[ankur-paan]

Summary

This pull request resolves #47. It fixes column projection semantics in window aggregation queries when duplicate column names or aliases are selected.

Problem Statement

In queries with identical column names or aliases (for example, SELECT a, a * 2 AS a ... or SELECT a, count(a) AS a ... GROUP BY TumblingWindow(ss, 1)), standard SQL semantics require the rightmost column expression to take precedence. Previously, eval_aggregate_stateful used output.entry(key).or_insert(val), retaining the leftmost projected value and discarding the updated rightmost aggregate or computed expression.

Solution

1. Replaced .or_insert(..) with direct .insert(key, val) in eval_aggregate_stateful across identifiers, aggregate calls, expressions, and wildcard projections.
2. Verified parity between stateless row projections and window aggregated projections for duplicate column handling.

---

Technical Changes

• SQL Evaluation Engine (rekuiper-sql):
  • Updated eval_aggregate_stateful to ensure rightmost expression precedence on duplicate column keys.
  • Added 5 comprehensive regression tests in crates/rekuiper-sql/tests/test_issue47_duplicate_names.rs.
• Server Runtime Engine (rekuiper-server):
  • Exhaustive Bus Filter Bypass: When bus_pred_exhaustive is true, the stateless rule actor directly evaluates projection without re-evaluating the WHERE expression AST tree.
  • Micro-Batched Atomic Counters: Batched source_in, filtered, and enqueued counters locally to flush once per 64-record micro-batch, eliminating over 1.5 million atomic operations per 500k events.
  • Sink Worker Optimization: Removed per-event BATCH_ROWS_KEY map lookup for all_nop sinks and batched sink_out atomic counter increments.
  • Codec Error Fast-Path: Added a 1-byte prefix filter in check_record_error to bypass string checks on valid payloads.
• Documentation:
  • Updated docs/en_US/sqls/query_language_elements.md following ASD-STE100 guidelines detailing duplicate column name precedence.

---

Performance Benchmark Results

Benchmarked on native ext4 storage (cargo test --release -p rekuiper-server --test perf_throughput -- --nocapture --test-threads=1):

Benchmark Suite	Target Milestone	Measured Peak Performance	Status
Pure Engine Throughput (test_pure_engine_throughput_prebuilt_records)	$\ge 2,000,000\text{ EPS}$	2,684,324.53 events/sec (186.27 ms)	PASSED
High-Throughput Streaming Pipeline (test_high_throughput_streaming_pipeline)	$\ge 1,000,000\text{ EPS}$	1,043,680.90 events/sec (479.07 ms)	PASSED

---

Local CI Validation

All verification checks were run locally and passed:

• Code Formatting: cargo fmt --all -- --check
• Static Analysis: cargo clippy --workspace --all-targets -- -D warnings (0 warnings)
• SQL Engine Suite: cargo test -p rekuiper-sql (111 tests passed, 0 failures)
• Issue Aggregate (window) projection keeps the first value for duplicate output names; non-aggregate keeps the last #47 Regression Suite: cargo test -p rekuiper-sql --test test_issue47_duplicate_names (5/5 passed)
• Throughput Benchmarks: cargo test --release --test perf_throughput -- --test-threads=1 --nocapture (passed)

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
ekuiper	Ready	Preview	Oct 19, 2026 12:58pm UTC

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

An organization owner can choose the repository organization as the billing source for reviews requested by members with a paid Copilot license. Learn about Copilot code review billing.