enum.Enum is False-y

[pitrou]

BPO	26748
Nosy	@gvanrossum, @warsaw, @pitrou, @ethanfurman, @serhiy-storchaka

^{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/ethanfurman'
closed_at = <Date 2016-04-14.21:00:43.068>
created_at = <Date 2016-04-13.17:22:01.365>
labels = ['type-bug', 'library']
title = 'enum.Enum is False-y'
updated_at = <Date 2016-04-14.21:00:43.066>
user = 'https://github.com/pitrou'

bugs.python.org fields:

activity = <Date 2016-04-14.21:00:43.066>
actor = 'ethan.furman'
assignee = 'ethan.furman'
closed = True
closed_date = <Date 2016-04-14.21:00:43.068>
closer = 'ethan.furman'
components = ['Library (Lib)']
creation = <Date 2016-04-13.17:22:01.365>
creator = 'pitrou'
dependencies = []
files = []
hgrepos = []
issue_num = 26748
keywords = []
message_count = 10.0
messages = ['263342', '263343', '263344', '263345', '263346', '263347', '263348', '263359', '263371', '263430']
nosy_count = 7.0
nosy_names = ['gvanrossum', 'barry', 'pitrou', 'eli.bendersky', 'ethan.furman', 'python-dev', 'serhiy.storchaka']
pr_nums = []
priority = 'normal'
resolution = 'fixed'
stage = 'resolved'
status = 'closed'
superseder = None
type = 'behavior'
url = 'https://bugs.python.org/issue26748'
versions = ['Python 3.4', 'Python 3.5']

[pitrou]

>>> import enum
>>> bool(enum.Enum)
False
>>> bool(enum.IntEnum)
False

This behaviour is relatively unexpected for classes, and can lead to subtle bugs such as the following:

https://bitbucket.org/ambv/singledispatch/issues/8/inconsistent-hierarchy-with-enum

[gvanrossum]

I guess it's marked 2.7 because of the enum34 backport? There's no enum in the 2.7 stdlib.

I believe this was brought up before on one of the lists but I don't recall the outcome of the discussion, except that for IntEnum the behavior is correct. I tend to agree that for plain Enum it's a problem, the question is whether we can fix it without breaking code that accidentally relies on this behavior.

[gvanrossum]

Oh wait. The *class* is False-y? That's definitely a bug, just fix it.

[pitrou]

Yes, I didn't find a separate bug tracker for the enum34 backport, which is why I included that version here.

for IntEnum the behavior is correct

Do you remember the argument? I agree that IntEnum *instances* may be falsy, but IntEnum classes I don't see why.

That said, if an IntEnum *class* has to be false-y, then there's no real point in fixing just Enum (you can pass an IntEnum instance to singledispatch() too).

[pitrou]

Ok, we posted at the same time :-) Yes, the class is false-y.

[pitrou]

Just reading the code now, the reason is that EnumMeta pretends to be a collection (it defines a __len__ and an __iter__).

[ethanfurman]

EnumMeta /is/ a collection (at least in the same sense the dict class is a collection). ;)

Fix is on it's way...

[serhiy-storchaka]

See also bpo-23008.

[python-dev]

New changeset 772805538caf by Ethan Furman in branch '3.4':
bpo-26748: Enum classes should evaluate as True
https://hg.python.org/cpython/rev/772805538caf

New changeset f840608f79da by Ethan Furman in branch '3.5':
bpo-26748: Enum classes should evaluate as True
https://hg.python.org/cpython/rev/f840608f79da

New changeset 2fc61f8ee2d2 by Ethan Furman in branch 'default':
bpo-26748: Enum classes should evaluate as True
https://hg.python.org/cpython/rev/2fc61f8ee2d2

[ethanfurman]

Enum classes are now Truth-y in 3.4, 3.5, enum34, and aenum. :)