Use mypyc to compile websocket reader

[Dreamsorcerer]

No description provided.

[codspeed-hq]

Merging this PR will degrade performance by 21.8%

⚠️ Different runtime environments detected

Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.

Open the report in CodSpeed to investigate

❌ 2 regressed benchmarks
✅ 95 untouched benchmarks
⏩ 83 skipped benchmarks¹

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Benchmark	BASE	HEAD	Efficiency
❌	test_read_one_hundred_masked_large_binary_websocket_messages	60.2 ms	86.2 ms	-30.24%
❌	test_read_and_drain_one_hundred_websocket_text_messages	1.1 ms	1.3 ms	-12.34%

Tip

Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing mypyc (876981a) with master (deab032)}

Footnotes

1. 83 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. ↩

[codecov]

⚠️ JUnit XML file not found

The CLI was unable to find any JUnit XML files to upload.
For more help, visit our troubleshooting guide.

[Dreamsorcerer]

This is resulting in GBs of memory use, while removing it is resulting in numerous segfaults. Do not merge until this is fixed.

[webknjaz]

Does mypyc also have a coveragepy plugin like Cython?

[Dreamsorcerer]

Does mypyc also have a coveragepy plugin like Cython?

I don't think so, but it's literally the same code now, as it's just compiling the Python code. So we already have coverage from that.

[Dreamsorcerer]

If we get to fully converting to mypyc, then I think the only part we'd need to worry about coverage is on the main HTTP parser, as that'd be separate code that uses llhttp.

[webknjaz]

I'm not so sure — once that code's in a C-extension, it'll start getting optimized on the compiler level and there will be fewer guarantees that it's getting executed. So we'll need to make sure to execute exactly the same tests against both implementations at all times (otherwise, the metrics would be unreliable).

[Dreamsorcerer]

I'm not so sure — once that code's in a C-extension, it'll start getting optimized on the compiler level and there will be fewer guarantees that it's getting executed.

I'm not sure I follow. The compiler would optimise out parts of C-level code, it'd not give us any real change at the Python level as that'd just be instrumented calls which would always be called.

So we'll need to make sure to execute exactly the same tests against both implementations at all times (otherwise, the metrics would be unreliable).

We already do this. Again, the only discrepancies we need to worry about is the llhttp vs pure Python HTTP parser. But, before we could even consider that, mypyc needs to do native optimisations of ctypes calls.