Reword inheritance diagram for `BufferedRandom` to encompass `_io` and `_pyio` implementations

[tbrugere]

Documentation

The documentation for io.BufferedRandom states

A buffered binary stream providing higher-level access to a seekable RawIOBase raw binary stream. It inherits from BufferedReader and BufferedWriter.

(emphasis mine)

Except when runing cpython,

Python 3.14.0rc3 (main, Sep 18 2025, 19:47:22) [Clang 20.1.4 ] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> from io import BufferedRandom, BufferedReader
>>> issubclass(BufferedRandom, BufferedReader)
False
>>> BufferedRandom.mro()
[<class '_io.BufferedRandom'>, <class '_io._BufferedIOBase'>, <class '_io._IOBase'>, <class 'object'>]

It seems that BufferedRandom does not inherit from either BufferedReader or BufferedWriter

Linked PRs

• gh-141627: Fix BufferedRandom inheritance documentation #141629
• [3.15] gh-141627: Fix BufferedRandom inheritance documentation (GH-141629) #150801
• [3.14] gh-141627: Fix BufferedRandom inheritance documentation (GH-141629) #150802
• [3.13] gh-141627: Fix BufferedRandom inheritance documentation (GH-141629) #150803

[picnixz]

This is actually half-true. It does inherit from those classes when it's using the pure Python implementation. However, I see how confusing this can be and the alternative is to register the C implementation of BufferedRandom as a virtual subclass of BufferedReader and BufferedWriter but this is an overkill IMO.

[cmaloney]

In the C case, BufferedReader, BufferedWriter, and BufferedRandom all have the same core struct buffered + many method implementations. So it is implemented as though it inherited, just doesn't expose that at the Python abstraction.

That it "behaves like" is already documented in the last paragraph of the BufferedRandom class documentation:

BufferedRandom is capable of anything BufferedReader or BufferedWriter can do. In addition, seek() and tell() are guaranteed to be implemented.

I lean for just removing the "It inherits from" sentence entirely. It doesn't inherit but in my going through topic-IO issues haven't seen any bugs filed against that until this one. Making it virtually inherit changes isinstance might break some code; removing the doc line is unlikely to break anything.

[mohsinm-dev]

@cmaloney Thanks, that matches what I’m seeing too. In the C path, they do it effectively “as if” they inherited, but the Python MRO doesn’t expose that.

I’ll take the least‑surprising fix and drop the "It inherits from …” sentence entirely. The next paragraph already communicates the important part.