[SPARK-50354][SQL] No need to set initialInputBufferOffset when Aggregate mode is Complete

[zhixingheyi-tian]

What changes were proposed in this pull request?

Here:

spark/sql/core/src/main/scala/org/apache/spark/sql/execution/ReplaceHashWithSortAgg.scala

Lines 73 to 78 in 87a5b37

	if isHashBasedAggWithKeys(partialAgg) && isPartialAgg(partialAgg, hashAgg) =>
	if (SortOrder.orderingSatisfies(
	partialAgg.child.outputOrdering, sortAgg.requiredChildOrdering.head)) {
	sortAgg.copy(
	aggregateExpressions = sortAgg.aggregateExpressions.map(_.copy(mode = Complete)),
	child = partialAgg.child)

            if isHashBasedAggWithKeys(partialAgg) && isPartialAgg(partialAgg, hashAgg) =>
            if (SortOrder.orderingSatisfies(
                partialAgg.child.outputOrdering, sortAgg.requiredChildOrdering.head)) {
              sortAgg.copy(
                aggregateExpressions = sortAgg.aggregateExpressions.map(_.copy(mode = Complete)),
                child = partialAgg.child) 

Here. sortAgg comes from Final Aggregate, now becomes Complete mode. So need to disable the original initialInputBufferOffset

Just a small flaw.

Why are the changes needed?

This makes the code more rigorous.

Does this PR introduce any user-facing change?

No

How was this patch tested?

Existed UTs

[zhixingheyi-tian]

cc @cloud-fan @MaxGekk @dongjoon-hyun

[dongjoon-hyun]

Thank you for making a PR, @zhixingheyi-tian .

Could you elaborate about this claim in the PR description?

This makes the code more rigorous.

[cloud-fan]

are you saying that this parameter is not used in complete-mode aggregation, so we should set it to 0?

[zhixingheyi-tian]

Yes， just directly restore to the default value "0".

cc @dongjoon-hyun