jayzhan211 commented on code in PR #26211:
URL: https://github.com/apache/datafusion/pull/26211#discussion_r4373338441
##########
datafusion/sqllogictest/test_files/aggregate.slt:
##########
@@ -10295,3 +10295,16 @@ SELECT avg(DISTINCT v ORDER BY v), var(DISTINCT v
ORDER BY v) FROM agg_order_by_
statement ok
DROP TABLE agg_order_by_t;
+
+# Repro for issue #17699 / #26661: distinct count with sum(decimal) preserves
declared type across rewrites
+statement ok
+CREATE TABLE s_decimal_distinct (g varchar, k varchar, x decimal(18,6)) AS
VALUES ('a', 'u1', 1.5), ('a', 'u2', 2.5), ('b', 'u1', 3.0);
+
+query T R I I
Review Comment:
The sqllogictest parser rejects `query T R I I` (it fails with `unexpected
token: "I"`, which is the CI failure), and the decimal column prints with its
scale. With these two changes the file passes locally:
```diff
-query T R I I
+query TRII
SELECT g, nullif(sum(x), 0) AS s, count(*) AS n, count(DISTINCT k) AS d
FROM s_decimal_distinct GROUP BY g ORDER BY g;
----
-a 4 2 2
-b 3 1 1
+a 4.000000 2 2
+b 3.000000 1 1
```
##########
datafusion/sqllogictest/test_files/single_distinct_to_groupby.slt:
##########
@@ -453,3 +453,35 @@ DROP TABLE empty_t;
statement ok
DROP TABLE t;
+
+# Repro for issue #26193: single distinct with sum(decimal) preserves declared
type
+statement ok
+CREATE TABLE t_decimal_distinct AS SELECT * FROM (VALUES (CAST(1.25 AS
DECIMAL(5,2)), 1), (CAST(2.50 AS DECIMAL(5,2)), 2)) AS v(d, x);
+
+query TT
+SELECT arrow_typeof(s), arrow_typeof(c) FROM (SELECT sum(d) AS s,
count(DISTINCT x) AS c FROM t_decimal_distinct);
+----
+Decimal128(15, 2) Int64
+
+query TT
Review Comment:
This expected plan doesn't match what the rule produces, so the file fails
in CI (the outer aggregate reads `sum(alias2)`, and the current physical plan
has `DataSourceExec` and no `CoalesceBatchesExec`). Running `cargo test -p
datafusion-sqllogictest --test sqllogictests -- single_distinct_to_groupby
--complete` regenerates it; on this branch it gives:
```diff
logical_plan
-01)Projection: CAST(sum(t_decimal_distinct.d) AS Decimal128(15, 2)) AS s,
count(alias1) AS c
-02)--Aggregate: groupBy=[[]], aggr=[[sum(t_decimal_distinct.d),
count(alias1)]]
-03)----Aggregate: groupBy=[[t_decimal_distinct.x AS alias1]],
aggr=[[sum(t_decimal_distinct.d)]]
+01)Projection: CAST(sum(alias2) AS Decimal128(15, 2)) AS s, count(alias1)
AS c
+02)--Aggregate: groupBy=[[]], aggr=[[sum(alias2), count(alias1)]]
+03)----Aggregate: groupBy=[[t_decimal_distinct.x AS alias1]],
aggr=[[sum(t_decimal_distinct.d) AS alias2]]
04)------TableScan: t_decimal_distinct projection=[d, x]
physical_plan
-01)ProjectionExec: expr=[CAST(sum(t_decimal_distinct.d)@0 AS Decimal128(15,
2)) as s, count(alias1)@1 as c]
-02)--AggregateExec: mode=Final, gby=[], aggr=[sum(t_decimal_distinct.d),
count(alias1)]
+01)ProjectionExec: expr=[CAST(sum(alias2)@0 AS Decimal128(15, 2)) as s,
count(alias1)@1 as c]
+02)--AggregateExec: mode=Final, gby=[], aggr=[sum(alias2), count(alias1)]
03)----CoalescePartitionsExec
-04)------AggregateExec: mode=Partial, gby=[],
aggr=[sum(t_decimal_distinct.d), count(alias1)]
-05)--------AggregateExec: mode=FinalPartitioned, gby=[alias1@0],
aggr=[sum(t_decimal_distinct.d)]
-06)----------CoalesceBatchesExec: target_batch_size=8192
-07)------------RepartitionExec: partitioning=Hash([alias1@0], 4),
input_partitions=4
-08)--------------AggregateExec: mode=Partial, gby=[alias1@1],
aggr=[sum(t_decimal_distinct.d)]
-09)----------------RepartitionExec: partitioning=RoundRobinBatch(4),
input_partitions=1
-10)------------------MemoryExec: partitions=1, partition_sizes=[1]
+04)------AggregateExec: mode=Partial, gby=[], aggr=[sum(alias2),
count(alias1)]
+05)--------AggregateExec: mode=FinalPartitioned, gby=[alias1@0 as alias1],
aggr=[sum(t_decimal_distinct.d) as alias2]
+06)----------RepartitionExec: partitioning=Hash([alias1@0], 4),
input_partitions=1
+07)------------AggregateExec: mode=Partial, gby=[x@1 as alias1],
aggr=[sum(t_decimal_distinct.d) as alias2]
+08)--------------DataSourceExec: partitions=1, partition_sizes=[1]
```
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
