delattr __slots__ inconsistancy

[ferringb]

BPO	7604
Nosy	@benjaminp

^{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 = None
closed_at = <Date 2009-12-30.19:36:07.821>
created_at = <Date 2009-12-30.18:46:47.743>
labels = ['interpreter-core', 'type-bug']
title = 'delattr __slots__ inconsistancy'
updated_at = <Date 2009-12-30.19:58:25.815>
user = 'https://bugs.python.org/ferringb'

bugs.python.org fields:

activity = <Date 2009-12-30.19:58:25.815>
actor = 'benjamin.peterson'
assignee = 'none'
closed = True
closed_date = <Date 2009-12-30.19:36:07.821>
closer = 'benjamin.peterson'
components = ['Interpreter Core']
creation = <Date 2009-12-30.18:46:47.743>
creator = 'ferringb'
dependencies = []
files = []
hgrepos = []
issue_num = 7604
keywords = []
message_count = 4.0
messages = ['97051', '97055', '97057', '97058']
nosy_count = 2.0
nosy_names = ['ferringb', 'benjamin.peterson']
pr_nums = []
priority = 'normal'
resolution = 'fixed'
stage = None
status = 'closed'
superseder = None
type = 'behavior'
url = 'https://bugs.python.org/issue7604'
versions = ['Python 2.6', 'Python 3.1']

[ferringb]

Everything I've read about __slots__, seen w/ them, etc, they're
effectively just a change in the underlying allocation- yes they can
limit the attributes, but that's about it.

Specifically, for general attribute access/mangling, best I can tell,
they're *supposed* to be exactly equivalent when manipulating the
object- the only difference being the backing store.

That said, delattr against slotted objects vs non slotted differs in a
rather buggy way- even worse, the behaviour differs on slotted classes
dependant on if you're delattr'ing a slotted attr or a nonslotted-
consider the following code.

class nonslotted(object):
  pass

class slotted(object):
  __slots__ = ("monkeys",)

try:
ns = nonslotted()
assert not hasattr(ns, 'monkeys')
del ns.monkeys
raise AssertionError("this is unreachable")
except AttributeError:
pass
try:
s = slotted()
assert not hasattr(s, 'monkeys')
del s.monkeys
print "slotting causes delattr to not throw an AttributeError"
# and now for the kicker
del s.some_attr_that_is_not_slotted
except AttributeError:
print "so... delattr results in AttributeError on a non-slotted attr,
but throws no AttributeError on a slotted attr..."

I'm guessing this isn't intentional/desired?

Confirmed on py2.6/py3.1 also; that said, I'd assume it affects all
versions of python that support __slots__...

[benjaminp]

Thank you for the report. Fixed in r77157.

[ferringb]

Any chance of this landing in py2.7? It's really a nasty wrench in the
works for some a mapping object essentially maps onto __slots__ for
memory efficiency...

[benjaminp]

2009/12/30 Brian Harring <report@bugs.python.org>:

Brian Harring <ferringb@gmail.com> added the comment:

Any chance of this landing in py2.7?  It's really a nasty wrench in the
works for some a mapping object essentially maps onto __slots__ for
memory efficiency...

It will be in 2.6.5, 2.7, 3.1.2, and 3.2.