bare websockets

[benedikt-bartscher]

saves ~11.6 KB gz / ~41 KB raw frontend size (a small js chunk replaces whole socketio)
dropped python deps: wsproto, simple-websocket, python-socketio, python-engineio, bidict
new benchmarks in test_event_transport.py show about 2x performance increase compared to socketio

also migrates AppHarness to granian for consistency. otherwise downstream tests using AppHarness still need uvicorn and wsproto/websockets

[codspeed-hq]

Merging this PR will not alter performance

✅ 32 untouched benchmarks
🆕 4 new benchmarks
⏩ 8 skipped benchmarks¹

Performance Changes

	Benchmark	BASE	HEAD	Efficiency
🆕	test_transport_inbound_socketio	N/A	7.9 ms	N/A
🆕	test_transport_inbound_websocket	N/A	4.4 ms	N/A
🆕	test_transport_outbound_socketio	N/A	15.4 ms	N/A
🆕	test_transport_outbound_websocket	N/A	6.7 ms	N/A

---

_{Comparing benedikt-bartscher:make-sio-optional (5e739cc) 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. ↩

[greptile-apps]

Greptile Summary

This PR replaces the default Socket.IO transport with a lightweight plain-WebSocket protocol while retaining Socket.IO as an optional transport.

• Adds matching browser and ASGI WebSocket implementations with handshake, heartbeat, reconnect, message validation, and token lifecycle handling.
• Makes Socket.IO and Uvicorn optional dependencies and updates configuration and startup paths accordingly.
• Migrates AppHarness to Granian and improves dependency-error reporting during cleanup.
• Fixes streamed-upload event-context isolation and adds transport unit tests and benchmarks.
• Documents the user-facing transport and dependency changes.

Confidence Score: 5/5

The PR appears safe to merge; no new actionable regressions or outstanding previous findings remain.

The changes since the previous review do not introduce a concrete transport failure, and every previous Greptile thread is resolved and therefore not outstanding.

Important Files Changed

Filename	Overview
packages/reflex-base/src/reflex_base/.templates/web/utils/helpers/websocket.js	Implements the browser-side plain-WebSocket transport, including framing, heartbeat handling, buffering, and lifecycle cleanup.
packages/reflex-base/src/reflex_base/.templates/web/utils/state.js	Selects the configured transport, lazily imports Socket.IO, and safely manages asynchronous connection and unmount cleanup.
reflex/event_namespace.py	Implements the server-side plain-WebSocket event protocol and its connection, validation, logging, and token-management behavior.
reflex/socketio_namespace.py	Extracts the legacy Socket.IO namespace and ASGI integration behind the optional Socket.IO dependency.
reflex/app.py	Configures either the native WebSocket route or optional Socket.IO application according to transport settings.
reflex/testing.py	Migrates the test harness backend to Granian and reports a clear optional-dependency error when cleanup lacks psutil.
packages/reflex-base/src/reflex_base/event/processor/event_processor.py	Forks the root event context for streaming deltas so unrelated events cannot become children of a completed future.
pyproject.toml	Moves Socket.IO and Uvicorn stacks into explicit extras while updating the required Granian version.
tests/units/test_event_namespace.py	Adds broad coverage for WebSocket framing, connection validation, event dispatch, rate limiting, and lifecycle behavior.

_{Reviews (32): Last reviewed commit: "Merge remote-tracking branch 'upstream/m..." | Re-trigger Greptile}

[cubic-dev-ai]

All reported issues were addressed across 19 files

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

Re-trigger cubic</sub>

[cubic-dev-ai]

All reported issues were addressed across 17 files

<sub>Requires human review: Auto-approval blocked because this review re-detected 1 unresolved issue already reported by Cubic.

Re-trigger cubic</sub>

[cubic-dev-ai]

All reported issues were addressed across 4 files (changes from recent commits).

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

Re-trigger cubic</sub>

[cubic-dev-ai]

All reported issues were addressed across 4 files (changes from recent commits).

<sub>Tip: Review your code locally with the cubic CLI to iterate faster.

Re-trigger cubic</sub>

[cubic-dev-ai]

All reported issues were addressed across 6 files (changes from recent commits).

<sub>Tip: Review your code locally with the cubic CLI to iterate faster.

Re-trigger cubic</sub>

[cubic-dev-ai]

All reported issues were addressed across 6 files (changes from recent commits).

<sub>Tip: Review your code locally with the cubic CLI to iterate faster.

Re-trigger cubic</sub>

[cubic-dev-ai]

All reported issues were addressed across 2 files (changes from recent commits).

<sub>Tip: Review your code locally with the cubic CLI to iterate faster.

Re-trigger cubic</sub>

[cubic-dev-ai]

All reported issues were addressed across 2 files (changes from recent commits).

<sub>Tip: Review your code locally with the cubic CLI to iterate faster.

Re-trigger cubic</sub>

[abulvenz]

I really like this lightweight approach 👍 .
One thing that is dropped, correct me if I see this wrong: under the hood socketio supports HTTP long-polling in case a websocket connection cannot be established (e.g. due to badly configured load balancers, reverse proxies or simply disabled client support). In most cases that should be fixed anyways on server side for performance reasons.

[benedikt-bartscher]

I really like this lightweight approach 👍 . One thing that is dropped, correct me if I see this wrong: under the hood socketio supports HTTP long-polling in case a websocket connection cannot be established (e.g. due to badly configured load balancers, reverse proxies or simply disabled client support). In most cases that should be fixed anyways on server side for performance reasons.

Thanks 🙏

Yes, the new bare websockets implementation does not support http long-polling yet. However one could still choose the old socketio based transport to archive this. Also iirc longpolling was disabled by default anyway, i guess not a lot of users use it. If this is really needed i could add support to our bare websockets implementation as well.

[benedikt-bartscher]

check out #6984

edit: done in 9caa24b

[masenf]

reflex-xy currently uses socketio multiplexing to send binary chart data and interaction payloads over the same reflex websocket, so we can't bring this in directly, yet.

i think there's potential here though. and i like the idea of dependency reduction.