[SPARK-38407][SQL][3.2] ANSI Cast: loosen the limitation of casting non-null complex types #35754
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…ll complex types When ANSI mode is off, `ArrayType(DoubleType, containsNull = false)` can't cast as `ArrayType(IntegerType, containsNull = false)` since there can be overflow thus result in null results and breaks the non-null constraint. When ANSI mode is on, currently Spark SQL has the same behavior. However, this is not correct since the non-null constraint won't be break. Spark SQL can just execute the cast and throw runtime error on overflow, just like casting DoubleType as IntegerType. This applies to MapType and StructType as well. This PR is to loosen the limitation of casting non-null Array/Map/Struct types. For ANSI mode compliance <img width="559" alt="image" src="https://user-images.githubusercontent.com/1097932/156600154-24e3817b-ba22-45ce-a16f-faf7461af73e.png"> Yes, for Cast under ANSI mode or table insertion, a complex type which don't contain null can be cast as another non-null complex type as long as the element types are castable. Before changes, this is only allowed when the source element type can be upcast to the target element type. UT Closes apache#35724 from gengliangwang/canCast. Authored-by: Gengliang Wang <gengliang@apache.org> Signed-off-by: Gengliang Wang <gengliang@apache.org>
|
This PR is to port #35724 to branch-3.2 |
|
Merging to branch-3.2 |
gengliangwang
added a commit
that referenced
this pull request
Mar 9, 2022
…on-null complex types ### What changes were proposed in this pull request? When ANSI mode is off, `ArrayType(DoubleType, containsNull = false)` can't cast as `ArrayType(IntegerType, containsNull = false)` since there can be overflow thus result in null results and breaks the non-null constraint. When ANSI mode is on, currently Spark SQL has the same behavior. However, this is not correct since the non-null constraint won't be break. Spark SQL can just execute the cast and throw runtime error on overflow, just like casting DoubleType as IntegerType. This applies to MapType and StructType as well. This PR is to loosen the limitation of casting non-null Array/Map/Struct types. ### Why are the changes needed? For ANSI mode compliance <img width="559" alt="image" src="https://user-images.githubusercontent.com/1097932/156600154-24e3817b-ba22-45ce-a16f-faf7461af73e.png"> ### Does this PR introduce _any_ user-facing change? Yes, for Cast under ANSI mode or table insertion, a complex type which don't contain null can be cast as another non-null complex type as long as the element types are castable. Before changes, this is only allowed when the source element type can be upcast to the target element type. ### How was this patch tested? UT Closes #35754 from gengliangwang/port-35724. Authored-by: Gengliang Wang <gengliang@apache.org> Signed-off-by: Gengliang Wang <gengliang@apache.org>
kazuyukitanimura
pushed a commit
to kazuyukitanimura/spark
that referenced
this pull request
Aug 10, 2022
…on-null complex types ### What changes were proposed in this pull request? When ANSI mode is off, `ArrayType(DoubleType, containsNull = false)` can't cast as `ArrayType(IntegerType, containsNull = false)` since there can be overflow thus result in null results and breaks the non-null constraint. When ANSI mode is on, currently Spark SQL has the same behavior. However, this is not correct since the non-null constraint won't be break. Spark SQL can just execute the cast and throw runtime error on overflow, just like casting DoubleType as IntegerType. This applies to MapType and StructType as well. This PR is to loosen the limitation of casting non-null Array/Map/Struct types. ### Why are the changes needed? For ANSI mode compliance <img width="559" alt="image" src="https://user-images.githubusercontent.com/1097932/156600154-24e3817b-ba22-45ce-a16f-faf7461af73e.png"> ### Does this PR introduce _any_ user-facing change? Yes, for Cast under ANSI mode or table insertion, a complex type which don't contain null can be cast as another non-null complex type as long as the element types are castable. Before changes, this is only allowed when the source element type can be upcast to the target element type. ### How was this patch tested? UT Closes apache#35754 from gengliangwang/port-35724. Authored-by: Gengliang Wang <gengliang@apache.org> Signed-off-by: Gengliang Wang <gengliang@apache.org> (cherry picked from commit d99cbbf) Signed-off-by: Dongjoon Hyun <dongjoon@apache.org>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
What changes were proposed in this pull request?
When ANSI mode is off,
ArrayType(DoubleType, containsNull = false)can't cast asArrayType(IntegerType, containsNull = false)since there can be overflow thus result in null results and breaks the non-null constraint.When ANSI mode is on, currently Spark SQL has the same behavior. However, this is not correct since the non-null constraint won't be break. Spark SQL can just execute the cast and throw runtime error on overflow, just like casting DoubleType as IntegerType.
This applies to MapType and StructType as well. This PR is to loosen the limitation of casting non-null Array/Map/Struct types.
Why are the changes needed?
For ANSI mode compliance
Does this PR introduce any user-facing change?
Yes, for Cast under ANSI mode or table insertion, a complex type which don't contain null can be cast as another non-null complex type as long as the element types are castable. Before changes, this is only allowed when the source element type can be upcast to the target element type.
How was this patch tested?
UT