Fix rounds over decimal in Spark 330+ #5786
Merged
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
Signed-off-by: sperlingxx <lovedreamf@gmail.com>
|
build |
jlowe
approved these changes
Jun 8, 2022
Comment on lines
+586
to
+587
| rounded.castTo( | ||
| DecimalUtil.createCudfDecimal(dataType.asInstanceOf[DecimalType])) |
There was a problem hiding this comment.
This cast seems unnecessary in most cases, may be worth checking if the desired cudf type is already the type computed to avoid an unnecessary copy.
nartal1
approved these changes
Jun 8, 2022
tgravescs
added a commit
that referenced
this pull request
Jun 8, 2022
* Correct the error message for test_mod_pmod_by_zero (#5781) Signed-off-by: Firestarman <firestarmanllc@gmail.com> * Update the error string for test_cast_neg_to_decimal_err on 330[databricks] (#5784) * Update the error string for test_cast_neg_to_decimal_err on 330 Signed-off-by: Firestarman <firestarmanllc@gmail.com> * address comments Signed-off-by: Firestarman <firestarmanllc@gmail.com> * Throw an exception when attempting to read columnar encrypted Parquet files on the GPU [databricks] (#5761) * Throw useful message when parquet columnar encryption enabled * update message * fix message * handle native encrypted * move variable * cleanup * fix native check * cleanup imports * fix import order * Sign off Signed-off-by: Thomas Graves <tgraves@nvidia.com> * Shim the parquet crypto exception check Signed-off-by: Thomas Graves <tgraves@nvidia.com> * shim 320cdh * Add test for parquet encryption Signed-off-by: Thomas Graves <tgraves@nvidia.com> * fix rounds over decimal in Spark 330+ (#5786) Passes the datatype of round-like functions directly to GPU overrides, so as to adapt different Spark versions. Signed-off-by: sperlingxx <lovedreamf@gmail.com> * Fix the overflow of container type when casting floats to decimal (#5766) Fixes #5765 Fix the potential overflow when casting float/double to decimal. The overflow occurs on the container decimal for HALF_UP round. Signed-off-by: sperlingxx <lovedreamf@gmail.com> Co-authored-by: Liangcai Li <firestarmanllc@gmail.com> Co-authored-by: Alfred Xu <lovedreamf@gmail.com>
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.
SPARK-39226 changed the way to compute output precision and scale for round-like functions, which fails the round-like tests: #5769.
This PR passes the datatype of round-like functions directly to GPU overrides, so as to adapt different Spark versions. The shim method approach doesn't work very well in this case because the computition of decimal precision and scale requiring member variables of rounds-like functions.
This PR also modifies the round logic over decimal type to adapt both original and new manner of datatype. In addition, some test cases are skipped in Spark 330+ since they are not supported by Spark any more.
Signed-off-by: sperlingxx lovedreamf@gmail.com