Skip to content

Adjust case_when DivideByZeroProtection benchmark so that "percentage of zeroes" corresponds to "number of times protection is needed" #20105

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
alamb merged 2 commits into from
Feb 2, 2026
Merged

Adjust case_when DivideByZeroProtection benchmark so that "percentage of zeroes" corresponds to "number of times protection is needed" #20105

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
09e5658
Adjust case_when DivideByZeroProtection so that "percentage of zeroes…
pepijnve Feb 2, 2026
1ed9d7a
Add SLTs covering divide-by-zero protection
pepijnve Feb 2, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
31 changes: 1 addition & 30 deletions datafusion/physical-expr/benches/case_when.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -564,7 +564,6 @@ fn benchmark_divide_by_zero_protection(c: &mut Criterion, batch_size: usize) {

let numerator_col = col("numerator", &batch.schema()).unwrap();
let divisor_col = col("divisor", &batch.schema()).unwrap();
let divisor_copy_col = col("divisor_copy", &batch.schema()).unwrap();

// DivideByZeroProtection: WHEN condition checks `divisor_col > 0` and division
// uses `divisor_col` as divisor. Since the checked column matches the divisor,
Expand All @@ -578,35 +577,7 @@ fn benchmark_divide_by_zero_protection(c: &mut Criterion, batch_size: usize) {
|b| {
let when = Arc::new(BinaryExpr::new(
Arc::clone(&divisor_col),
Operator::Gt,
lit(0i32),
));
let then = Arc::new(BinaryExpr::new(
Arc::clone(&numerator_col),
Operator::Divide,
Arc::clone(&divisor_col),
));
let else_null: Arc<dyn PhysicalExpr> = lit(ScalarValue::Int32(None));
let expr =
Arc::new(case(None, vec![(when, then)], Some(else_null)).unwrap());

b.iter(|| black_box(expr.evaluate(black_box(&batch)).unwrap()))
},
);

// ExpressionOrExpression: WHEN condition checks `divisor_copy_col > 0` but
// division uses `divisor_col` as divisor. Since the checked column does NOT
// match the divisor, this falls back to ExpressionOrExpression evaluation.
group.bench_function(
format!(
"{} rows, {}% zeros: ExpressionOrExpression",
batch_size,
(zero_percentage * 100.0) as i32
),
|b| {
let when = Arc::new(BinaryExpr::new(
Arc::clone(&divisor_copy_col),
Operator::Gt,
Operator::NotEq,
lit(0i32),
));
let then = Arc::new(BinaryExpr::new(
Expand Down
21 changes: 21 additions & 0 deletions datafusion/sqllogictest/test_files/case.slt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -621,6 +621,27 @@ a
b
c

query I
SELECT CASE WHEN d != 0 THEN n / d ELSE NULL END FROM (VALUES (1, 1), (1, 0), (1, -1)) t(n,d)
----
1
NULL
-1

query I
SELECT CASE WHEN d > 0 THEN n / d ELSE NULL END FROM (VALUES (1, 1), (1, 0), (1, -1)) t(n,d)
----
1
NULL
NULL

query I
SELECT CASE WHEN d < 0 THEN n / d ELSE NULL END FROM (VALUES (1, 1), (1, 0), (1, -1)) t(n,d)
----
NULL
NULL
-1

# EvalMethod::WithExpression using subset of all selected columns in case expression
query III
SELECT CASE a1 WHEN 1 THEN a1 WHEN 2 THEN a2 WHEN 3 THEN b END, b, c
Expand Down