Subclassing from Modules

[malemburg]

BPO	765228
Nosy	@malemburg, @gvanrossum, @birkenfeld, @tiran

^{Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.}

Show more details

GitHub fields:

assignee = 'https://github.com/gvanrossum'
closed_at = <Date 2007-11-26.04:16:17.312>
created_at = <Date 2003-07-03.10:32:45.000>
labels = ['interpreter-core']
title = 'Subclassing from Modules'
updated_at = <Date 2007-11-26.04:16:17.311>
user = 'https://github.com/malemburg'

bugs.python.org fields:

activity = <Date 2007-11-26.04:16:17.311>
actor = 'gvanrossum'
assignee = 'gvanrossum'
closed = True
closed_date = <Date 2007-11-26.04:16:17.312>
closer = 'gvanrossum'
components = ['Interpreter Core']
creation = <Date 2003-07-03.10:32:45.000>
creator = 'lemburg'
dependencies = []
files = []
hgrepos = []
issue_num = 765228
keywords = []
message_count = 4.0
messages = ['16833', '55196', '57776', '57837']
nosy_count = 4.0
nosy_names = ['lemburg', 'gvanrossum', 'georg.brandl', 'christian.heimes']
pr_nums = []
priority = 'normal'
resolution = 'wont fix'
stage = None
status = 'closed'
superseder = None
type = None
url = 'https://bugs.python.org/issue765228'
versions = ['Python 2.4']

[malemburg]

In Python 2.2.3 there's a problem with accidental
subclassing from a Python module, e.g.

import MyStuff

class A(MyStuff):
     pass

this gives no error until you try to instantiate the
class:

o = A()

TypeError: 'module' object is not callable

In Python 2.3 the error is generated at module
startup time:

class A(MyStuff):
     pass
TypeError: function takes at most 2 arguments (3 given)

Since it is rather common that you create modules which
have the same name as their most important class, I
would find it more appropriate to raise a TypeError
with a message "can't subclass a module instance"
in both versions.

[birkenfeld]

In 2.5, the message says "module.__init__() takes at most 2 arguments (3
given)", which is at least a bit more specific.

You get similar errors when "deriving" from other arbitrary objects, so
I don't know if a special case makes sense here.

[tiran]

I agree. Python can't stop the developer from doing stupid things. We
could remove Py_TPFLAGS_BASETYPE from the module type but that could
cause incompatibilities with existing code.

I'm assigning the bug to our beloved dictator to ask for his opinion.

[gvanrossum]

I don't see an issue to be fixed here; adding special tests in order to
provide more detailed error messages is rarely a good idea.

Also, PEP-8 has said for years now that modules should *not* be named
the same as classes. Yes, there are a few such modules in the standard
library. They're historical mistakes that will be fixed in 3.0.