RamakrishnaChilaka opened a new pull request, #21177:
URL: https://github.com/apache/datafusion/pull/21177

   ## Which issue does this PR close?
   
   - Closes #21176
   
   ## Rationale for this change
   
   When `SELECT` projects columns in a different order than the table schema 
(e.g., `SELECT col_c, col_b` vs schema order `col_a, col_b, col_c`), the 
`LIMIT` clause is silently ignored — all rows are returned instead of the 
requested limit.
   
   The physical optimizer runs `LimitPushdown` (which correctly sets `fetch=N` 
on `MemorySourceConfig`) **before** `ProjectionPushdown`. When the projection 
is pushed down, `try_swapping_with_projection()` creates a new 
`MemorySourceConfig` via `try_new()`, which resets `fetch` to `None`, silently 
dropping the limit.
   
   ## What changes are included in this PR?
   
   One-line fix in `MemorySourceConfig::try_swapping_with_projection` to 
preserve `self.fetch` on the newly created config:
   
   ```rust
   // Before:
   .map(|s| Arc::new(s) as Arc<dyn DataSource>)
   // After:
   .map(|s| Arc::new(s.with_limit(self.fetch)) as Arc<dyn DataSource>)
   ```
   
   ## Are these changes tested?
   
   Yes.
   
   - **Unit test** (`try_swapping_with_projection_preserves_fetch`): directly 
verifies `fetch` is preserved after projection pushdown. Confirmed it fails 
without the fix (`None` vs `Some(5)`).
   - **SQL logic test** (`limit.slt`, table `t21176`): end-to-end regression 
test for `SELECT col_c, col_b FROM t LIMIT 5` with reverse column order.
   
   ## Are there any user-facing changes?
   
   No API changes. This is a correctness fix — `LIMIT` now works correctly 
regardless of `SELECT` column order against in-memory tables.
   


-- 
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]

Reply via email to