Debug SparkR test failure #15

HyukjinKwon · 2020-07-30T07:55:42Z

No description provided.

This reverts commit 0450ae3.

…itHub Actions ### What changes were proposed in this pull request? This PR proposes to manually install R instead of using `setup-r` which seems broken. Currently, GitHub Actions uses its default R 3.4.4 installed, which we dropped as of SPARK-32073. While I am here, I am also upgrading R version to 4.0. Jenkins will test the old version and GitHub Actions tests the new version. AppVeyor uses R 4.0 but it does not check CRAN which is important when we make a release. ### Why are the changes needed? To recover GitHub Actions build. ### Does this PR introduce _any_ user-facing change? No, dev-only ### How was this patch tested? Manually tested at HyukjinKwon#15 Closes #29302 from HyukjinKwon/SPARK-32493. Authored-by: HyukjinKwon <gurwls223@apache.org> Signed-off-by: HyukjinKwon <gurwls223@apache.org>

…itHub Actions This PR proposes to manually install R instead of using `setup-r` which seems broken. Currently, GitHub Actions uses its default R 3.4.4 installed, which we dropped as of SPARK-32073. While I am here, I am also upgrading R version to 4.0. Jenkins will test the old version and GitHub Actions tests the new version. AppVeyor uses R 4.0 but it does not check CRAN which is important when we make a release. To recover GitHub Actions build. No, dev-only Manually tested at #15 Closes apache#29302 from HyukjinKwon/SPARK-32493. Authored-by: HyukjinKwon <gurwls223@apache.org> Signed-off-by: HyukjinKwon <gurwls223@apache.org>

…onnect ### What changes were proposed in this pull request? Implement Arrow-optimized Python UDFs in Spark Connect. Please see apache#39384 for motivation and performance improvements of Arrow-optimized Python UDFs. ### Why are the changes needed? Parity with vanilla PySpark. ### Does this PR introduce _any_ user-facing change? Yes. In Spark Connect Python Client, users can: 1. Set `useArrow` parameter True to enable Arrow optimization for a specific Python UDF. ```sh >>> df = spark.range(2) >>> df.select(udf(lambda x : x + 1, useArrow=True)('id')).show() +------------+ |<lambda>(id)| +------------+ | 1| | 2| +------------+ # ArrowEvalPython indicates Arrow optimization >>> df.select(udf(lambda x : x + 1, useArrow=True)('id')).explain() == Physical Plan == *(2) Project [pythonUDF0#18 AS <lambda>(id)#16] +- ArrowEvalPython [<lambda>(id#14L)#15], [pythonUDF0#18], 200 +- *(1) Range (0, 2, step=1, splits=1) ``` 2. Enable `spark.sql.execution.pythonUDF.arrow.enabled` Spark Conf to make all Python UDFs Arrow-optimized. ```sh >>> spark.conf.set("spark.sql.execution.pythonUDF.arrow.enabled", True) >>> df.select(udf(lambda x : x + 1)('id')).show() +------------+ |<lambda>(id)| +------------+ | 1| | 2| +------------+ # ArrowEvalPython indicates Arrow optimization >>> df.select(udf(lambda x : x + 1)('id')).explain() == Physical Plan == *(2) Project [pythonUDF0#30 AS <lambda>(id)#28] +- ArrowEvalPython [<lambda>(id#26L)#27], [pythonUDF0#30], 200 +- *(1) Range (0, 2, step=1, splits=1) ``` ### How was this patch tested? Parity unit tests. Closes apache#40725 from xinrong-meng/connect_arrow_py_udf. Authored-by: Xinrong Meng <xinrong@apache.org> Signed-off-by: Hyukjin Kwon <gurwls223@apache.org>

…pressions in `buildAggExprList` ### What changes were proposed in this pull request? Trim aliases before matching Sort/Having/Filter expressions with semantically equal expression from the Aggregate below in `buildAggExprList` ### Why are the changes needed? For a query like: ``` SELECT course, year, GROUPING(course) FROM courseSales GROUP BY CUBE(course, year) ORDER BY GROUPING(course) ``` Plan after `ResolveReferences` and before `ResolveAggregateFunctions` looks like: ``` !Sort [cast((shiftright(tempresolvedcolumn(spark_grouping_id#18L, spark_grouping_id, false), 1) & 1) as tinyint) AS grouping(course)#22 ASC NULLS FIRST], true +- Aggregate [course#19, year#20, spark_grouping_id#18L], [course#19, year#20, cast((shiftright(spark_grouping_id#18L, 1) & 1) as tinyint) AS grouping(course)#21 AS grouping(course)#15] .... ``` Because aggregate list has `Alias(Alias(cast((shiftright(spark_grouping_id#18L, 1) & 1) as tinyint))` expression from `SortOrder` won't get matched as semantically equal and it will result in adding an unnecessary `Project`. By stripping inner aliases from aggregate list (that are going to get removed anyways in `CleanupAliases`) we can match `SortOrder` expression and resolve it as `grouping(course)#15` ### Does this PR introduce _any_ user-facing change? No ### How was this patch tested? Existing tests ### Was this patch authored or co-authored using generative AI tooling? No Closes apache#51339 from mihailotim-db/mihailotim-db/fix_inner_aliases_semi_structured. Authored-by: Mihailo Timotic <mihailo.timotic@databricks.com> Signed-off-by: Wenchen Fan <wenchen@databricks.com>

…onicalized expressions ### What changes were proposed in this pull request? Make PullOutNonDeterministic use canonicalized expressions to dedup group and aggregate expressions. This affects pyspark udfs in particular. Example: ``` from pyspark.sql.functions import col, avg, udf pythonUDF = udf(lambda x: x).asNondeterministic() spark.range(10)\ .selectExpr("id", "id % 3 as value")\ .groupBy(pythonUDF(col("value")))\ .agg(avg("id"), pythonUDF(col("value")))\ .explain(extended=True) ``` Currently results in a plan like this: ``` Aggregate [_nondeterministic#15](#15), [_nondeterministic#15 AS dummyNondeterministicUDF(value)#12, avg(id#0L) AS avg(id)#13, dummyNondeterministicUDF(value#6L)#8 AS dummyNondeterministicUDF(value)#14](#15%20AS%20dummyNondeterministicUDF(value)#12,%20avg(id#0L)%20AS%20avg(id)#13,%20dummyNondeterministicUDF(value#6L)#8%20AS%20dummyNondeterministicUDF(value)#14) +- Project [id#0L, value#6L, dummyNondeterministicUDF(value#6L)#7 AS _nondeterministic#15](#0L,%20value#6L,%20dummyNondeterministicUDF(value#6L)#7%20AS%20_nondeterministic#15) +- Project [id#0L, (id#0L % cast(3 as bigint)) AS value#6L](#0L,%20(id#0L%20%%20cast(3%20as%20bigint))%20AS%20value#6L) +- Range (0, 10, step=1, splits=Some(2)) ``` and then it throws: ``` [[MISSING_AGGREGATION] The non-aggregating expression "value" is based on columns which are not participating in the GROUP BY clause. Add the columns or the expression to the GROUP BY, aggregate the expression, or use "any_value(value)" if you do not care which of the values within a group is returned. SQLSTATE: 42803 ``` - how canonicalized fixes this: - nondeterministic PythonUDF expressions always have distinct resultIds per udf - The fix is to canonicalize the expressions when matching. Canonicalized means that we're setting the resultIds to -1, allowing us to dedup the PythonUDF expressions. - for deterministic UDFs, this rule does not apply and "Post Analysis" batch extracts and deduplicates the expressions, as expected ### Why are the changes needed? - the output of the query with the fix applied still makes sense - the nondeterministic UDF is invoked only once, in the project. ### Does this PR introduce _any_ user-facing change? Yes, it's additive, it enables queries to run that previously threw errors. ### How was this patch tested? - added unit test ### Was this patch authored or co-authored using generative AI tooling? No Closes apache#52061 from benrobby/adhoc-fix-pull-out-nondeterministic. Authored-by: Ben Hurdelhey <ben.hurdelhey@databricks.com> Signed-off-by: Wenchen Fan <wenchen@databricks.com>

HyukjinKwon added 13 commits July 30, 2020 16:26

Do not install SparkR in test-only mode in testing script

b831229

Debug

cc954f1

Show R version

4bcd5a9

explicit r version print

a482085

Use master

4d9468c

Manually installs R 3.6

0450ae3

Revert "Manually installs R 3.6"

1ec49ec

This reverts commit 0450ae3.

debug

3c1526d

print R version

7c00c30

debug

3020eda

try different version

9a49621

manual installation

c5cc523

try

a1ce047

HyukjinKwon mentioned this pull request Jul 30, 2020

[SPARK-32493][INFRA] Manually install R instead of using setup-r in GitHub Actions apache/spark#29302

Closed

HyukjinKwon closed this Jul 30, 2020

HyukjinKwon deleted the install-r-debug branch December 7, 2020 02:06

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Debug SparkR test failure #15

Debug SparkR test failure #15

Uh oh!

HyukjinKwon commented Jul 30, 2020

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Debug SparkR test failure #15

Debug SparkR test failure #15

Uh oh!

Conversation

HyukjinKwon commented Jul 30, 2020

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants