Dandandan opened a new pull request, #21964:
URL: https://github.com/apache/datafusion/pull/21964

   ## Which issue does this PR close?
   
   - Closes #.
   
   ## Rationale for this change
   
   Some optimizer paths were returning `Transformed::yes` even when they 
preserved the original plan or expression. That can cause unnecessary follow-up 
optimizer work and makes the transformed flag less useful as a fixed-point 
signal.
   
   ## What changes are included in this PR?
   
   - Reports `Transformed::no` when filter pushdown through joins, aggregates, 
windows, or table scans does not actually change the plan.
   - Reports `Transformed::no` when constant simplification preserves an 
expression after a runtime simplification error.
   - Tracks whether placeholder type inference actually fills in placeholder 
type information before reporting a transformation.
   - Preserves `Transformed::no` in `AggregateStatistics` recursive child 
optimization when the optimized child is the same plan handle.
   
   ## Are these changes tested?
   
   Existing and targeted tests:
   
   ```bash
   cargo fmt --all
   cargo test --locked -p datafusion-optimizer filter_with_table_provider
   cargo test --locked -p datafusion-optimizer test_simplify_divide_zero_by_zero
   cargo test --locked -p datafusion-expr placeholder
   cargo test --locked -p datafusion aggregate_statistics --test 
core_integration
   cargo check --locked -p datafusion-expr -p datafusion-optimizer -p 
datafusion-physical-optimizer
   cargo clippy --locked -p datafusion-expr -p datafusion-optimizer -p 
datafusion-physical-optimizer --all-targets --message-format=short -- -D 
warnings
   cargo clippy --locked --workspace --exclude datafusion-benchmarks 
--all-targets --all-features -- -D warnings
   git diff --check
   ```
   
   Also attempted:
   
   ```bash
   cargo clippy --all-targets --all-features -- -D warnings
   ```
   
   That command fails in `datafusion-benchmarks` because `--all-features` 
enables both `snmalloc` and `mimalloc`, which the benchmark binaries explicitly 
reject.
   
   ## Are there any user-facing changes?
   
   No.
   


-- 
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