[issue39978] Vectorcall implementation should conform to PEP 590.

[STINNER Victor]

STINNER Victor  added the comment:

> The change has a negative impact on performance

How did you notice the slowdown? I ran benchmarks in 
https://github.com/python/cpython/pull/17052 and I didn't notice any 
performance overhead.

--

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue39978] Vectorcall implementation should conform to PEP 590.

[Jeremy Kloth]

Change by Jeremy Kloth :


--
nosy: +jkloth

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue39978] Vectorcall implementation should conform to PEP 590.

[Mark Shannon]

New submission from Mark Shannon :

The implementation of `PyObject_Vectorcall` adds unnecessary overhead to PEP 
590, which undermines its purpose.
The implementation was changed in https://github.com/python/cpython/pull/17052, 
which I objected to at the time.

The change has a negative impact on performance as it add calls to both 
`PyThreadState_GET()` and `_Py_CheckFunctionResult()`.

This is practically an invitation for callers to skip `PyObject_Vectorcall` and 
access the underlying function pointer directly, making `PyObject_Vectorcall` 
pointless.

https://github.com/python/cpython/pull/17052 should be reverted.

--
keywords: 3.9regression
messages: 364325
nosy: Mark.Shannon, petr.viktorin, vstinner
priority: normal
severity: normal
status: open
title: Vectorcall implementation should conform to PEP 590.
versions: Python 3.9

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com