[SPARK-39476][SQL] Disable Unwrap cast optimize when casting from Long to Float/ Double or from Integer to Float #36873
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
…ng from Long to Double/Float
WangGuangxin
force-pushed
the
SPARK-24994-followup
branch
from
8ff6461 to
cce6c87
Compare
|
you should file a new ticket since SPARK-24994 is already released .. |
WangGuangxin
changed the title
Thanks for remind, updated. @ulysses-you @sunchao @dongjoon-hyun @cloud-fan Please help review this |
cloud-fan
reviewed
Jun 15, 2022
|
|
||
| private def canUnwrapCast(from: DataType, to: DataType): Boolean = (from, to) match { | ||
| case (BooleanType, _) => true | ||
| case (IntegerType, FloatType) => false |
There was a problem hiding this comment.
can we add some code comments to explain the reason?
|
Can one of the admins verify this patch? |
sunchao
approved these changes
Jun 15, 2022
cloud-fan
pushed a commit
that referenced
this pull request
Jun 16, 2022
…g to Float/ Double or from Integer to Float
### What changes were proposed in this pull request?
Cast from Integer to Float or from Long to Double/Float may loss precision if the length of Integer/Long beyonds the **significant digits** of a Double(which is 15 or 16 digits) or Float(which is 7 or 8 digits).
For example, ```select *, cast(a as int) from (select cast(33554435 as foat) a )``` gives `33554436` instead of `33554435`.
When it comes the optimization rule `UnwrapCastInBinaryComparison`, it may result in incorrect (confused) result .
We can reproduce it with following script.
```
spark.range(10).map(i => 64707595868612313L).createOrReplaceTempView("tbl")
val df = sql("select * from tbl where cast(value as double) = cast('64707595868612313' as double)")
df.explain(true)
df.show()
```
With we disable this optimization rule , it returns 10 records.
But if we enable this optimization rule, it returns empty, since the sql is optimized to
```
select * from tbl where value = 64707595868612312L
```
### Why are the changes needed?
Fix the behavior that may confuse users (or maybe a bug?)
### Does this PR introduce _any_ user-facing change?
No
### How was this patch tested?
Add a new UT
Closes #36873 from WangGuangxin/SPARK-24994-followup.
Authored-by: wangguangxin.cn <[email protected]>
Signed-off-by: Wenchen Fan <[email protected]>
(cherry picked from commit 9612db3)
Signed-off-by: Wenchen Fan <[email protected]>
cloud-fan
pushed a commit
that referenced
this pull request
Jun 16, 2022
…g to Float/ Double or from Integer to Float
Cast from Integer to Float or from Long to Double/Float may loss precision if the length of Integer/Long beyonds the **significant digits** of a Double(which is 15 or 16 digits) or Float(which is 7 or 8 digits).
For example, ```select *, cast(a as int) from (select cast(33554435 as foat) a )``` gives `33554436` instead of `33554435`.
When it comes the optimization rule `UnwrapCastInBinaryComparison`, it may result in incorrect (confused) result .
We can reproduce it with following script.
```
spark.range(10).map(i => 64707595868612313L).createOrReplaceTempView("tbl")
val df = sql("select * from tbl where cast(value as double) = cast('64707595868612313' as double)")
df.explain(true)
df.show()
```
With we disable this optimization rule , it returns 10 records.
But if we enable this optimization rule, it returns empty, since the sql is optimized to
```
select * from tbl where value = 64707595868612312L
```
Fix the behavior that may confuse users (or maybe a bug?)
No
Add a new UT
Closes #36873 from WangGuangxin/SPARK-24994-followup.
Authored-by: wangguangxin.cn <[email protected]>
Signed-off-by: Wenchen Fan <[email protected]>
(cherry picked from commit 9612db3)
Signed-off-by: Wenchen Fan <[email protected]>
cloud-fan
pushed a commit
that referenced
this pull request
Jun 16, 2022
…g to Float/ Double or from Integer to Float
Cast from Integer to Float or from Long to Double/Float may loss precision if the length of Integer/Long beyonds the **significant digits** of a Double(which is 15 or 16 digits) or Float(which is 7 or 8 digits).
For example, ```select *, cast(a as int) from (select cast(33554435 as foat) a )``` gives `33554436` instead of `33554435`.
When it comes the optimization rule `UnwrapCastInBinaryComparison`, it may result in incorrect (confused) result .
We can reproduce it with following script.
```
spark.range(10).map(i => 64707595868612313L).createOrReplaceTempView("tbl")
val df = sql("select * from tbl where cast(value as double) = cast('64707595868612313' as double)")
df.explain(true)
df.show()
```
With we disable this optimization rule , it returns 10 records.
But if we enable this optimization rule, it returns empty, since the sql is optimized to
```
select * from tbl where value = 64707595868612312L
```
Fix the behavior that may confuse users (or maybe a bug?)
No
Add a new UT
Closes #36873 from WangGuangxin/SPARK-24994-followup.
Authored-by: wangguangxin.cn <[email protected]>
Signed-off-by: Wenchen Fan <[email protected]>
(cherry picked from commit 9612db3)
Signed-off-by: Wenchen Fan <[email protected]>
|
thanks, merging to master/3.3/3.2/3.1! |
sunchao
pushed a commit
to sunchao/spark
that referenced
this pull request
Jun 2, 2023
…g to Float/ Double or from Integer to Float
Cast from Integer to Float or from Long to Double/Float may loss precision if the length of Integer/Long beyonds the **significant digits** of a Double(which is 15 or 16 digits) or Float(which is 7 or 8 digits).
For example, ```select *, cast(a as int) from (select cast(33554435 as foat) a )``` gives `33554436` instead of `33554435`.
When it comes the optimization rule `UnwrapCastInBinaryComparison`, it may result in incorrect (confused) result .
We can reproduce it with following script.
```
spark.range(10).map(i => 64707595868612313L).createOrReplaceTempView("tbl")
val df = sql("select * from tbl where cast(value as double) = cast('64707595868612313' as double)")
df.explain(true)
df.show()
```
With we disable this optimization rule , it returns 10 records.
But if we enable this optimization rule, it returns empty, since the sql is optimized to
```
select * from tbl where value = 64707595868612312L
```
Fix the behavior that may confuse users (or maybe a bug?)
No
Add a new UT
Closes apache#36873 from WangGuangxin/SPARK-24994-followup.
Authored-by: wangguangxin.cn <[email protected]>
Signed-off-by: Wenchen Fan <[email protected]>
(cherry picked from commit 9612db3)
Signed-off-by: Wenchen Fan <[email protected]>
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?
Cast from Integer to Float or from Long to Double/Float may loss precision if the length of Integer/Long beyonds the significant digits of a Double(which is 15 or 16 digits) or Float(which is 7 or 8 digits).
For example,
select *, cast(a as int) from (select cast(33554435 as foat) a )gives33554436instead of33554435.When it comes the optimization rule
UnwrapCastInBinaryComparison, it may result in incorrect (confused) result .We can reproduce it with following script.
With we disable this optimization rule , it returns 10 records.
But if we enable this optimization rule, it returns empty, since the sql is optimized to
Why are the changes needed?
Fix the behavior that may confuse users (or maybe a bug?)
Does this PR introduce any user-facing change?
No
How was this patch tested?
Add a new UT