Issues with get_id_pack, get_methods, a race and stdin, stdout issues

[rlehfeld]

Describe the issue briefly here, including:

• expected result versus actual result
  When working on a new Remote Library for Robot Framework (https://github.com/rlehfeld/robotframework-rpycremote) which takes the RPyC functionality into use, I run into several issues and also fixed them on my fork of RPyC. These changes I would like to describe but and if I remember right, I found 4 issues and unfortunately only one has a simple example which I will describe below.

1. for objects which do not provide a class property, RPyC is not usable. E.g. this is the case for some c-extensions or when it is hidden. (for this problem I will provide a simple example)
   Here the problem is within rpyc.lib.get_id_pack
2. When using Pydantic BaseModel objects, under some not completely understood cicumstances, the rpyc.lib.get_methods fails
3. There is a race in rpyc.core.protocol.Connection._unbox between the lines:

            if id_pack in self._proxy_cache:
                proxy = self._proxy_cache[id_pack]

as the garbage collector can and will kick in in between these two lines.
4. While fixing 3. and executing the unit tests I run into the following unrelated issue.
PipeStream and Win32PipeStream are taking ownership of stdin and stdout if from_std is used. When e.g. peer is closing the connection, the stdin and stdout of the python process is closed which can lead to exit code 1.

Further, I would like to say thank you for this super solution. Together with my Robot Framework Remote Library, this is a game changer.

Many thanks

René

Environment

• rpyc version
  pypy3.10-7.3.15-2.3.10.fc40.x86_64
• python version
  pypy3.10-7.3.15-2.3.10.fc40.x86_64
  python3-3.12.2-2.fc40.x86_64
• operating system
  Fedora 40

Minimal example for problem 1

Server:
Objects from Dummy class below, cannot be used together with RPyC.

    class Dummy:
        """dummy class"""
        def __init__(self):
            self._value = 1
            self.value2 = 5

        def __getattribute__(self, attr):
            if attr in ('__class__'):
                raise AttributeError(f'catched {attr}')
            return super().__getattribute__(attr)

        @property
        def value(self):
            """value getter"""
            print('called value getter')
            return self._value

        @value.setter
        def value(self, v):
            print('called value setter')
            self._value = v

        def __call__(self, *args, **kwargs):
            """callable object"""
            print(f'called __call__({args}, {kwargs})')
            return '__call__'

        def method(self, *args, **kwargs):
            """and some callable method"""
            print(f'called method({args}, {kwargs})')
            return 'method'