language reference should specify restrictions on class namespace

[ericsnowcurrently]

BPO	17422
Nosy	@ncoghlan, @ned-deily, @durban, @ericsnowcurrently, @miss-islington, @cdonovick, @ananthan-123
PRs	bpo-17422:language reference should specify restrictions on class namespace #18559 [3.8] bpo-17422: Language reference should specify restrictions on class namespace (GH-18559) #18610 [3.7] bpo-17422: Language reference should specify restrictions on class namespace (GH-18559) #18611 bpo-17422: slightly more precise language #18682 [3.8] bpo-17422: slightly more precise language (GH-18682) #18812 [3.7] bpo-17422: slightly more precise language (GH-18682) #18813
Files	issue17422.stoneleaf.01.patch

^{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 2020-03-06.18:38:42.459>
created_at = <Date 2013-03-14.20:33:35.170>
labels = ['3.8', 'type-feature', '3.7', '3.9', 'docs']
title = 'language reference should specify restrictions on class namespace'
updated_at = <Date 2020-03-10.07:48:52.510>
user = 'https://github.com/ericsnowcurrently'

bugs.python.org fields:

activity = <Date 2020-03-10.07:48:52.510>
actor = 'ned.deily'
assignee = 'docs@python'
closed = True
closed_date = <Date 2020-03-06.18:38:42.459>
closer = 'eric.snow'
components = ['Documentation']
creation = <Date 2013-03-14.20:33:35.170>
creator = 'eric.snow'
dependencies = []
files = ['37712']
hgrepos = []
issue_num = 17422
keywords = ['patch']
message_count = 13.0
messages = ['184188', '362272', '362470', '362479', '362480', '362853', '362854', '362856', '363530', '363531', '363532', '363534', '363807']
nosy_count = 8.0
nosy_names = ['ncoghlan', 'ned.deily', 'daniel.urban', 'docs@python', 'eric.snow', 'miss-islington', 'donovick', 'Ananthakrishnan']
pr_nums = ['18559', '18610', '18611', '18682', '18812', '18813']
priority = 'normal'
resolution = 'fixed'
stage = 'resolved'
status = 'closed'
superseder = None
type = 'enhancement'
url = 'https://bugs.python.org/issue17422'
versions = ['Python 3.7', 'Python 3.8', 'Python 3.9']

[ericsnowcurrently]

(related to PEP-422 and issue bpo-17044)

The execution namespace from a class definition is passed as the third argument to the metaclass (see builtin___build_class__() in Python/bltinmodule.c). When applicable, which is almost always, the subsequent call to type.__new__() copies that namespace into a new dict which is in turn exposed via the class's __dict__.

This matters in the case where a metaclass with __prepare__() gets used (or the PEP-422 equivalent). The language reference[1] should note that the object returned by __prepare__() is copied into a new dict when the class is finally created via the metaclass.

[1] http://docs.python.org/3.4/reference/datamodel.html#customizing-class-creation

[ananthan-123]

I just addded a PR for this issue.

[gvanrossum]

New changeset fbe2e0b by ananthan-123 in branch 'master':
bpo-17422: Language reference should specify restrictions on class namespace (bpo-18559)
fbe2e0b

[miss-islington]

New changeset 91ba446 by Miss Islington (bot) in branch '3.8':
bpo-17422: Language reference should specify restrictions on class namespace (GH-18559)
91ba446

[miss-islington]

New changeset 36a120d by Miss Islington (bot) in branch '3.7':
bpo-17422: Language reference should specify restrictions on class namespace (GH-18559)
36a120d

[ericsnowcurrently]

Thanks for working on this. Sorry I didn't get a chance to see your PR sooner. There was one small thing that needs to be changed back, as I implied in my comment on the PR [1]. Please undo the change in the text from "ordered mapping" to "dict".

The original "is initialised as an empty ordered mapping" text line should be restored. "dict" is currently still not correct (the language spec does not dictate that dict be ordered). However, PEP-520 specifies that it must be ordered.

Thanks!

[1] #18559 (review)

[gvanrossum]

As I mentioned in my response on the PR, dict *is* required to be ordered since 3.7. So I think it is okay as is.

[cdonovick]

While It is perfectly fine behavior for CPython to use a dict, as dict is an ordered mapping. I think Eric is right that the data model should not specify that a dict is necessarily used in the absence of __prepare__.

[miss-islington]

New changeset e59334e by Caleb Donovick in branch 'master':
bpo-17422: slightly more precise language (GH-18682)
e59334e

[miss-islington]

New changeset f1b7964 by Miss Islington (bot) in branch '3.7':
bpo-17422: slightly more precise language (GH-18682)
f1b7964

[miss-islington]

New changeset 6df0c47 by Miss Islington (bot) in branch '3.8':
bpo-17422: slightly more precise language (GH-18682)
6df0c47

[ericsnowcurrently]

Thanks for fixing that, Caleb!

FWIW, I've opened a separate issue (bpo-39879) for adding a note in the language reference about dict ordering. Sorry for the confusion.

[ned-deily]

New changeset 2244814 by Ned Deily (Miss Islington (bot)) in branch '3.7':
bpo-17422: slightly more precise language (GH-18682)
2244814