### How was this patch tested?
Added a test to `ParquetFilterSuite`
Closes#27574 from MaxGekk/parquet-StringStartsWith-case-sens.
Authored-by: Maxim Gekk <max.gekk@gmail.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>
### What changes were proposed in this pull request?
1. `InMemoryTable` was flatting the nested columns, and then the flatten columns was used to look up the indices which is not correct.
This PR implements partitioned by nested column for `InMemoryTable`.
### Why are the changes needed?
This PR implements partitioned by nested column for `InMemoryTable`, so we can test this features in DSv2
### Does this PR introduce any user-facing change?
No.
### How was this patch tested?
Existing unit tests and new tests.
Closes#26929 from dbtsai/addTests.
Authored-by: DB Tsai <d_tsai@apple.com>
Signed-off-by: DB Tsai <d_tsai@apple.com>
### What changes were proposed in this pull request?
In the PR, I propose to use Java 8 time API in timestamp truncations to the levels of `HOUR` and `DAY`. The problem is in the usage of `timeZone.getOffset(millis)` in days/hours truncations where the combined calendar (Julian + Gregorian) is used underneath.
### Why are the changes needed?
The change fix wrong truncations. For example, the following truncation to hours should print `0010-01-01 01:00:00` but it outputs wrong timestamp:
```scala
Seq("0010-01-01 01:02:03.123456").toDF()
.select($"value".cast("timestamp").as("ts"))
.select(date_trunc("HOUR", $"ts").cast("string"))
.show(false)
+------------------------------------+
|CAST(date_trunc(HOUR, ts) AS STRING)|
+------------------------------------+
|0010-01-01 01:30:17 |
+------------------------------------+
```
### Does this PR introduce any user-facing change?
Yes. After the changes, the result of the example above is:
```scala
+------------------------------------+
|CAST(date_trunc(HOUR, ts) AS STRING)|
+------------------------------------+
|0010-01-01 01:00:00 |
+------------------------------------+
```
### How was this patch tested?
- Added new test to `DateFunctionsSuite`
- By `DateExpressionsSuite` and `DateTimeUtilsSuite`
Closes#27512 from MaxGekk/fix-trunc-old-timestamp.
Authored-by: Maxim Gekk <max.gekk@gmail.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>
### What changes were proposed in this pull request?
This PR fixes `DataFrameReader.csv(dataset: Dataset[String])` API to take a `Dataset[String]` originated from a column name different from `value`. This is a long-standing bug started from the very first place.
`CSVUtils.filterCommentAndEmpty` assumed the `Dataset[String]` to be originated with `value` column. This PR changes to use the first column name in the schema.
### Why are the changes needed?
For `DataFrameReader.csv(dataset: Dataset[String])` to support any `Dataset[String]` as the signature indicates.
### Does this PR introduce any user-facing change?
Yes,
```scala
val ds = spark.range(2).selectExpr("concat('a,b,', id) AS text").as[String]
spark.read.option("header", true).option("inferSchema", true).csv(ds).show()
```
Before:
```
org.apache.spark.sql.AnalysisException: cannot resolve '`value`' given input columns: [text];;
'Filter (length(trim('value, None)) > 0)
+- Project [concat(a,b,, cast(id#0L as string)) AS text#2]
+- Range (0, 2, step=1, splits=Some(2))
```
After:
```
+---+---+---+
| a| b| 0|
+---+---+---+
| a| b| 1|
+---+---+---+
```
### How was this patch tested?
Unittest was added.
Closes#27561 from HyukjinKwon/SPARK-30810.
Authored-by: HyukjinKwon <gurwls223@apache.org>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>
### What changes were proposed in this pull request?
We should convert Spark InternalRows to hive data via `HiveInspectors.wrapperFor`.
### Why are the changes needed?
We may hit below exception without this change:
```
[info] org.apache.spark.SparkException: Job aborted due to stage failure: Task 0 in stage 1.0 failed 1 times, most recent failure: Lost task 0.0 in stage 1.0 (TID 1, 192.168.1.6, executor driver): java.lang.ClassCastException: org.apache.spark.sql.types.Decimal cannot be cast to org.apache.hadoop.hive.common.type.HiveDecimal
[info] at org.apache.hadoop.hive.serde2.objectinspector.primitive.JavaHiveDecimalObjectInspector.getPrimitiveJavaObject(JavaHiveDecimalObjectInspector.java:55)
[info] at org.apache.hadoop.hive.serde2.lazy.LazyUtils.writePrimitiveUTF8(LazyUtils.java:321)
[info] at org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe.serialize(LazySimpleSerDe.java:292)
[info] at org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe.serializeField(LazySimpleSerDe.java:247)
[info] at org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe.doSerialize(LazySimpleSerDe.java:231)
[info] at org.apache.hadoop.hive.serde2.AbstractEncodingAwareSerDe.serialize(AbstractEncodingAwareSerDe.java:55)
[info] at org.apache.spark.sql.hive.execution.ScriptTransformationWriterThread.$anonfun$run$2(ScriptTransformationExec.scala:300)
[info] at org.apache.spark.sql.hive.execution.ScriptTransformationWriterThread.$anonfun$run$2$adapted(ScriptTransformationExec.scala:281)
[info] at scala.collection.Iterator.foreach(Iterator.scala:941)
[info] at scala.collection.Iterator.foreach$(Iterator.scala:941)
[info] at scala.collection.AbstractIterator.foreach(Iterator.scala:1429)
[info] at org.apache.spark.sql.hive.execution.ScriptTransformationWriterThread.$anonfun$run$1(ScriptTransformationExec.scala:281)
[info] at scala.runtime.java8.JFunction0$mcV$sp.apply(JFunction0$mcV$sp.java:23)
[info] at org.apache.spark.util.Utils$.logUncaughtExceptions(Utils.scala:1932)
[info] at org.apache.spark.sql.hive.execution.ScriptTransformationWriterThread.run(ScriptTransformationExec.scala:270)
```
### Does this PR introduce any user-facing change?
No.
### How was this patch tested?
Added new test. But please note that this test returns different result between Hive1.2 and Hive2.3 due to `HiveDecimal` or `SerDe` difference(don't know the root cause yet).
Closes#27556 from Ngone51/script_transform.
Lead-authored-by: yi.wu <yi.wu@databricks.com>
Co-authored-by: Wenchen Fan <wenchen@databricks.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>
### What changes were proposed in this pull request?
This PR makes sure AQE is either enabled or disabled for the entire query, including the main query and all subqueries.
Currently there are unsupported queries by AQE, e.g., queries that contain DPP filters. We need to make sure that if the main query is unsupported, none of the sub-queries should apply AQE, otherwise it can lead to performance regressions due to missed opportunity of sub-query reuse.
### Why are the changes needed?
To get rid of potential perf regressions when AQE is turned on.
### Does this PR introduce any user-facing change?
No.
### How was this patch tested?
Updated DynamicPartitionPruningSuite:
1. Removed the existing workaround `withSQLConf(SQLConf.ADAPTIVE_EXECUTION_ENABLED.key, "false")`
2. Added `DynamicPartitionPruningSuiteAEOn` and `DynamicPartitionPruningSuiteAEOff` to enable testing this suite with AQE on and off options
3. Added a check in `checkPartitionPruningPredicate` to verify that the subqueries are always in sync with the main query in terms of whether AQE is applied.
Closes#27554 from maryannxue/spark-30801.
Authored-by: maryannxue <maryannxue@apache.org>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>
### What changes were proposed in this pull request?
This PR scopes `SparkSession.active` to prevent problems with processing queries with possibly different spark sessions (and different configs). A new method, `withActive` is introduced on `SparkSession` that restores the previous spark session after the block of code is executed.
### Why are the changes needed?
`SparkSession.active` is a thread local variable that points to the current thread's spark session. It is important to note that the `SQLConf.get` method depends on `SparkSession.active`. In the current implementation it is possible that `SparkSession.active` points to a different session which causes various problems. Most of these problems arise because part of the query processing is done using the configurations of a different session. For example, when creating a data frame using a new session, i.e., `session.sql("...")`, part of the data frame is constructed using the currently active spark session, which can be a different session from the one used later for processing the query.
### Does this PR introduce any user-facing change?
The `withActive` method is introduced on `SparkSession`.
### How was this patch tested?
Unit tests (to be added)
Closes#27387 from dbaliafroozeh/UseWithActiveSessionInQueryExecution.
Authored-by: Ali Afroozeh <ali.afroozeh@databricks.com>
Signed-off-by: herman <herman@databricks.com>
<!--
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.
-->
### 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 is a followup of https://github.com/apache/spark/pull/26434
This PR use one special shuffle reader for skew join, so that we only have one join after optimization. In order to do that, this PR
1. add a very general `CustomShuffledRowRDD` which support all kind of partition arrangement.
2. move the logic of coalescing shuffle partitions to a util function, and call it during skew join optimization, to totally decouple with the `ReduceNumShufflePartitions` rule. It's too complicated to interfere skew join with `ReduceNumShufflePartitions`, as you need to consider the size of split partitions which don't respect target size already.
### 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.
-->
The current skew join optimization has a serious performance issue: the size of the query plan depends on the number and size of skewed partitions.
### Does this PR introduce any user-facing change?
<!--
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 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.
-->
existing tests
test UI manually:
explain output
```
AdaptiveSparkPlan(isFinalPlan=true)
+- OverwriteByExpression org.apache.spark.sql.execution.datasources.noop.NoopTable$403a2ed5, [AlwaysTrue()], org.apache.spark.sql.util.CaseInsensitiveStringMap1f
+- *(5) SortMergeJoin(skew=true) [key1#2L], [key2#6L], Inner
:- *(3) Sort [key1#2L ASC NULLS FIRST], false, 0
: +- SkewJoinShuffleReader 2 skewed partitions with size(max=5 KB, min=5 KB, avg=5 KB)
: +- ShuffleQueryStage 0
: +- Exchange hashpartitioning(key1#2L, 200), true, [id=#53]
: +- *(1) Project [(id#0L % 2) AS key1#2L]
: +- *(1) Filter isnotnull((id#0L % 2))
: +- *(1) Range (0, 100000, step=1, splits=6)
+- *(4) Sort [key2#6L ASC NULLS FIRST], false, 0
+- SkewJoinShuffleReader 2 skewed partitions with size(max=5 KB, min=5 KB, avg=5 KB)
+- ShuffleQueryStage 1
+- Exchange hashpartitioning(key2#6L, 200), true, [id=#64]
+- *(2) Project [((id#4L % 2) + 1) AS key2#6L]
+- *(2) Filter isnotnull(((id#4L % 2) + 1))
+- *(2) Range (0, 100000, step=1, splits=6)
```
Closes#27493 from cloud-fan/aqe.
Authored-by: Wenchen Fan <wenchen@databricks.com>
Signed-off-by: herman <herman@databricks.com>
### What changes were proposed in this pull request?
Although Spark SQL support bracketed comments, but `SQLQueryTestSuite` can't treat bracketed comments well and lead to generated golden files can't display bracketed comments well.
This PR will improve the treatment of bracketed comments and add three test case in `PlanParserSuite`.
Spark SQL can't support nested bracketed comments and https://github.com/apache/spark/pull/27495 used to support it.
### Why are the changes needed?
Golden files can't display well.
### Does this PR introduce any user-facing change?
No
### How was this patch tested?
New UT.
Closes#27481 from beliefer/ansi-brancket-comments.
Authored-by: beliefer <beliefer@163.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>
### What changes were proposed in this pull request?
This PR proposes to support Hive-style `ALTER TABLE ... REPLACE COLUMNS ...` as described in https://cwiki.apache.org/confluence/display/Hive/LanguageManual+DDL#LanguageManualDDL-Add/ReplaceColumns
The user now can do the following:
```SQL
CREATE TABLE t (col1 int, col2 int) USING Foo;
ALTER TABLE t REPLACE COLUMNS (col2 string COMMENT 'comment2', col3 int COMMENT 'comment3');
```
, which drops the existing columns `col1` and `col2`, and add new columns `col2` and `col3`.
### Why are the changes needed?
This is a new DDL statement. Spark currently supports the Hive-style `ALTER TABLE ... CHANGE COLUMN ...`, so this new addition can be useful.
### Does this PR introduce any user-facing change?
Yes, adding a new DDL statement.
### How was this patch tested?
More tests to be added.
Closes#27482 from imback82/replace_cols.
Authored-by: Terry Kim <yuminkim@gmail.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>
### What changes were proposed in this pull request?
This PR adds a config for Dynamic Partition Pruning subquery duplication and turns it off by default due to its potential performance regression.
When planning a DPP filter, it seeks to reuse the broadcast exchange relation if the corresponding join is a BHJ with the filter relation being on the build side, otherwise it will either opt out or plan the filter as an un-reusable subquery duplication based on the cost estimate. However, the cost estimate is not accurate and only takes into account the table scan overhead, thus adding an un-reusable subquery duplication DPP filter can sometimes cause perf regression.
This PR turns off the subquery duplication DPP filter by:
1. adding a config `spark.sql.optimizer.dynamicPartitionPruning.reuseBroadcastOnly` and setting it `true` by default.
2. removing the existing meaningless config `spark.sql.optimizer.dynamicPartitionPruning.reuseBroadcast` since we always want to reuse broadcast results if possible.
### Why are the changes needed?
This is to fix a potential performance regression caused by DPP.
### Does this PR introduce any user-facing change?
No.
### How was this patch tested?
Updated DynamicPartitionPruningSuite to test the new configuration.
Closes#27551 from maryannxue/spark-30528.
Authored-by: maryannxue <maryannxue@apache.org>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>
### What changes were proposed in this pull request?
`spark.sql("select map()")` returns {}.
After these changes it will return map<null,null>
### Why are the changes needed?
After changes introduced due to #27521, it is important to maintain consistency while using map().
### Does this PR introduce any user-facing change?
Yes. Now map() will give map<null,null> instead of {}.
### How was this patch tested?
UT added. Migration guide updated as well
Closes#27542 from iRakson/SPARK-30790.
Authored-by: iRakson <raksonrakesh@gmail.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>
### What changes were proposed in this pull request?
This is another PR for stage level scheduling. In particular this adds changes to the dynamic allocation manager and the scheduler backend to be able to track what executors are needed per ResourceProfile. Note the api is still private to Spark until the entire feature gets in, so this functionality will be there but only usable by tests for profiles other then the DefaultProfile.
The main changes here are simply tracking things on a ResourceProfile basis as well as sending the executor requests to the scheduler backend for all ResourceProfiles.
I introduce a ResourceProfileManager in this PR that will track all the actual ResourceProfile objects so that we can keep them all in a single place and just pass around and use in datastructures the resource profile id. The resource profile id can be used with the ResourceProfileManager to get the actual ResourceProfile contents.
There are various places in the code that use executor "slots" for things. The ResourceProfile adds functionality to keep that calculation in it. This logic is more complex then it should due to standalone mode and mesos coarse grained not setting the executor cores config. They default to all cores on the worker, so calculating slots is harder there.
This PR keeps the functionality to make the cores the limiting resource because the scheduler still uses that for "slots" for a few things.
This PR does also add the resource profile id to the Stage and stage info classes to be able to test things easier. That full set of changes will come with the scheduler PR that will be after this one.
The PR stops at the scheduler backend pieces for the cluster manager and the real YARN support hasn't been added in this PR, that again will be in a separate PR, so this has a few of the API changes up to the cluster manager and then just uses the default profile requests to continue.
The code for the entire feature is here for reference: https://github.com/apache/spark/pull/27053/files although it needs to be upmerged again as well.
### Why are the changes needed?
Needed for stage level scheduling feature.
### Does this PR introduce any user-facing change?
No user facing api changes added here.
### How was this patch tested?
Lots of unit tests and manually testing. I tested on yarn, k8s, standalone, local modes. Ran both failure and success cases.
Closes#27313 from tgravescs/SPARK-29148.
Authored-by: Thomas Graves <tgraves@nvidia.com>
Signed-off-by: Thomas Graves <tgraves@apache.org>
### What changes were proposed in this pull request?
This is a follow-up to the PR #26956. In #26956, the patch proposed to preserve path permission when truncating table. When setting up original ACLs, we need to set user/group/other permission as ACL entries too, otherwise if the path doesn't have default user/group/other ACL entries, ACL API will complain an error `Invalid ACL: the user, group and other entries are required.`.
In short this change makes sure:
1. Permissions for user/group/other are always kept into ACLs to work with ACL API.
2. Other custom ACLs are still kept after TRUNCATE TABLE (#26956 did this).
### Why are the changes needed?
Without this fix, `TRUNCATE TABLE` will get an error when setting up ACLs if there is no default default user/group/other ACL entries.
### Does this PR introduce any user-facing change?
No
### How was this patch tested?
Update unit test.
Manual test on dev Spark cluster.
Set ACLs for a table path without default user/group/other ACL entries:
```
hdfs dfs -setfacl --set 'user:liangchi:rwx,user::rwx,group::r--,other::r--' /user/hive/warehouse/test.db/test_truncate_table
hdfs dfs -getfacl /user/hive/warehouse/test.db/test_truncate_table
# file: /user/hive/warehouse/test.db/test_truncate_table
# owner: liangchi
# group: supergroup
user::rwx
user:liangchi:rwx
group::r--
mask::rwx
other::r--
```
Then run `sql("truncate table test.test_truncate_table")`, it works by normally truncating the table and preserve ACLs.
Closes#27548 from viirya/fix-truncate-table-permission.
Lead-authored-by: Liang-Chi Hsieh <liangchi@uber.com>
Co-authored-by: Liang-Chi Hsieh <viirya@gmail.com>
Signed-off-by: Dongjoon Hyun <dhyun@apple.com>
### What changes were proposed in this pull request?
In the PR, I propose to rewrite the `millisToDays` and `daysToMillis` of `DateTimeUtils` using Java 8 time API.
I removed `getOffsetFromLocalMillis` from `DateTimeUtils` because it is a private methods, and is not used anymore in Spark SQL.
### Why are the changes needed?
New implementation is based on Proleptic Gregorian calendar which has been already used by other date-time functions. This changes make `millisToDays` and `daysToMillis` consistent to rest Spark SQL API related to date & time operations.
### Does this PR introduce any user-facing change?
Yes, this might effect behavior for old dates before 1582 year.
### How was this patch tested?
By existing test suites `DateTimeUtilsSuite`, `DateFunctionsSuite`, DateExpressionsSuite`, `SQLQuerySuite` and `HiveResultSuite`.
Closes#27494 from MaxGekk/millis-2-days-java8-api.
Authored-by: Maxim Gekk <max.gekk@gmail.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>
### What changes were proposed in this pull request?
Added new test to `RegexpExpressionsSuite` which checks that `cache` of compiled pattern is set when the `right` expression (pattern in `LIKE`) is a foldable expression.
### Why are the changes needed?
To be sure that `cache` in `StringRegexExpression` is initialized for foldable patterns.
### Does this PR introduce any user-facing change?
No
### How was this patch tested?
By running the added test in `RegexpExpressionsSuite`.
Closes#27547 from MaxGekk/regexp-cache-test.
Authored-by: Maxim Gekk <max.gekk@gmail.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>
### What changes were proposed in this pull request?
In the PR, I propose to add legacy date/timestamp formatters based on `SimpleDateFormat` and `FastDateFormat`:
- `LegacyFastTimestampFormatter` - uses `FastDateFormat` and supports parsing/formatting in microsecond precision. The code was borrowed from Spark 2.4, see https://github.com/apache/spark/pull/26507 & https://github.com/apache/spark/pull/26582
- `LegacySimpleTimestampFormatter` uses `SimpleDateFormat`, and support the `lenient` mode. When the `lenient` parameter is set to `false`, the parser become much stronger in checking its input.
### Why are the changes needed?
Spark 2.4.x uses the following parsers for parsing/formatting date/timestamp strings:
- `DateTimeFormat` in CSV/JSON datasource
- `SimpleDateFormat` - is used in JDBC datasource, in partitions parsing.
- `SimpleDateFormat` in strong mode (`lenient = false`), see https://github.com/apache/spark/blob/branch-2.4/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/util/DateTimeUtils.scala#L124. It is used by the `date_format`, `from_unixtime`, `unix_timestamp` and `to_unix_timestamp` functions.
The PR aims to make Spark 3.0 compatible with Spark 2.4.x in all those cases when `spark.sql.legacy.timeParser.enabled` is set to `true`.
### Does this PR introduce any user-facing change?
This shouldn't change behavior with default settings. If `spark.sql.legacy.timeParser.enabled` is set to `true`, users should observe behavior of Spark 2.4.
### How was this patch tested?
- Modified tests in `DateExpressionsSuite` to check the legacy parser - `SimpleDateFormat`.
- Added `CSVLegacyTimeParserSuite` and `JsonLegacyTimeParserSuite` to run `CSVSuite` and `JsonSuite` with the legacy parser - `FastDateFormat`.
Closes#27524 from MaxGekk/timestamp-formatter-legacy-fallback.
Authored-by: Maxim Gekk <max.gekk@gmail.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>
### What changes were proposed in this pull request?
The current implement of `regexp_extract` will throws a unprocessed exception show below:
`SELECT regexp_extract('1a 2b 14m', 'd+')`
```
java.lang.IndexOutOfBoundsException: No group 1
[info] at java.util.regex.Matcher.group(Matcher.java:538)
[info] at org.apache.spark.sql.catalyst.expressions.GeneratedClass$GeneratedIteratorForCodegenStage1.processNext(Unknown Source)
[info] at org.apache.spark.sql.execution.BufferedRowIterator.hasNext(BufferedRowIterator.java:43)
[info] at org.apache.spark.sql.execution.WholeStageCodegenExec$$anon$1.hasNext(WholeStageCodegenExec.scala:729)
```
I think should treat this exception well.
### Why are the changes needed?
Fix a bug `java.lang.IndexOutOfBoundsException No group 1 `
### Does this PR introduce any user-facing change?
Yes
### How was this patch tested?
New UT
Closes#27508 from beliefer/fix-regexp_extract-bug.
Authored-by: beliefer <beliefer@163.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>
### What changes were proposed in this pull request?
This PR proposes to make the `code` string interpolator treat escapes the same way as Scala's builtin `StringContext.s()` string interpolator. This will remove the need for an ugly workaround in `Like` expression's codegen.
### Why are the changes needed?
The `code()` string interpolator in Spark SQL's code generator should treat escapes like Scala's builtin `StringContext.s()` interpolator, i.e. it should treat escapes in the code parts, and should not treat escapes in the input arguments.
For example,
```scala
val arg = "This is an argument."
val str = s"This is string part 1. $arg This is string part 2."
val code = code"This is string part 1. $arg This is string part 2."
assert(code.toString == str)
```
We should expect the `code()` interpolator to produce the same result as the `StringContext.s()` interpolator, where only escapes in the string parts should be treated, while the args should be kept verbatim.
But in the current implementation, due to the eager folding of code parts and literal input args, the escape treatment is incorrectly done on both code parts and literal args.
That causes a problem when an arg contains escape sequences and wants to preserve that in the final produced code string. For example, in `Like` expression's codegen, there's an ugly workaround for this bug:
```scala
// We need double escape to avoid org.codehaus.commons.compiler.CompileException.
// '\\' will cause exception 'Single quote must be backslash-escaped in character literal'.
// '\"' will cause exception 'Line break in literal not allowed'.
val newEscapeChar = if (escapeChar == '\"' || escapeChar == '\\') {
s"""\\\\\\$escapeChar"""
} else {
escapeChar
}
```
### Does this PR introduce any user-facing change?
No.
### How was this patch tested?
Added a new unit test case in `CodeBlockSuite`.
Closes#27544 from rednaxelafx/fix-code-string-interpolator.
Authored-by: Kris Mok <kris.mok@databricks.com>
Signed-off-by: HyukjinKwon <gurwls223@apache.org>
### What changes were proposed in this pull request?
This is a small follow-up for https://github.com/apache/spark/pull/27400. This PR makes an empty `LocalTableScanExec` return an `RDD` without partitions.
### Why are the changes needed?
It is a bit unexpected that the RDD contains partitions if there is not work to do. It also can save a bit of work when this is used in a more complex plan.
### Does this PR introduce any user-facing change?
No
### How was this patch tested?
Added test to `SparkPlanSuite`.
Closes#27530 from hvanhovell/SPARK-30780.
Authored-by: herman <herman@databricks.com>
Signed-off-by: HyukjinKwon <gurwls223@apache.org>
### What changes were proposed in this pull request?
In the PR, I propose to revert the commit 8aebc80e0e.
### Why are the changes needed?
See the concerns https://github.com/apache/spark/pull/27355#issuecomment-584344438
### Does this PR introduce any user-facing change?
No
### How was this patch tested?
By existing test suites.
Closes#27531 from MaxGekk/revert-like-3-args.
Authored-by: Maxim Gekk <max.gekk@gmail.com>
Signed-off-by: Dongjoon Hyun <dhyun@apple.com>
### What changes were proposed in this pull request?
This PR adds some more information and context to `spark.sql.defaultUrlStreamHandlerFactory.enabled`.
### Why are the changes needed?
It is a bit difficult to understand the documentation of `spark.sql.defaultUrlStreamHandlerFactory.enabled`.
### Does this PR introduce any user-facing change?
Nope, internal doc only fix.
### How was this patch tested?
Nope. I only tested linter.
Closes#27541 from HyukjinKwon/SPARK-29462-followup.
Authored-by: HyukjinKwon <gurwls223@apache.org>
Signed-off-by: Dongjoon Hyun <dhyun@apple.com>
### What changes were proposed in this pull request?
In the case of back-to-back calculation of `floorDiv` and `floorMod` with the same arguments, the result of `foorDiv` can be reused in calculation of `floorMod`. The `floorMod` method is defined as the following in Java standard library:
```java
public static int floorMod(int x, int y) {
int r = x - floorDiv(x, y) * y;
return r;
}
```
If `floorDiv(x, y)` has been already calculated, it can be reused in `x - floorDiv(x, y) * y`.
I propose to modify 2 places in `DateTimeUtils`:
1. `microsToInstant` which is widely used in many date-time functions. `Math.floorMod(us, MICROS_PER_SECOND)` is just replaced by its definition from Java Math library.
2. `truncDate`: `Math.floorMod(oldYear, divider) == 0` is replaced by `Math.floorDiv(oldYear, divider) * divider == oldYear` where `floorDiv(...) * divider` is pre-calculated.
### Why are the changes needed?
This reduces the number of arithmetic operations, and can slightly improve performance of date-time functions.
### Does this PR introduce any user-facing change?
No
### How was this patch tested?
By existing test suites `DateTimeUtilsSuite`, `DateFunctionsSuite` and `DateExpressionsSuite`.
Closes#27491 from MaxGekk/opt-microsToInstant.
Authored-by: Maxim Gekk <max.gekk@gmail.com>
Signed-off-by: Sean Owen <srowen@gmail.com>
### What changes were proposed in this pull request?
Add class document for PruneFileSourcePartitions and PruneHiveTablePartitions.
### Why are the changes needed?
To describe these two classes.
### Does this PR introduce any user-facing change?
no
### How was this patch tested?
no
Closes#27535 from fuwhu/SPARK-15616-FOLLOW-UP.
Authored-by: fuwhu <bestwwg@163.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>
### What changes were proposed in this pull request?
This brings https://github.com/apache/spark/pull/26324 back. It was reverted basically because, firstly Hive compatibility, and the lack of investigations in other DBMSes and ANSI.
- In case of PostgreSQL seems coercing NULL literal to TEXT type.
- Presto seems coercing `array() + array(1)` -> array of int.
- Hive seems `array() + array(1)` -> array of strings
Given that, the design choices have been differently made for some reasons. If we pick one of both, seems coercing to array of int makes much more sense.
Another investigation was made offline internally. Seems ANSI SQL 2011, section 6.5 "<contextually typed value specification>" states:
> If ES is specified, then let ET be the element type determined by the context in which ES appears. The declared type DT of ES is Case:
>
> a) If ES simply contains ARRAY, then ET ARRAY[0].
>
> b) If ES simply contains MULTISET, then ET MULTISET.
>
> ES is effectively replaced by CAST ( ES AS DT )
From reading other related context, doing it to `NullType`. Given the investigation made, choosing to `null` seems correct, and we have a reference Presto now. Therefore, this PR proposes to bring it back.
### Why are the changes needed?
When empty array is created, it should be declared as array<null>.
### Does this PR introduce any user-facing change?
Yes, `array()` creates `array<null>`. Now `array(1) + array()` can correctly create `array(1)` instead of `array("1")`.
### How was this patch tested?
Tested manually
Closes#27521 from HyukjinKwon/SPARK-29462.
Lead-authored-by: HyukjinKwon <gurwls223@apache.org>
Co-authored-by: Aman Omer <amanomer1996@gmail.com>
Signed-off-by: HyukjinKwon <gurwls223@apache.org>
### What changes were proposed in this pull request?
- Fix the scope of `Logging.initializeForcefully` so that it doesn't appear in subclasses' public methods. Right now, `sc.initializeForcefully(false, false)` is allowed to called.
- Don't show classes under `org.apache.spark.internal` package in API docs.
- Add missing `since` annotation.
- Fix the scope of `ArrowUtils` to remove it from the API docs.
### Why are the changes needed?
Avoid leaking APIs unintentionally in Spark 3.0.0.
### Does this PR introduce any user-facing change?
No. All these changes are to avoid leaking APIs unintentionally in Spark 3.0.0.
### How was this patch tested?
Manually generated the API docs and verified the above issues have been fixed.
Closes#27528 from zsxwing/audit-ss-apis.
Authored-by: Shixiong Zhu <zsxwing@gmail.com>
Signed-off-by: Xiao Li <gatorsmile@gmail.com>
### What changes were proposed in this pull request?
Follow up for #27267, reset the status changed in SQLExecution.withThreadLocalCaptured.
### Why are the changes needed?
For code safety.
### Does this PR introduce any user-facing change?
No.
### How was this patch tested?
Existing UT.
Closes#27516 from xuanyuanking/SPARK-30556-follow.
Authored-by: Yuanjian Li <xyliyuanjian@gmail.com>
Signed-off-by: herman <herman@databricks.com>
### What changes were proposed in this pull request?
In the PR, I propose to fix `cache` initialization in `StringRegexExpression` by changing `case Literal(value: String, StringType)` to `case p: Expression if p.foldable`
### Why are the changes needed?
Actually, the case doesn't work at all because of:
1. Literals value has type `UTF8String`
2. It doesn't work for foldable expressions like in the example:
```sql
SELECT '%SystemDrive%\Users\John' _FUNC_ '%SystemDrive%\\Users.*';
```
<img width="649" alt="Screen Shot 2020-02-08 at 22 45 50" src="https://user-images.githubusercontent.com/1580697/74091681-0d4a2180-4acb-11ea-8a0d-7e8c65f4214e.png">
### Does this PR introduce any user-facing change?
No
### How was this patch tested?
By the `check outputs of expression examples` test from `SQLQuerySuite`.
Closes#27502 from MaxGekk/str-regexp-foldable-pattern.
Authored-by: Maxim Gekk <max.gekk@gmail.com>
Signed-off-by: Dongjoon Hyun <dhyun@apple.com>
### What changes were proposed in this pull request?
This reverts commit 8ce7962931. There's variable name conflicts with 8aebc80e0e (diff-39298b470865a4cbc67398a4ea11e767).
This can be cleanly ported back to branch-3.0.
### Why are the changes needed?
Performance investigation were not made enough and it's not clear if it really beneficial or now.
### Does this PR introduce any user-facing change?
No.
### How was this patch tested?
Jenkins tests.
Closes#27514 from HyukjinKwon/revert-cache-PR.
Authored-by: HyukjinKwon <gurwls223@apache.org>
Signed-off-by: Xiao Li <gatorsmile@gmail.com>
### What changes were proposed in this pull request?
This is a follow-up for #24938 to tweak error message and migration doc.
### Why are the changes needed?
Making user know workaround if SHOW CREATE TABLE doesn't work for some Hive tables.
### Does this PR introduce any user-facing change?
No
### How was this patch tested?
Existing unit tests.
Closes#27505 from viirya/SPARK-27946-followup.
Authored-by: Liang-Chi Hsieh <viirya@gmail.com>
Signed-off-by: Liang-Chi Hsieh <liangchi@uber.com>
### What changes were proposed in this pull request?
Enhance RuleExecutor strategy to take different actions when exceeding max iterations. And raise exception if analyzer exceed max iterations.
### Why are the changes needed?
Currently, both analyzer and optimizer just log warning message if rule execution exceed max iterations. They should have different behavior. Analyzer should raise exception to indicates the plan is not fixed after max iterations, while optimizer just log warning to keep the current plan. This is more feasible after SPARK-30138 was introduced.
### Does this PR introduce any user-facing change?
No
### How was this patch tested?
Add test in AnalysisSuite
Closes#26977 from Eric5553/EnhanceMaxIterations.
Authored-by: Eric Wu <492960551@qq.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>
### What changes were proposed in this pull request?
This is a follow up in [#27452](https://github.com/apache/spark/pull/27452).
Add a unit test to verify whether the log warning is print when intentionally skip AQE.
### Why are the changes needed?
Add unit test
### Does this PR introduce any user-facing change?
No
### How was this patch tested?
adding unit test
Closes#27515 from JkSelf/aqeLoggingWarningTest.
Authored-by: jiake <ke.a.jia@intel.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>
### What changes were proposed in this pull request?
This PR updates the documentation on `TableCatalog.alterTable`s behavior on the order by which the requested changes are applied. It now explicitly mentions that the changes are applied in the order given.
### Why are the changes needed?
The current documentation on `TableCatalog.alterTable` doesn't mention which order the requested changes are applied. It will be useful to explicitly document this behavior so that the user can expect the behavior. For example, `REPLACE COLUMNS` needs to delete columns before adding new columns, and if the order is guaranteed by `alterTable`, it's much easier to work with the catalog API.
### Does this PR introduce any user-facing change?
Yes, document change.
### How was this patch tested?
Not added (doc changes).
Closes#27496 from imback82/catalog_table_alter_table.
Authored-by: Terry Kim <yuminkim@gmail.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>
### What changes were proposed in this pull request?
Add round-trip tests for CSV and JSON functions as https://github.com/apache/spark/pull/27317#discussion_r376745135 asked.
### Why are the changes needed?
improve test coverage
### Does this PR introduce any user-facing change?
no
### How was this patch tested?
add uts
Closes#27510 from yaooqinn/SPARK-30592-F.
Authored-by: Kent Yao <yaooqinn@hotmail.com>
Signed-off-by: HyukjinKwon <gurwls223@apache.org>
This reverts commit a0e63b61e7.
### What changes were proposed in this pull request?
This reverts the patch at #26978 based on gatorsmile's suggestion.
### Why are the changes needed?
Original patch #26978 has not considered a corner case. We may need to put more time on ensuring we can cover all cases.
### Does this PR introduce any user-facing change?
No
### How was this patch tested?
Unit test.
Closes#27504 from viirya/revert-SPARK-29721.
Authored-by: Liang-Chi Hsieh <viirya@gmail.com>
Signed-off-by: Xiao Li <gatorsmile@gmail.com>
### What changes were proposed in this pull request?
Simplify the changes for adding metrics description for WholeStageCodegen in https://github.com/apache/spark/pull/27405
### Why are the changes needed?
In https://github.com/apache/spark/pull/27405, the UI changes can be made without using the function `adjustPositionOfOperationName` to adjust the position of operation name and mark as an operation-name class.
I suggest we make simpler changes so that it would be easier for future development.
### Does this PR introduce any user-facing change?
No
### How was this patch tested?
Manual test with the queries provided in https://github.com/apache/spark/pull/27405
```
sc.parallelize(1 to 10).toDF.sort("value").filter("value > 1").selectExpr("value * 2").show
sc.parallelize(1 to 10).toDF.sort("value").filter("value > 1").selectExpr("value * 2").write.format("json").mode("overwrite").save("/tmp/test_output")
sc.parallelize(1 to 10).toDF.write.format("json").mode("append").save("/tmp/test_output")
```
Closes#27490 from gengliangwang/wholeCodegenUI.
Authored-by: Gengliang Wang <gengliang.wang@databricks.com>
Signed-off-by: Gengliang Wang <gengliang.wang@databricks.com>
### What changes were proposed in this pull request?
This is a follow-up for #25029, in this PR we throw an AnalysisException when name conflict is detected in nested WITH clause. In this way, the config `spark.sql.legacy.ctePrecedence.enabled` should be set explicitly for the expected behavior.
### Why are the changes needed?
The original change might risky to end-users, it changes behavior silently.
### Does this PR introduce any user-facing change?
Yes, change the config `spark.sql.legacy.ctePrecedence.enabled` as optional.
### How was this patch tested?
New UT.
Closes#27454 from xuanyuanking/SPARK-28228-follow.
Authored-by: Yuanjian Li <xyliyuanjian@gmail.com>
Signed-off-by: Dongjoon Hyun <dhyun@apple.com>
### What changes were proposed in this pull request?
The current ALTER COLUMN syntax allows to change multiple properties at a time:
```
ALTER TABLE table=multipartIdentifier
(ALTER | CHANGE) COLUMN? column=multipartIdentifier
(TYPE dataType)?
(COMMENT comment=STRING)?
colPosition?
```
The SQL standard (section 11.12) only allows changing one property at a time. This is also true on other recent SQL systems like [snowflake](https://docs.snowflake.net/manuals/sql-reference/sql/alter-table-column.html) and [redshift](https://docs.aws.amazon.com/redshift/latest/dg/r_ALTER_TABLE.html). (credit to cloud-fan)
This PR proposes to change ALTER COLUMN to follow SQL standard, thus allows altering only one column property at a time.
Note that ALTER COLUMN syntax being changed here is newly added in Spark 3.0, so it doesn't affect Spark 2.4 behavior.
### Why are the changes needed?
To follow SQL standard (and other recent SQL systems) behavior.
### Does this PR introduce any user-facing change?
Yes, now the user can update the column properties only one at a time.
For example,
```
ALTER TABLE table1 ALTER COLUMN a.b.c TYPE bigint COMMENT 'new comment'
```
should be broken into
```
ALTER TABLE table1 ALTER COLUMN a.b.c TYPE bigint
ALTER TABLE table1 ALTER COLUMN a.b.c COMMENT 'new comment'
```
### How was this patch tested?
Updated existing tests.
Closes#27444 from imback82/alter_column_one_at_a_time.
Authored-by: Terry Kim <yuminkim@gmail.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>
### What changes were proposed in this pull request?
- Rewrite the `convertTz` method of `DateTimeUtils` using Java 8 time API
- Change types of `convertTz` parameters from `TimeZone` to `ZoneId`. This allows to avoid unnecessary conversions `TimeZone` -> `ZoneId` and performance regressions as a consequence.
### Why are the changes needed?
- Fixes incorrect behavior of `to_utc_timestamp` on daylight saving day. For example:
```scala
scala> df.select(to_utc_timestamp(lit("2019-11-03T12:00:00"), "Asia/Hong_Kong").as("local UTC")).show
+-------------------+
| local UTC|
+-------------------+
|2019-11-03 03:00:00|
+-------------------+
```
but the result must be 2019-11-03 04:00:00:
<img width="1013" alt="Screen Shot 2020-02-06 at 20 09 36" src="https://user-images.githubusercontent.com/1580697/73960846-a129bb00-491c-11ea-92f5-45831cb28a62.png">
- Simplifies the code, and make it more maintainable
- Switches `convertTz` on Proleptic Gregorian calendar used by Java 8 time classes by default. That makes the function consistent to other date-time functions.
### Does this PR introduce any user-facing change?
Yes, after the changes `to_utc_timestamp` returns the correct result `2019-11-03 04:00:00`.
### How was this patch tested?
- By existing test suite `DateTimeUtilsSuite`, `DateFunctionsSuite` and `DateExpressionsSuite`.
- Added `convert time zones on a daylight saving day` to DateFunctionsSuite
Closes#27474 from MaxGekk/port-convertTz-on-Java8-api.
Authored-by: Maxim Gekk <max.gekk@gmail.com>
Signed-off-by: Wenchen Fan <wenchen@databricks.com>
### What changes were proposed in this pull request?
This is a followup of https://github.com/apache/spark/pull/26656.
We don't support window aggregate function with filter predicate yet and we should fail explicitly.
Observable metrics has the same issue. This PR fixes it as well.
### Why are the changes needed?
If we simply ignore filter predicate when we don't support it, the result is wrong.
### Does this PR introduce any user-facing change?
yea, fix the query result.
### How was this patch tested?
new tests
Closes#27476 from cloud-fan/filter.
Authored-by: Wenchen Fan <wenchen@databricks.com>
Signed-off-by: Dongjoon Hyun <dhyun@apple.com>
