Jean-Paul Calderone

The `random` parameter was removed because "no one ever uses" it "nor would they have a valid reason to do so". Also, Raymond has "long been annoyed" by it. https://github.com/python/cpython/issues/84645

Twisted is getting involved via `twisted.persisted.styles` which modifies the copyreg module to add support for pickling and unpickling methods. `modin.pandas.DataFrame` does a lot of stuff dynamically (eg it has `__getattribute__`)...

Also twisted.internet.test.test_tcp.AbortConnectionTests_CFReactorTests.test_fullWriteBufferAfterByteExchange

`ProtocolWrapper` is a ... protocol (`IProtocol`) wrapper. It's not an `IHandshakeListener` wrapper. It's not an anything-else wrapper, either. If you want a different kind of wrapper, why not use one?...

I think I agree with something _like_ this issue but I'm not sure exactly what. Recall that `twisted.internet.interfaces.IHandshakeListener` exists and its `handshakeCompleted` will only be called after the handshake completes...

If your protocol implements `twisted.internet.interfaces.IHandshakeListener` then you can wait for `handshakeCompleted` instead of `connectionMade`. It's a bit unfortunate that this separate interface exists but if `connectionMade` (note - actually `makeConnection`)...

> The author of the PR needs to run the test locally anyway and can generate a local diff report. This is impossible for many contributions. I cannot run the...

> For example, consider a case of 10000 tests that access a database. Currently tests can only be run in sequence, each using the same database and cleaning up after...

> Also, just a note. The approach of having a worker index is one of the standard approaches of optimizing parallel execution. In such case it's possible to avoid all...

> Another example: we started to design workerIndex() API. For some users it's an error if test runs without parallelization. For me it's perfectly fine - I would want workerIndex()...