tohuya6 opened a new pull request, #23849:
URL: https://github.com/apache/datafusion/pull/23849
## Which issue does this PR close?
- Part of #22715.
## Rationale for this change
`multi_group_by::group_column_supported_type` gates which GROUP BY columns
may
use the column-wise `GroupValuesColumn` fast path, and the gate is
all-or-nothing: a single unsupported column forces the **entire** grouping
onto
the byte-encoded `GroupValuesRows` fallback, even when every other key column
would have qualified. A `Decimal256` key triggers exactly that today.
Although `i256` is a 32-byte native rather than a machine-word scalar, it is
`Copy` and already implements `ArrowNativeTypeOp` + `HashValue`, so
`Decimal256`
reuses the existing `PrimitiveGroupValueBuilder` with no new builder type
and no
`T: Copy` relaxation — it mirrors the `Decimal128` arm that already exists.
## What changes are included in this PR?
- Dispatch `Decimal256Type` in `make_group_column`, mirroring the
`Decimal128` arm.
- Accept `Decimal256(_, _)` in `group_column_supported_type`.
- Update the `group_column_supported_type` ↔ `make_group_column` consistency
fuzz: `Decimal256` previously served as the stock *unsupported* example,
so it
moves into the supported set (`Float16` / invalid `Time` unit combinations
remain as the negative examples).
- Add a `Decimal256` group-count benchmark to `benches/multi_group_by.rs`.
## Are these changes tested?
Yes.
- New unit test: a `Decimal256` key (precision > 38, so genuinely 256-bit)
stays
on the `GroupValuesColumn` path, dedups including nulls, and round-trips
with
the `Decimal256(precision, scale)` output type preserved.
- The consistency fuzz now asserts `Decimal256` routes through the
dispatcher.
- New single- and multi-column `Decimal256` `GROUP BY` coverage in
`aggregate.slt`.
## Are there any user-facing changes?
No API changes. `GROUP BY` queries with a `Decimal256` key now use the
column-wise fast path instead of the row-encoded fallback; results are
unchanged.
--
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]
