perf: avoid draining the event queue when prepending events

[Alek99]

Prepending events currently drains the pending queue with repeated shift() calls, builds a combined array, filters it, and pushes it back. Insert the filtered incoming events with unshift() instead, preserving queue identity and event order while removing the extra traversal and copies.

The ordinary append expression and event-processing condition are unchanged. This change adds no Python backend work or WebSocket protocol changes.

Validation

• Fourteen behavioral tests load the complete frontend module with Node’s native ES-module loader and explicit dependency stubs. They cover ordering, filtering, prepend, reconnect, async/reentrant dispatch, REST, mismatch handling, and rejection/retry. The original implementation fails the prepend-work regression: five shifts for three dispatches; this implementation uses three. Four loader cases cover normal and compact formatting, function declarations, and nested closures. All 18 cases pass on Node 22 and Node 26; no parser or npm dependency is required.
• uv run --no-sync pytest tests/units --cov --no-cov-on-fail --cov-report=: 8,298 passed, 18 skipped; 75.95% coverage (72% required).
• All pre-commit hooks passed: Ruff, codespell, stub generation, Pyright, ty, and Biome. Full pyright reflex tests and repository-wide Ruff lint/format checks also passed.
• Five paired Chrome 152 trials on an Apple M5 Pro, against c49a85d, showed prepend capacity improvements of 19.2% for 32 incoming + 32 pending events and 18.5% for 2,048 + 2,048. A higher-arrival confirmation reduced measured queue work by 16.2% and median completion p99 from 2.865 to 2.659 ms.

The benchmarks exercise queue processing with counted local callbacks/socket emits, excluding real network traffic and React rendering. Light-load timer results were noisy, including in an identical-source A/A control, so these measurements do not establish exact frontend latency equivalence or overall application speedup.

Submission checklist

• Followed the contributing guidelines and checked for overlapping open PRs.
• Added a performance changelog fragment for reflex-base.
• Added behavioral and work-count regression coverage.
• Reviewed the patch adversarially.

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.

[greptile-apps]

Greptile Summary

Replaces the prepend queue drain-and-rebuild operation with an in-place filtered unshift, preserving queue identity and event ordering while reducing work. The follow-up changes replace formatting-sensitive source extraction with complete-module evaluation through Node VM modules.

• Preserves append and event-processing behavior.
• Adds queue ordering, filtering, reconnect, reentrancy, failure, and work-count regression coverage.
• Resolves the previous formatting-sensitive test-harness finding.
• Renames the performance changelog fragment to the PR-numbered form.

Confidence Score: 5/5

The PR appears safe to merge with no outstanding correctness, security, or repository-rule issues.

The in-place prepend preserves filtering and FIFO order, the complete-module harness removes the previously reported formatting dependency, and that prior thread is manually resolved.

Important Files Changed

Filename	Overview
packages/reflex-base/src/reflex_base/.templates/web/utils/state.js	Uses filtered in-place prepend or append operations without changing dispatch behavior.
tests/units/reflex_base/client_event_queue_runtime.mjs	Evaluates the complete frontend module with explicit synthetic dependency and browser stubs.
tests/units/reflex_base/client_event_queue.mjs	Adds broad behavioral coverage and verifies that prepending no longer drains pending events.
tests/units/reflex_base/test_client_event_queue.py	Runs the Node regression suite with VM-module support and reports subprocess failures.
packages/reflex-base/news/7053.performance.md	Adds a correctly named concise performance changelog fragment.

_{Reviews (2): Last reviewed commit: "test: load the full client state module ..." | Re-trigger Greptile}

[codspeed]

Merging this PR will not alter performance

✅ 32 untouched benchmarks
⏩ 8 skipped benchmarks¹

---

_{Comparing codex/non-regressing-event-path (f15b0ad) with main (c49a85d)}

Footnotes

1. 8 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

[cubic-dev-ai]

All reported issues were addressed across 5 files

<sub>Reply with feedback, questions, or to request a fix.

Re-trigger cubic</sub>