[New-bugs-announce] [issue42406] Importing multiprocessing breaks pickle.whichmodule

[Renato Cunha]

New submission from Renato Cunha <renato at renatocunha.com>:

Importing multiprocessing prior to other modules that define `ufunc`s breaks pickle.whichmodule for those functions.

Example:

    Python 3.8.6 (default, Sep 30 2020, 04:00:38)
    [GCC 10.2.0] on linux
    Type "help", "copyright", "credits" or "license" for more information.
    >>> import multiprocessing
    >>> from scipy.special import gdtrix
    >>> from pickle import whichmodule
    >>> whichmodule(gdtrix, gdtrix.__name__)
    '__mp_main__'

Now, if we change the order of imports, whichmodule works fine:

    Python 3.8.6 (default, Sep 30 2020, 04:00:38)
    [GCC 10.2.0] on linux
    Type "help", "copyright", "credits" or "license" for more information.
    >>> from scipy.special import gdtrix
    >>> from pickle import whichmodule
    >>> import multiprocessing
    >>> whichmodule(gdtrix, gdtrix.__name__)
    'scipy.special._ufuncs'

This happens because multiprocessing creates an alias to `__main__` when imported:

    Lib/multiprocessing/__init__.py
    37:    sys.modules['__mp_main__'] = sys.modules['__main__']

But doing so changes the ordering of `sys.modules`, and `whichmodule` returns `__mp_main__` instead of the correct `scipy.special._ufuncs`.

This bug has arisen in the context of using `dill` (which imports multiprocessing) for serialization and is further documented at https://github.com/uqfoundation/dill/issues/392.

I intend to submit a patch/PR with a fix soon.

----------
components: Interpreter Core
messages: 381410
nosy: renatolfc
priority: normal
severity: normal
status: open
title: Importing multiprocessing breaks pickle.whichmodule
type: behavior
versions: Python 3.10, Python 3.6, Python 3.7, Python 3.8, Python 3.9

_______________________________________
Python tracker <report at bugs.python.org>
<https://bugs.python.org/issue42406>
_______________________________________