bpo-40304: Correct type(name, bases, dict) doc

[verhovsky]

https://bugs.python.org/issue40304

[verhovsky]

Please correct me if there's some gotcha I'm missing, but if classes inherit from object, then don't they also do that when created via type?

>>> type('X', (object,), dict(a=1)).__bases__
(<class 'object'>,)
>>> type('X', (), dict(a=1)).__bases__
(<class 'object'>,)

[merwok]

No gotcha, in Python 3 both are equivalent.

This means that this bit of the docs (a couple paragraphs before your change) is not strictly accurate:

the bases class name and becomes the :attr:~definition.__name__ attribute; the bases tuple itemizes the base classes and becomes the :attr:~class.__bases__ tuple itemizes the base classes and becomes the :attr:~class.__bases__ attribute

object ends up in __bases__ even when implicitly added by the metaclass!

[terryjreedy]

E'ric, can you finish this off, with blurb and possible additional doc change?

[taleinat]

Ping, @merwok?

[merwok]

Hi! I will come back to this.

[merwok]

@taleinat @terryjreedy Please review additional changes!

• class dict does not become instance __dict__ at all
• itemizes is an obscure term, I thought about using enumerates but went for the simpler contains
• I think this does not need a NEWS note, unless this is the first contribution from the PR author in which case I will add one to give attribution as a thank-you 🙂

[merwok]

I have added the skip news and backport labels. If there is no negative review about the wording I will merge this in a couple days!

[taleinat]

This is an important fix and I like the direction!

However, the new wording is still not accurate: When providing one or more base classes, object is not added to __bases__. For example:

>>> class A:
	pass
>>> type('B', (A,), {}).__bases__
(<class '__main__.A'>,)

Also, if the attributes dict doesn't become __dict__, what exactly is done with it? Perhaps this would be a good place for a link to the docs about class instantiation?

[bedevere-bot]

A Python core developer has requested some changes be made to your pull request before we can consider merging it. If you could please address their requests along with any other requests in other reviews from core developers that would be appreciated.

Once you have made the requested changes, please leave a comment on this pull request containing the phrase I have made the requested changes; please review again. I will then notify any core developers who have left a review that you're ready for them to take another look at this pull request.

[merwok]

Also, if the attributes dict doesn't become dict, what exactly is done with it?

It is passed to the metaclass so that functions become methods, other descriptors do their thing, and all attributes are stored in the class object in the end… in TheClass.__dict__! (instance __dict__ is only for instance attributes)

[merwok]

Text I removed:

the dict dictionary is the namespace containing definitions
for class body and is copied to a standard dictionary to become the
:attr:~object.__dict__ attribute.

so it was not wrong but could be confusing; what do you think about this change:

to become the :attr:~object.__dict__ attribute of the class object.

(let me check if the link target only talks about class instances)

[taleinat]

LGTM. Small suggestion about the NEWS entry, @terryjreedy.

[bedevere-bot]

A Python core developer has requested some changes be made to your pull request before we can consider merging it. If you could please address their requests along with any other requests in other reviews from core developers that would be appreciated.

Once you have made the requested changes, please leave a comment on this pull request containing the phrase I have made the requested changes; please review again. I will then notify any core developers who have left a review that you're ready for them to take another look at this pull request.

And if you don't make the requested changes, you will be put in the comfy chair!

Made requested changes

[merwok]

Where does this link go?

class.__bases__ is at https://docs.python.org/3/library/stdtypes.html#class.__bases__
does class.__dict__ go to https://docs.python.org/3/reference/datamodel.html#index-48 ?

[merwok]

No link is created!

Ideas:

• keep object.__dict__ but amend the doc there (in library/stdtypes) to describe __dict__ for class objects
• change the reference to be to __dict__ (class attribute)

[terryjreedy]

Since I changed object to class and that does not work, I changed it back for this issue, so it continues to link to
file:///F:/dev/3x/Doc/build/html/library/stdtypes.html#object.dict
Since 'objects' include 'classes', the current text is not incorrect. I consider changing it a separate issue.
If classes were the only objects with dict attribute (I am not sure), and object.dict should be class.dict, a change would require grepping for other references to change. This would be a different issue also.

one change needed

[miss-islington]

Thanks @verhovsky for the PR, and @terryjreedy for merging it 🌮🎉.. I'm working now to backport this PR to: 3.8, 3.9.
🐍🍒⛏🤖

[bedevere-bot]

GH-24295 is a backport of this pull request to the 3.9 branch.

[bedevere-bot]

GH-24296 is a backport of this pull request to the 3.8 branch.

[terryjreedy]

Boris, you have 6 merged PRs before this. After checking them, I think you qualify for a listing in Misc/acks. If you make a PR with no issue or news to add yourself, with the improved English transliteration, and request my review, I will merge it, and ask for a 3.9 backport.