Outdated PyObject_HasAttr documentation (it no longer behaves like builtin hasattr)

Documentation

The documentation says

Lines 50 to 52 in c31943a

    
              Returns ``1`` if *o* has the attribute *attr_name*, and ``0`` otherwise.  This 
        
              is equivalent to the Python expression ``hasattr(o, attr_name)``.  This function 
        
              always succeeds.

but since #53875 this is no longer equivalent to hasattr(o, attr_name) but, if I understood correctly, to something like:

try:
  return hasattr(o, attr_name)
except:
  return False

Linked PRs

gh-113391: fix outdated PyObject_HasAttr sphinx docs #113420

Isn't the following note enough?

cpython/Doc/c-api/object.rst

Lines 54 to 59 in c31943a

    
              .. note:: 
        
                 Exceptions that occur when this calls :meth:`~object.__getattr__` and 
        
                 :meth:`~object.__getattribute__` methods are silently ignored. 
        
                 For proper error handling, use :c:func:`PyObject_HasAttrWithError`, 
        
                 :c:func:`PyObject_GetOptionalAttr` or :c:func:`PyObject_GetAttr` instead.

pythongh-113391: fix outdated PyObject_HasAttr sphinx docs

Isn't the following note enough?

I think the sentence "This is equivalent to the Python expression hasattr(o, attr_name)." - must be removed. It actually belongs to the PyObject_HasAttrWithError().

Removing that sentence seems good, I was confused by that part because it contradicts the note below. Thank you !

gh-113391: fix outdated PyObject_HasAttr docs (#113420)

@skirpichev

@skirpichev I saw that a pull request to fix this was merged, but the issue is still open, is it OK to close it now ? Thanks again for the fix and sorry for this question, I'm not familiar with how python uses issue tracker

@encukou

Yeah, I think so.

But maybe it's stay opened intentionally. @encukou ?

I want to look at the docs for 3.12 (and 3.11) too, but didn't get to it yet.

On Thu, Jan 11, 2024 at 12:54:10AM -0800, Petr Viktorin wrote: I want to look the docs for 3.12 (and 3.11) too, but didn't get to it yet.

Ah, does make sense. BTW, patch could be cleanly applied to even earlier releases.

Just got back to this.
On 3.12 and below, PyObject_HasAttr is the closest thing to hasattr(o, attr_name). The difference from the Python function is in error handling, and the sentence just after the removed note makes things very clear.
So, I'll not backport.

Effectively, in 3.13 the “equivalent to hasattr” note was moved from PyObject_HasAttr to PyObject_HasAttrWithError.

Feel free to reopen/comment if you disagree.

pythongh-113391: fix outdated PyObject_HasAttr docs (python#113420)

Navigation Menu

Repository navigation

Outdated PyObject_HasAttr documentation (it no longer behaves like builtin `hasattr`) #113391

Documentation

Linked PRs

Metadata

Assignees

Labels

Projects

Milestone

Relationships

Development

	Returns ``1`` if o has the attribute attr_name, and ``0`` otherwise. This
	is equivalent to the Python expression ``hasattr(o, attr_name)``. This function
	always succeeds.

	.. note::

	Exceptions that occur when this calls :meth:`~object.__getattr__` and
	:meth:`~object.__getattribute__` methods are silently ignored.
	For proper error handling, use :c:func:`PyObject_HasAttrWithError`,
	:c:func:`PyObject_GetOptionalAttr` or :c:func:`PyObject_GetAttr` instead.

Outdated PyObject_HasAttr documentation (it no longer behaves like builtin hasattr) #113391

Description

Documentation

Linked PRs

Activity

serhiy-storchaka commented on Dec 22, 2023

skirpichev commented on Dec 23, 2023

perrinjerome commented on Dec 23, 2023

perrinjerome commented on Jan 11, 2024

skirpichev commented on Jan 11, 2024

encukou commented on Jan 11, 2024

skirpichev commented on Jan 11, 2024

encukou commented on Jan 16, 2024

Metadata

Metadata

Assignees

Labels

Projects

Milestone

Relationships

Development

Issue actions

Outdated PyObject_HasAttr documentation (it no longer behaves like builtin `hasattr`) #113391